INTERMEDIATE

DYNAMICS
Second Edition

Patrick Hamill
Contents

Part 1. Principles of Mechanics 7

Chapter 1. A Brief Review of Introductory Concepts 9

1.1. Kinematics 9
1.2. Newton’s Second Law 11
1.3. Work and Energy 13
1.4. Momentum 14
1.5. Rotational Motion 15
1.6. Summary 22
1.7. Problems 23

Chapter 2. Kinematics 29
2.1. Galileo Galilei (Optional Historical Note) 30
2.2. The Principle of Inertia 31
2.3. Basic Concepts in Kinematics 32
2.4. The Position of a Particle on a Plane 43
2.5. Unit Vectors 44
2.6. Kinematics in Two Dimensions 47
2.7. Kinematics in Three Dimensions 51
2.8. Summary 60
2.9. Problems 61

Chapter 3. Newton’s Laws: Determining the Motion 69

3.1. Isaac Newton(Optional Historical Note) 70
3.2. The Law of Inertia 71
3.3. Newton’s Second Law and the Equation of Motion 74
3.4. Newton’s Third Law: Action Equals Reaction 78
3.5. Is Rotation Absolute or Relative? 81
3.6. Determining the Motion 83
3.7. Numerical Method to Determine the Motion (Optional) 94
iii
iv CONTENTS

3.8. Summary 98
3.9. Problems 99

Chapter 4. The Lagrangian Method 109

4.1. The Equation of Motion by Inspection 109
4.2. The Lagrangian 111
4.3. Lagrange’s Equations 117
4.4. Degrees of Freedom 121
4.5. Generalized Momentum 123
4.6. Generalized Force (Optional) 126
4.7. The Calculus of Variations (Optional) 128
4.8. Hamilton’s Equations (Optional) 134
4.9. Summary 138
4.10. Problems 140

Chapter 5. The Conservation of Energy 149

5.1. The Work-Energy Theorem 149
5.2. Work Along a Path. The Line Integral 151
5.3. Potential Energy 155
5.4. Force, Work, and Potential Energy 167
5.5. The Conservation of Energy 172
5.6. Energy Diagrams 176
5.7. Solving for the Motion: The Energy Integral 178
5.8. The Kinetic Energy of a System of Particles 181
5.9. Work on an Extended Body. Pseudowork 183
5.10. Summary 184
5.11. Problems 187

Chapter 6. Conservation of Linear Momentum 195

6.1. The Law of Conservation of Momentum 195
6.2. The Motion of a Rocket 196
6.3. Collisions 200
6.4. Inelastic Collisions. The Coefficient of Restitution 210
6.5. Impulse 211
6.6. Momentum of a System of Particles 212
6.7. Relative Motion and the Reduced Mass 214
6.8. Collisions in Center of Mass Coordinates (Optional) 216
6.9. Summary 221
6.10. Problems 223

Chapter 7. Conservation of Angular Momentum 233

7.1. Definition of Angular Momentum 233
7.2. Conservation of Angular Momentum 235
 CONTENTS v

7.3. Angular Momentum of a System of Particles 237

7.4. Rotation of a Rigid Body about a Fixed Axis 244
7.5. The Moment of Inertia 247
7.6. The Gyroscope 250
7.7. Angular Momentum is an Axial Vector 253
7.8. Summary 255
7.9. Problems 257
Chapter 8. Conservation Laws and Symmetries 265
8.1. Symmetry 265
8.2. Symmetry and the Laws of Physics 267
8.3. Symmetries and Conserved Physical Quantities 268
8.4. Are the Laws of Physics Symmetrical? 270
8.5. Strangeness 272
8.6. Symmetry Breaking 274
8.7. Summary 274
8.8. Problems 274

Part 2. The Gravitational Field 277

Chapter 9. The Gravitational Field 279
9.1. Newton’s Law of Universal Gravitation 280
9.2. The Gravitational Field 282
9.3. The Gravitational Field of an Extended Body 286
9.4. The Gravitational Potential 289
9.5. Field Lines and Equipotential Surfaces 293
9.6. The Newtonian Gravitational Field Equations 294
9.7. The Equations of Poisson and Laplace 298
9.8. Einstein’s Theory of Gravitation (Optional) 299
9.9. Summary 306
9.10. Problems 308

Part 3. The Mechanics of Particles 315

Chapter 10. Central Force Motion: The Kepler Problem 317
10.1. Johannes Kepler (Optional Historical Note) 318
10.2. Kepler’s Laws 321
10.3. Central Forces 321
10.4. The Equation of Motion 328
10.5. Energy and the Effective Potential Energy 331
10.6. Solving the Radial Equation of Motion 336
10.7. The Equation of the Orbit 337
10.8. The Equation of an Ellipse 342
vi CONTENTS

10.9. Kepler’s Laws Revisited 348

10.10. A Perturbed Circular Orbit 355
10.11. Resonances 361
10.12. Summary 362
10.13. Problems 363
Chapter 11. Harmonic Motion 373
11.1. Springs and Pendulums 373
11.2. Solving the Differential Equation (Optional) 376
11.3. The Damped Harmonic Oscillator 383
11.4. The Forced Harmonic Oscillator 389
11.5. Coupled Oscillators 402
11.6. Summary 409
11.7. Problems 410
Chapter 12. The Pendulum 417
12.1. A Simple Pendulum with Arbitrary Amplitude 418
12.2. The Physical Pendulum 424
12.3. The Center of Percussion 429
12.4. The Spherical Pendulum 434
12.5. Summary 445
12.6. Problems 447
Chapter 13. Accelerated Reference Frames 453
13.1. A Linearly Accelerating Reference Frame 453
13.2. A Rotating Coordinate Frame 455
13.3. Fictitious Forces 457
13.4. Centrifugal Force and the Plumb Bob 460
13.5. The Coriolis Force 462
13.6. The Foucault Pendulum 469
13.7. Summary 476
13.8. Problems 476

Part 4. The Mechanics of Extended Bodies 481

Chapter 14. Statics (Optional) 483
14.1. Basic Concepts 483
14.2. Couples, Resultants and Equilibrants 487

14.3. Reduction to the Simplest Set of Forces 489

14.4. The Hanging Cable 489
14.5. Stress and Strain 494
14.6. The Centroid (Optional) 497
14.7. The Center of Gravity (Optional) 499
 CONTENTS vii

14.8. Equilibrium of Fluids 501

14.9. D’Alembert’s Principle and Virtual Work (Optional) 507
14.10. Summary 511
14.11. Problems 513

Chapter 15. Rotational Kinematics 519

15.1. Orientation of a Rigid Body 520
15.2. Orthogonal Transformations 522
15.3. The Euler Angles 530
15.4. Euler’s Theorem 534
15.5. Infinitesimal Rotations 545
15.6. Summary 547
15.7. Problems 549

Chapter 16. Rotational Dynamics 551

16.1. Angular Momentum 552
16.2. Kinetic Energy 556
16.3. Properties of the Inertia Tensor 558
16.4. The Euler Equations of Motion 571
16.5. Torque-Free Motion 572
16.6. The Spinning Top 574
16.7. Summary 582
16.8. Problems 585

Chapter 17. Waves 589

17.1. A Wave in a Stretched String 590
17.2. Direct Solution of the Wave Equation 593
17.3. Standing Waves 595
17.4. Traveling Waves 597
17.5. Standing Waves as a Special Case of Traveling Waves 600
17.6. Energy 603
17.7. Momentum 607
17.8. Summary 609
17.9. Problems 610

Chapter 18. Small Oscillations (Optional) 615

18.1. Introduction 615
18.2. Statement of the Problem 615
18.3. Normal Modes 621
18.4. Matrix Formulation 629
18.5. Normal Coordinates 632
18.6. Coupled Pendulums: An Example 634
18.7. Many Degrees of Freedom 638
viii CONTENTS

18.8. Transition to Continuous Systems 643

18.9. Summary 647
18.10. Problems 650

Part 5. Special Topics 653

Chapter 19. The Special Theory of Relativity 655
19.1. Albert Einstein (Optional Historical Note) 656
19.2. Experimental Background 657
19.3. The Postulates of Special Relativity 659
19.4. The Lorentz Transformations 660
19.5. The Addition of Velocities 668
19.6. Simultaneity and Causality 670
19.7. The Twin Paradox 673
19.8. Minkowski Space-Time Diagrams 675
19.9. 4-Vectors 679
19.10. Relativistic Dynamics 685
19.11. Summary 688
19.12. Problems 689
Chapter 20. Classical Chaos (Optional) 695
20.1. Configuration Space and Phase Space 696
20.2. Periodic Motion 698
20.3. Attractors 700
20.4. Chaotic Trajectories and Liapunov Exponents 702
20.5. Poincaré Maps 703
20.6. The Henon Heiles Hamiltonian 705
20.7. Summary 707
20.8. Problems 708
Chapter A. Formulas and Constants 713
Chapter B. Answers to Selected Problems 717
Preface

Although this book begins at an introductory level, by the end of

the book the student will have been exposed to all of the subject matter
usually found in an intermediate mechanics course as well as a few less
common advanced topics.
Organization
This book is divided into five parts, preceded by a short chapter
consisting of a review of a few essential introductory concepts. This
chapter can be skipped by well prepared students, or assigned as read-
ing for students who only need a quick refresher.
Part I (Chapters 2 through 8) is called “Principles of Mechanics.”
This part covers all the basic concepts in mechanics. The topics of kine-
matics, dynamics and the conservation principles are treated in depth.
The first chapter in this part is called “Kinematics.” This is the tradi-
tional starting point for courses in intermediate mechanics. Here, the
student is exposed to relations between acceleration, velocity and po-
sition in Cartesian, plane polar, cylindrical, and spherical coordinates.
A few simple concepts from vector analysis are introduced. A number
of reasonably difficult projectile problems are included in the problems.
The next chapter considers Newton’s laws. It includes a discussion on
“Determining the Motion” in which the student learns techniques for
integrating Newton’s second law to obtain the position as a function of
time. This is done for constant forces, and for forces that are functions
of time, of velocity, and of position. A short section called “Solving for
the Motion by Numerical Methods” gives a flavor for the use of compu-
tational techniques in physics. The role of computers in physics is not
emphasized in this course. However, I realize that many instructors
want to expose their students to computational methods, so I have in-
cluded a few discussions of numerical techniques. Furthermore, nearly

1
2 PREFACE

every chapter has a number of “Computational Projects.” However,

this text does not emphasize the role of computers in physics because
I find that teaching the traditional material of intermediate mechanics
takes most of two semesters and does not give enough time to delve
into computational physics. Furthermore, many universities have in-
cluded a computational physics course in the undergraduate physics
curriculum.
Chapter 4 is called “The Lagrangian Approach.” I think it is im-
portant for physics students to be exposed to the Lagrangian early on
in mechanics. The Lagrangian is presented, at first, as a simple tech-
nique for generating the equation of motion. Later in the chapter, I go
through a derivation of the Lagrange equations using the Calculus of
Variations. This section need not be covered if the instructor feels it
is too advanced. The chapter ends with an optional discussion of the
Hamiltonian and Hamilton’s equations. There are several reasons why
I chose to present the Lagrangian early in the course, perhaps the most
important is that it gives the student a simple (almost “cookbook”)
technique for obtaining the equations of motion for a complicated dy-
namical system. For example, the Lagrangian technique allows one to
easily determine the equations of motion for a double pendulum, for
a spherical pendulum, or for coupled oscillators. More importantly,
it allows one to introduce the concepts of generalized momentum and
ignorable coordinates and leads to the relation between conservation
laws and symmetries. Furthermores, it lets the student know that
this course is not simply a rehash of concepts learned in introductory
physics.
The following four chapters cover the conservations of energy, linear
momentum, and angular momentum, followed by a short chapter on
the relation between symmetry and conservation laws. The chapter on
the conservation of energy discusses potential energy and the concept
of effective potential. Potential energy naturally leads to a discussion
of the gradient of a scalar field. There is a section on the way the “Del”
operator can be expressed in cylindrical and spherical coordinates; this
allows one to discuss coordinate transformations in general. (I be-
lieve that introducing concepts from vector calculus as required by the
physics is much more effective than stuffing all of the vector concepts
into a single introductory chapter.) The chapters on the conservation
of linear and angular momentum cover the usual topics (rockets, col-
lisions, etc.) as well as some less usual topics such as the fact that
angular momentum is an axial vector.
Part II is a very short introduction to field theory. Since field theory
is treated exhaustively in courses on electromagnetism, this study is
 PREFACE 3

limited to a single chapter on the gravitational field. Additional vector

concepts are introduced here and the student is exposed to Gauss’s law
and the equations of Poisson and Laplace.
Next is Part III (Chapters 10 - 13) called “The Mechanics of Par-
ticles.” This important section covers central force motion, harmonic
motion and motion in accelerated reference frames. The student is ex-
posed to the techniques of expansions in power series and the idea of
using successive approximations to solve otherwise intractable physics
problems. Chapter 10 deals with central force motion, as illustrated by
the Kepler problem. Also considered is the stability of circular orbits
which shows the student how to deal with small perturbations. Specif-
ically, we imagine a comet striking a planet in a previously perfectly
circular orbit and analyze the subsequent motion to determine stability
and the frequency of radial oscillations.
The next chapter is called “Simple Harmonic Motion.” Damped and
driven harmonic oscillators are treated in depth. A rather thorough dis-
cussion on how to solve second order differential equations is included
here. Coupled oscillators and normal modes are touched upon.
The third chapter in this part (Chapter 12) is on the motion of a
pendulum. We begin with the motion of a simple pendulum of arbi-
trary amplitude and introduce elliptic integrals. Next we consider the
physical pendulum, centers of oscillation and percussion, the spherical
pendulum, and the conical pendulum. To spend a whole chapter on
the pendulum may seem excessive, but it is a simple, easily visualized
physical system that allows one to introduce many useful mathematical
techniques without having to spend time explaining the motion.
The fourth chapter in this part is called “Accelerated Reference
Frames” in which we (mainly) consider motion on the surface of the
rotating Earth. Coriolis forces and the Foucault pendulum are treated.
Once again, perturbation theory is used to solve these problems.
Part IV (Chapters 14-18) is called “The Mechanics of Extended
Bodies.” Portions of the material in this part may be too advanced
for some classes, but the instructor should try to cover the chapters
on rotational kinematics, rotational dynamics, and waves. This part
begins with an optional chapter on statics that includes a discussion
of d’Alembert’s principle and virtual work. The next chapter (on rota-
tional kinematics) introduces orthogonal transformations. We obtain
the Euler angles and consider Euler’s theorem. The following chapter
(Chapter 16) on rotational dynamics introduces the inertia tensor and
some simple methods from tensor analysis. The next chapter in this
part is an introduction to wave motion. This topic is not considered
4 PREFACE

in great detail since it is treated extensively in the undergraduate elec-

tromagnetic theory class, nevertheless it does show the student how to
solve a partial differential equation by separation of variables. The last
chapter in this part treats small oscillations. It is rather advanced and
the chapter is denoted as optional.
The last part of the book is called “Special Topics” and consists
of two chapters. The first of these is on special relativity. The second
chapter is an introduction to classical chaos. Neither of these chapters
is at all exhaustive; they are simply intended to give the student a
flavor of these interesting subjects.
A One-Semester Course
Many instructors will find that the intermediate mechanics course
in their department has been reduced to one semester. If such is the
situation at your institution, you may wish to cover the material in
Chapters 2, 4, 6, 7, 11, 13, 15, 16, 17 and include some material from
the skipped chapters, in particular, Sections 3.6, 5.6, and 12.2
Exercises and Problems
Learning physics requires doing physics, so I have included a large
number of “exercises.” These are found at the end of nearly every
section. Most of them are fairly easy. Some are merely “plug-ins” to
get the student to look at a formula and (hopefully) to think about
it. Others ask the student to fill in the missing steps in a derivation.
A few require a bit of clever thinking. Nearly all have answers given.
I hope that students studying this book will solve every one of these
exercises. At the end of each chapter is a collection of problems that
are of the degree of difficulty to be expected from a course at this level.
Most of these will require significant effort on the part of the student.
However, I believe that a student who has read the chapter and worked
the exercises will be prepared to attack the problems.
 PREFACE 5

Acknowledgements
I thank my colleagues at San Jose State University and NASA Ames
Research Center, particularly Drs. Alejandro Garcia and Michael Kauf-
man. I am especially indebted to the many students in my mechanics
courses whose influence on this book cannot be overestimated.
 Part 1

Principles of Mechanics
Chapter 1
A Brief Review of
Introductory Concepts

This introductory chapter is intended to remind you of some of

the basic concepts from your introductory physics course. You should
probably read through this chapter quickly, work out some of the ex-
ercises and try a few of the problems at the end of the chapter. If you
remember all of this material, go on to Chapter 2, but if you find that
you have forgotten some of these basic concepts, you should go back
to your introductory physics textbook and review the material there.
Most of the standard introductory physics texts are well written and
contain many instructive figures and diagrams. It is a good idea to go
back to that text whenever you are exposed to the same material on a
more advanced level.
As you may recall, the mechanics section of your introductory
physics book covered the following topics:
• Kinematics
• Newton’s Second Law
• Work and Energy
• Momentum
• Rotational Motion
We now very briefly review some concepts from each of these items.

1.1. Kinematics
Kinematics is defined as the study of motion. Essentially, kine-
matics involves determing the relationships between position, velocity,
acceleration, and time.
9
10 1. A BRIEF REVIEW OF INTRODUCTORY CONCEPTS

Position is denoted by the vector r and the change in position (or

displacement) can be written as ∆r. Velocity is defined as the displace-
ment with respect to time, so the average velocity is given by
∆r
< v >=
∆t
where ∆t is the time during which the object moved a distance ∆r.
As the time interval becomes very small, we replace the difference
(represented by ∆) with the derivative and write
dr
v= . (1.1)
dt
The change in velocity with respect to time is called the acceleration
and is given by
dv
a= . (1.2)
dt
Clearly, we can use the definitions of acceleration and velocity to
write the inverse relations:
Z Z
dv = adt,

and Z Z
dr = vdt.

1.1.1. Motion in a Straight Line at Constant Acceleration.

For motion in a straight line, we do not need to use vector notation.
Letting the position to be represented by x, if the acceleration is con-
stant the integrals above are easy to evaluate, leading to the familiar
relations
v(t) = at + v0 , (1.3)
and
1
x(t) = at2 + v0 t + x0 , (1.4)
2
where v0 and x0 are the initial velocity and the initial position.
You probably memorized Equations (1.3) and (1.4) in your intro-
ductory physics course. You should, however, keep in mind that they
are only valid if the acceleration is constant. In this course, we shall
frequently be concerned with non-constant accelerations and these re-
lations cannot be used.
Two other useful equations can be obtained from (1.3) and (1.4)
as follows. Solve one equation for t and substitute it in the other to
obtain
2a(x − x0 ) = v 2 − v02 ,
 1.2. NEWTON’S SECOND LAW 11

or solve one equation for a and substitute it in the other equation to

obtain
1
x − x0 = (v + v0 )t.
2

Comment 1. In this book you will find a large number of

“exercises.” They are not difficult. They are intended to
give you a chance to review and understand the concepts
in the preceding section. It would be a good idea to work
out the solution for each exercise as you read through
the text. At the end of each chapter you will find a set
of problems that are significantly more difficult than the
exercises. However, if you solve the exercises, you will be
well prepared for solving the problems. Many students
find working the exercises to be a good preparation for
the examinations.

Exercise 1.1. You were driving your new Ferrari at 62 mph (= 100
km/hr) when you spotted a police car. Naturally, you hit the brakes.
You slowed to 31 mph, covering a distance of 50 m. (a) What is your
constant acceleration? (b) How much time did it take to slow to 31
mph? Answers: (a) -5.79 m/s2 (b) 2.4 sec.

1.2. Newton’s Second Law

The study of the relation between the forces acting on a body and
the motion of the body is called dynamics. In your introductory course
you were exposed to dynamics in the form of Newton’s Second Law.
This law states that a body of mass m that is acted upon by a force F
will accelerate at
a = F/m.
This relationship is usually remembered as
F = ma. (1.5)
As you shall see, this form of Newton’s second law is only valid if the
mass is constant.
12 1. A BRIEF REVIEW OF INTRODUCTORY CONCEPTS

Figure 1.1. A block sliding down an inclined plane

with friction. Sketch (a) shows the forces acting on the
block. Sketch (b) shows the free-body diagram. Sketch
(c) shows the x-and y-axes inclined so that they are par-
allel and perpendicular to the plane.

Worked Example 1.1. Determine the acceleration of a block

of mass m sliding down an inclined plane of angle θ. Assume the
coefficient of sliding friction is µ. See Figure 1.1.
Solution: The forces acting on the block are gravity mg (down-
wards), the normal force N (perpendicular to the plane), and the
frictional force µN (parallel to the plane). These forces are illus-
trated in Figure 1.1(a). The free-body diagram with all of the
forces acting at the center of mass of the block is shown in Figure
1.1(b). In a problem such as this one, it is convenient to assume the
axes are parallel to and perpendicular to the surface of the plane
as indicated in Figure 1.1(c).
There is no acceleration perpendicular to the plane so the net
force in that direction must be zero. Consequently, N = mg cos θ.
The net force down the plane is Fd = mg sin θ − µN . The acceler-
ation of the block is
Fd
a= = g sin θ − µg cos θ.
m

Exercise 1.2. Two blocks of mass M1 and M2 are tied together.

They are sitting on a smooth frictionless surface as shown in Figure
1.2. A force F is applied to the free string attached to M1 . What
is the tension in the string between the two blocks? Answer: T =
M2 F/(M1 + M2 ).

Exercise 1.3. A block of mass 25 kg is held in place on an inclined

plane of angle 30◦ as shown in Figure 1.3. The coefficient of static
 1.3. WORK AND ENERGY 13

Figure 1.2. Two blocks on a smooth frictionless surface

connected by a massless string.

friction is 0.4. a) Draw the free body diagram. What forces act on the
block? b) What is the tension in the string? c) If the string is cut,
what is the acceleration of the block? Answers (b) T = 37.63 N (c)
a = 1.51 m/s2 .

30o

Figure 1.3. A block on an inclined plane. What is the

acceleration of the block if the string is cut?

1.3. Work and Energy

Imagine pushing a box along a horizontal surface, such as a table-
top. The force you are applying can be denoted by F. As you might
expect, there will be opposing forces such as friction, air resistance,
etc., but for now we are only interested in the force you are exerting.
If you push the box from x1 to x2 , the work you do on it is
Z x2
W = F · dx. (1.6)
x1
When we do work on an object, we often change its energy. The
energy of an object is either kinetic energy (energy of motion) or po-
tential energy (energy of position). For a particle of mass m moving
with speed v the kinetic energy (denoted by T ) is given by
1
T = mv 2 . (1.7)
2
14 1. A BRIEF REVIEW OF INTRODUCTORY CONCEPTS

The potential energy of an object of mass m raised a height h above

the surface of the Earth is given by
V = mgh. (1.8)
We shall often be interested in the potential energy of a stretched
or compressed spring. The potential energy of such a system is given
by
1
V = kx2 , (1.9)
2
where k is the spring constant and x represents the amount the spring
is stretched or compressed.
Power is defined as the work done per unit time. Suppose a force F
acts on a particle for an infinitesimal time interval, from t to t+dt. The
work done during this interval is dW = F · dx. Therefore,the power is

dW F · dx
P = = .
dt dt
Since dx/dt ≡ v, another expression for power is P = F · v.

Exercise 1.4. A rock is thrown upward from the top of a 30 m

building with a velocity of 5 m/s. Determine its velocity (a) When it
falls back past its original point, (b) When it is 15 m above the street,
and (c) Just before it hits the street. Answer: (a) -5 m/s, (c)24.76 m/s.

Exercise 1.5. A horse drags a 100 kg sled a distance of 4 km in

20 minutes. The horse exerts one horsepower, of course. What is the
coefficient of sliding friction between the sled and the ground? Answer:
µk = 0.23.

1.4. Momentum
A moving particle is characterized by having a particular momen-
tum. When we use the term “momentum” we usually are referring to
the linear momentum, not to be confused with the angular momentum,
which we will define in a few moments.
 1.5. ROTATIONAL MOTION 15

The momentum is simply defined as mass times velocity and is

denoted by the letter p. Thus,
p = mv. (1.10)
If the mass of a body is constant, the time derivative of the mo-
mentum is
dp d(mv) dv
= =m = ma
dt dt dt
But according to Newton’s second law, F = ma, so Newton’s second
law can be expressed as
dp
F= (1.11)
dt
Here F is the net or total vector sum of all forces acting on the body.
Consequently, we appreciate that if the net force is zero, the time de-
rivative of the momentum is zero. That is, the momentum of the body
is constant. This is called the law of conservation of momentum.

Exercise 1.6. A 1500 kg car traveling east at 40 km/hr turns a

corner and speeds up to a velocity of 50 km/hr due north. What is the
change in the car’s momentum? Answer: 26, 700 kg m/s at 38.7◦ West
of North.

1.5. Rotational Motion

The motion of a body rotating about a fixed axis is mathematically
identical to one dimensional linear motion. Recall that kinematics
studies the relationship between position, velocity, and acceleration.
Rotational kinematics deals with angular position (θ), angular velocity
(ω), and angular acceleration (α), where
dθ
ω≡ ,
dt
and
dω d2 θ
α≡ = 2.
dt dt
You will discover that rotational motion can be very complicated.
To keep things simple for the moment, consider the special case of
a symmetrical body rotating about a fixed axis, such as the wheel
illustrated on the left side of Figure 1.4.
16 1. A BRIEF REVIEW OF INTRODUCTORY CONCEPTS

Figure 1.4. A wheel mounted on a fixed axis. The

wheel is allowed to rotate but not to translate.

For a fixed, stationary axis, the center of the wheel is at rest. All
other points are moving in circles around it. If you looked straight down
the axis you would see the circle shown on the right side of Figure 1.4.
Point P is on the rim of the wheel. The angular position of P is given
by the angle between some fixed line and the radius vector to P. If the
wheel is turning, after a time dt point P will have moved a distance
ds to P 0 . Recall from geometry that ds = rdθ where dθ (in radians of
course!) is the angle subtended by the arc P P 0 = ds. Point P moves
with speed
ds rdθ
v= = = rω.
dt dt
The speed of point P is called the “tangential speed” because instan-
taneously P is moving tangent to the rim of the wheel. It is sometimes
convenient to write the tangential speed as vT . Then
vT = rω. (1.12)
Taking the time derivative of vT yields the tangential acceleration
aT ,
d2 θ
 
dv d dθ
aT = = r = r 2,
dt dt dt dt
dω
where we used the fact that r is constant. But dt
= α, so
aT = rα. (1.13)

Exercise 1.7. A wheel initially spinning at ω0 = 50.0 rad/s comes

to a halt in 20.0 seconds. Determine the constant angular acceleration
and the number of revolutions it makes before stopping. Answers: -2.5
rad/s2 ; 79.6 rev.
 1.5. ROTATIONAL MOTION 17

1.5.1. Rotational Dynamics. Rotational dynamics is the anal-

ysis of the rotational motion of a body subjected to external torques.
Consider a body constrained to rotate about a fixed axis as shown
in Figure 1.5. I drew the body in the shape of a plane lamina for
simplicity. Let a force F act on the body at a point on its rim. Let
us assume (again for simplicity) that F is perpendicular to the axis
of rotation. The point of application of F is specified by the vector r
whose origin is at the axis of rotation.
The body cannot accelerate linearly because the axis is fixed. The
applied force causes the body to rotate about the axis. The tendency
of a force to cause a rotation is called the moment of the force, or more
commonly, the torque. Just as you can think of a force as a pull, you
can think of a torque as a twist.
The ability of a force to produce a rotation depends not only on the
magnitude of the force, but also on its direction and on the location of
the point where the force is applied to the body.
To define torque draw a line having the direction of the force and
passing through the point of application. This is called the “line of
action of the force.” (See Figure 1.6.) Next draw a line that starts at
the axis of rotation and intersects the line of action at a 90◦ angle.
This line is called the lever arm. The lever arm is the shortest distance
from the axis of rotation to the line of action of the force. Its length
is r sin θ where θ is the angle between r and F. The magnitude of the

Axis of
rotation

r
F

Figure 1.5. Illustration of torque. The laminar body

is free to rotate around the fixed axis of rotation. The
vector r, with origin at the axis, specifies the point of
application of the force.
18 1. A BRIEF REVIEW OF INTRODUCTORY CONCEPTS

Figure 1.6. Definition of lever arm. The axis of rota-

tion is perpendicular to the plane of the figure.

torque is given by the relation:

Torque = Force times Lever Arm.
This elementary definition of torque is complicated and cumber-
some. Fortunately, vector concepts allow us to simplify the definition.
Torque (N) is the cross product of r and F, thus:

N ≡ r × F. (1.14)
In Equation (1.14) r is the vector from the axis of rotation to the
point of application of the force. The direction of the torque is perpen-
dicular to the plane defined by r and F and it is usually represented
as a vector along the axis of rotation.

Exercise 1.8. Show that a given force applied at any point along
the line of action yields the same torque.
Exercise 1.9. An athlete is holding a 2.5 meter pole by one end.
The pole makes an angle of 60◦ with the horizontal. The mass of the
pole is 4 kg. Determine the torque exerted by the pole on the athlete’s
hand. (The mass of the pole can be assumed to be concentrated at the
center of mass.) Answer: Torque = 24.5 N m.
 1.5. ROTATIONAL MOTION 19

1.5.2. Statics. This section is an overview of some of the basic

concepts of statics.
Newton’s first law tells us that if no external force acts on a body,
it will move at constant velocity. That is, zero force implies zero ac-
celeration. (This is also a consequence of the Newton’s second law.)
Consequently, a body that is not accelerating must have no net exter-
nal force acting on it. Similarly, a body that is rotating at a constant
angular velocity has no angular acceleration. Such a body has no net
external torque acting on it.
A body that has no linear acceleration and no angular acceleration
is said to be in equilibrium. The conditions for equilibrium are:
X
Fi = 0, (1.15)
and
X
Ni = 0, (1.16)
where Fi and Ni are the net external forces and torques acting on the
system.
A body in equilibrium has constant linear velocity and constant
angular velocity. A particular but important special case occurs when
both the linear and angular velocity are zero. This is called static
equilibrium.
The general conditions for equilibrium are given by Equations (1.15)
and (1.16). I will state these conditions in words for a situation in which
all of the forces applied to a body lie in the same plane (but it is not
difficult to generalize). If a body is in equilibrium then:
1. The algebraic sum of the components of the forces in each of two
mutually perpendicular directions is zero, and,
2. The algebraic sum of the torques about any point in the plane
of the forces is zero.
The first condition is obvious. The second condition will be proved
in Chapter 7. It is very useful for solving simple statics problems
because it allows us to take the torques about any convenient point.
This point does not even have to be in the body.

Worked Example 1.2. The captain of the swimming team

(who weighs 600 N) is posing on the end of a 4 m diving board
that weighs 150 N. The board is bolted to two pillars that are 1.5
m apart. The first pillar is at the end of the board. Determine the
tension (or compression) acting on the bolts. See Figure 1.7.
20 1. A BRIEF REVIEW OF INTRODUCTORY CONCEPTS

Figure 1.7. What forces are exerted on the bolts that

hold the diving board on the pillars?

Solution: The board is in equilibrium so

forces up = forces down
and
torques clockwise (CW) = torques counterclockwise (CCW).
The first condition leads to
600 + 150 = F1 + F2 .
The second condition leads to
600 × 4 + 150 × 2 = F2 × 1.5.
Therefore, F2 = 1800 N (up) and F1 = –1050N (down).

1.5.3. Rotational Kinetic Energy. The rotational kinetic en-

ergy of an extended rigid body is obtained (at least conceptually) by
summing over the kinetic energy of each particle in the body. If the
body is rotating about a fixed axis and if the body is symmetrical
about this axis, the rotational kinetic energy is given by a fairly simple
formula. Recall that every particle in the body is moving in a circle,
and that the linear velocity v of the particle is related to the angular
velocity of the body, ω, according to v = ωr⊥ where r⊥ is the per-
pendicular distance from the axis of rotation to the particle. So the
kinetic energy of the particle is T = 21 mv 2 = 12 m(ωr⊥ )2 . The total
energy of the body will be the sum over all of the particles (molecules)
in the body. If we denote the mass of the ith particle by mi and its
 1.5. ROTATIONAL MOTION 21

perpendicular distance from the axis by ri⊥ , we have

X1
2
T = mi ri⊥ ω2
i
2
2
P
The quantity i mi ri⊥ is called the moment of inertia of the body and
is denoted I. Therefore the rotational kinetic energy can be expressed
as

1
T = Iω 2 . (1.17)
2
Note the similarity with the translational kinetic energy T = 21 mv 2 .

Exercise 1.10. A disk of mass M and radius R is initially at rest. It

is acted upon by a constant torque N for a time T. Determine the final
angular velocity of the disk and the work required to spin it up to this
final state. If the energy was supplied by a motor, what was the average
power output of the motor? Answer: Average Power = N 2 T /M R2 .

1.5.4. Angular Momentum. In general, the equations for linear

motion go over to the equations for rotational motion if we substitute
I for m and ω for v. Thus, for example, the magnitude of the angular
momentum of a single particle of mass m moving in a circle of radius
r with speed v will be

l = Iω, (1.18)
2
where I = mr .
This relation can also be expressed as l = mr2 ω or as l = mvr.
But angular momentum is a vector, so
l = Iω
where ω is the angular velocity vector, directed along the axis of rota-
tion.
For an extended rigid body the total angular momentum is usually
denoted by L and, as you might expect, is given by
L = Iω
2
P
where now I = i mi ri⊥ .
22 1. A BRIEF REVIEW OF INTRODUCTORY CONCEPTS

The rotational analogue of Newton’s second law is obtained by sub-

stituting torque for force, moment of inertia for mass and angular ac-
celeration for linear acceleration. This yields
N = Iα. (1.19)
Recall that Newton’s second law can be expressed as
dp
F= .
dt
Similarly, for rotational motion we can write
dL
N= .
dt
Just as linear momentum is constant if the net force is zero, so too,
angular momentum is constant if the net torque is zero.

Exercise 1.11. A spinning ice skater speeds up by pulling her

hands to the side of her body. Approximate the ice skater by a cylinder
of radius 10 cm and mass 30 kg. Her hands are two point masses of
0.25 kg each and her (massless) arms extend 90 cm from her torso (or
1 m from the axis of rotation). If she was initially rotating at 2 rev/s,
what is her final angular velocity? Answer: 8.4 rev/s.

1.6. Summary
The position of a particle is given by the vector r. Velocity is
defined by
dr
v= ,
dt
and the acceleration is
dv
a= .
dt
Newton’s second law for a particle with constant mass is
F = ma. (1.20)
Work is defined by the integral of the dot product of force and
displacement, Z
W = F · dr. (1.21)
 1.7. PROBLEMS 23

Kinetic energy is given by

1
T = mv 2 ,
2
for linear motion, and
1
T = Iω 2 ,
2
for rotational motion.
Potential energy of an object raised a height h above the Earth’s
surface is
V = mgh.
The potential energy stored in a spring is
1
V = kx2 .
2
The linear momentum of a particle of mass m moving at velocity v
is defined as
p = mv. (1.22)
The relation between angular velocity and linear velocity is given
by the cross product:
v = ω × r, (1.23)
where ω is a vector pointing along the axis of rotation whose magnitude
is equal to the angular velocity.
Torque is defined by
N = r × F. (1.24)
The angular momentum of a particle can be written as
l = r × p. (1.25)
The rotational version of Newton’s Second Law is
dl
N= . (1.26)
dt

1.7. Problems
Problem 1.1. Two ships are sailing in a thick fog. Initially, ship
A is 10 miles north of ship B. Ship A sails directly east at 30 miles
per hour. Ship B sails due east at constant speed vB then turns and
sails due north at the same speed. After two hours, the ships collide.
Determine vB .
24 1. A BRIEF REVIEW OF INTRODUCTORY CONCEPTS

Problem 1.2. You carefully observe an object moving along the

x−axis and determine that its position as a function of time is given
by
x(t) = 2t − 3t2 + t3 .
(a) What is the position at time t = 2 seconds?
(b) What is the velocity at time t = 2 seconds?
(c) What is the acceleration at time t = 2 seconds?
(d) How far did it travel between times t = 0 and t = 2 seconds?
(Note: Distance, not displacement! It might be helpful to plot x vs t
for 0 ≤ t ≤ 2.)
Problem 1.3. The first train leaves the station and accelerates at a
constant rate to its maximum speed of 100 km/hr, reaching this speed
at a distance of 2 km from the station. Five minutes later, a second
train leaves the station and accelerates to 100 km/hr in 4 km. What is
the distance between the two trains when they both reach maximum
speed?
Problem 1.4. A rocket is trying to land on a barge in ocean. The
rocket engines deliver an upward force of 36,000 N and the rocket is
observed to be descending at a constant safe speed of 3 m/s when it is
100 m above the barge. But suddenly there is a malfunction and the
rocket engine thrust is reduced to 30,000 N.
(a) What is the mass of the rocket?
(b) What is the acceleration of the rocket after the malfunction?
(c) Assume this acceleration is maintained constant during the final
descent. What is the speed of the rocket when it contacts the barge?
Problem 1.5. You travel a distance d in time t. (a) If you travel
at speed v1 for half the time and at v2 for the other half of the time,
what is your average speed? (b) If you travel at speed v1 for half the
distance and at speed v2 for the other half of the distance, what is your
average speed?
Problem 1.6. There is a long straight road out in the desert and
it goes through a small town that has just one police car. The police
car accelerates at 2 m/sec2 unitl it reaches a maximum speed of 200
km/hour. A car full of escaped criminals skpeeds through the town
at its top speed which is 150 km/hour. The police car, starting from
rest, gives chase. How far from the town do the police catch up to the
criminals?
Problem 1.7. A police car is at rest on the side of the road when
a wild teenager comes speeding by at 75 miles per hour. The police
 1.7. PROBLEMS 25

car starts immediately and accelerates at 8 miles per hour per second.
At that same moment the teenager steps on the gas, but his car only
accelerates at 2 miles per hour per second. (a) How far from the starting
point does the police car overtake the speeder? (b) How fast are they
going at that time? (c) Why is the speed you calculated for the police
car unrealistic?
Problem 1.8. A brick is on a wooden plank that is resting on a
table. One end of the plank is slowly raised so that it forms an angle
θ with the horizontal table top. When θ = 60◦ the brick starts to slide
down the plane. (a) Draw the free body diagram. (b) Determine the
coefficient of static friction between brick and plank. (c) If the coef-
ficient of sliding friction is one half of the coefficient of static friction,
determine the acceleration of the block.
Problem 1.9. A block of mass M is on an inclined plane with a
coefficient of sliding friction µ. At a given instant of time the block is
at some point P on the plane and is moving up the plane with a speed
v0 . (a) Obtain the time for the block to reach its highest point relative
to P. (b) Obtain the time for the block to slide back down to point P.
(c) Obtain an expression for the velocity of the block at the time it
returns to P.
Problem 1.10. A toy rocket burns fuel for 1.5 seconds. During
that time, it accelerates upwards at 60 m/s2 . It then coasts upwards
to some maximum altitude before falling back down. Determine the
maximum altitude reached.
Problem 1.11. Atwood’s machine consists of two weights (M1 and
M2 ) suspended at the ends of a string that passes over a pulley. Assume
massless, inextensible strings and a frictionless pulley. Let M1 =6 kg
and M2 = 5.5 kg. The masses are released from rest. Determine the
distance descended by the 6 kg mass when its velocity reaches 0.5 m/s.
Problem 1.12. (a) Determine the rotational kinetic energy of a
wheel of your bicycle when your linear speed is 20 km/hour. You
may assume the wheel is a hoop of mass 1.5 kg and radius 30 cm. (b)
Compare your result with the translational kinetic energy of the wheel.
(c) Give a theoretical explanation for the relationship found in parts
(a) and (b).
Problem 1.13. The frictional force between water and seabed in
shallow seas causes an increase in the day by about 1 ms/century.
Determine the torque that causes this change. Assume the Earth is a
uniform sphere.
26 1. A BRIEF REVIEW OF INTRODUCTORY CONCEPTS

Problem 1.14. A meter stick has a pivot at one end. It is found

to be in static equilibrium when acted upon by three forces that act
at different points along the meter stick and act in different directions.
We conclude that the net torque about the pivot is zero. Prove that
the net torque about any other (arbitrary) point is also zero.
Problem 1.15. To build the pyramids it was necessary to pull
heavy stones up inclined planes. Suppose a 2000 kg stone was dragged
up a 20◦ incline at a speed of 0.25 m/s by a gang of 20 laborers. The
coefficient of kinetic friction between the stone and the incline was 0.4.
How much power was exerted by each laborer?
Problem 1.16. Two wooden blocks, both of mass 3 kg, are sliding
in the same direction at 2 m/s on a frictionless horizontal surface. Let
M1 = M2 = 3 kg and assume their speed is 2 m/s. A third block of
mass 1 kg is also sliding in the same direction at a speed of 10 m/s,
and it collides with the trailing 3 kg block. The third block is covered
with a sticky, gooey substance, so it sticks to the trailing block. This
combination block catches up with and collides elastically with the
leading 3 kg block. Determine the final speed of the leading block.
(Note that this is a one-dimensional problem. You may find it easier
to solve the problem if you insert numerical values sooner rather than
later.)
Problem 1.17. You are driving along at 60 mph (=100 km/hr).
Your car’s wheels have a radius of 35 cm. (a) Determine the angular
velocity of the wheels. (b) What is the angular displacement of a wheel
when you travel 1 km?
Problem 1.18. Aeronautical engineers have developed “tip jet”
helicopters in which small jet engines are attached to the tips of the
rotor. One such helicopter was powered by two ramjets. For a ramjet
to develop thrust, it needs to be moving through the air quite rapidly,
and is not efficient until it is moving at about 1000 km/hr. Assume
the rotor diameter is 10 meters. Determine the angular speed of the
rotor when the ramjet is moving through the air at 1000 km/hr.
Problem 1.19. Figure 1.8 shows an angle iron of mass M hang-
ing from the pivot at point P. The three segments have equal length.
Determine whether or not this is a stable equilibrium orientation.

Problem 1.20. A solid ball of mass M and radius R rolls down an

inclined plane. What is its translational speed when it has descended
a vertical distance h?
 1.7. PROBLEMS 27

Figure 1.8. An angle iron. Determine equilibrium.

Problem 1.21. A disk of mass 72 kg and radius 50 cm is rotating

at 2000 rpm. (a) Determine its angular momentum. (b) If acted upon
by a retarding force of 20 N acting tangent to the rim of the disk,
determine the time required to stop the disk.

COMPUTATIONAL PROJECTS: The following problems can

be solved easily if you know a computer language such as MATLAB,
Python, FORTRAN, or C++. A few can also be solved using a spread
sheet program such as EXCEL. Although some of the computational
projects in this book can be solved analytically, they will require a
significant amount of brainless labor that can be done much more con-
veniently using a computer.
Computational Project 1.1. The position of a particle as a
function of time is given by x = 5t3 − 2t (meters). Plot the position
as a function of time for the interval t = −5 sec to t = +10 sec. Using
the relationship v = ∆x/∆t, obtain the average velocity at one second
intervals. Plot the average velocity as a function of time and on the
same graph, plot the analytical expression for the velocity.
Computational Project 1.2. This is a more realistic version
of problem 1.4. In that problem a teenager driving at 33.5 m/s (∼75
mph) speeds past a parked policeman. The policeman and the teenager
then accelerate at given constant rates and you are required to deter-
mine how far from the starting point the policeman catches up to the
teenager. To make the problem somewhat more realistic, assume the
teenager accelerates at aT = kT e−bT t and the policeman accelerates at
aP = kP e−bP t , where kT = 2.0 m/s2 and kP = 3.5 m/s2 . By plotting
the positions as functions of time for various values of bT and bP obtain
reasonable values for these constants. (Make sure your answers are
28 1. A BRIEF REVIEW OF INTRODUCTORY CONCEPTS

reasonable. Obviously, the policeman will catch the teenager, but your
answer is not reasonable if the distance required is hundreds of miles!)
Chapter 2
Kinematics

Kinematics in one dimension is fairly simple because position, ve-

locity and acceleration can be treated as scalars. Motion in a straight
line and rotation about a fixed axis are examples of one-dimensional
motion. The motion of a projectile in a constant gravitational field can
be resolved into two linked one-dimensional problems. Going to two
or three dimensions complicates the problem significantly because now
you need to express the basic quantities as vectors. In this chapter you
will learn the relations between the position, velocity and acceleration
vectors in the three main coordinate systems used in physics: Cartesian
coordinates, cylindrical coordinates and spherical coordinates. (Rota-
tional motion in three dimensions is considerably more difficult than
linear motion and will be left to Chapters 13 and 16.)
Although some of the material presented in this chapter will be
familiar to you, you will also find many new concepts. These concepts
are used throughout the course, so please make sure you understand
this chapter thoroughly. Be aware that many students find this material
rather difficult.
I think it is important for you as a physicist to know something
about the people upon whose shoulders you are standing, so in this
book I have included a few sections marked “Optional Historical Note.”
You will encounter very little physics in these sections, but you might
find them interesting and helpful in placing a few famous physicists in
historical context. The first Historical Note describes the life of the
person who invented your profession.

29
30 2. KINEMATICS

2.1. Galileo Galilei (Optional Historical Note)

Galileo Galilei (1564 - 1642) was a brilliant but difficult man whose
studies of the physical world launched a scientific and cultural revolu-
tion. This cantankerous Italian genius was the first modern scientist.
He rejected authority and based his conclusions on observation, exper-
iment, and rational analysis. His ideas were opposed by the powers
of the Church, and he ended his days under house arrest. Neverthe-
less, his view of the universe and his methods for discovering scientific
truth eventually won out and today every physicist is an intellectual
descendent of Galileo.1
Galileo did not invent the telescope, but he built one and was prob-
ably the first person to make a scientific study of what he saw in the
sky. In short order, he discovered that the Moon has mountains and
valleys, that the Milky Way is made up of individual stars, that Jupiter
has satellites orbiting it, and that the Sun rotates and has spots on its
surface. He became the first “celebrity scientist.” He got into trouble,
however, because he was outspoken and had little patience with peo-
ple who put faith above reason. He was particularly offensive towards
those who believed the Earth was the center of the universe and that
the Sun, planets and stars revolved around it. This “geocentric” uni-
verse was based on the philosophy of Aristotle which had been adapted
to Christianity by Thomas Aquinas and was subsequently accepted as
the official philosophy of the Church. Therefore, Galileo’s outspoken
attacks on the geocentric theory were considered by many to be hereti-
cal. Galileo was eventually called to Rome and ordered by Church
authorities to neither teach nor defend the Copernican theory that the
Sun is the center of the Universe. Shortly thereafter Galileo wrote a
book in the form of a dialogue between three men. In this book he
ridiculed the Aristotelians and even put some arguments used by the
Pope in the mouth of a character named “Simplicio” who was pictured
as a bit of a dunce. Galileo was called back to Rome by the Inquisition
and was eventually condemned to life imprisonment, to take place in
own home. He was 70 years old at the time. He continued to work,
publishing his last book, “Discourses and Mathematical Demonstra-
tions Relating to Two New Sciences” in 1638. This book describes
much of the work in physics that he had carried out over the previous
30 years. He died at age seventy eight.
Galileo’s life is an instructive example of the conflict between a
brilliant individual and the powers that rule society. There is no doubt
1Anexcellent biography of Galileo is “Galileo’s Daughter” by Dava Sobel,
Walker and Co., New York, 1999.
 2.2. THE PRINCIPLE OF INERTIA 31

that Galileo changed the world and our understanding of it. Some
people condemn the Church authorities for trying to stop scientific
progress; others are more understanding of the authorities who saw
their entire world view threatened by a revolutionary iconoclast.

2.2. The Principle of Inertia

Among his many accomplishments, Galileo was arguably the first
person to truly understand motion. He performed a series of experi-
ments with pendulums of different lengths, as well as falling bodies and
balls rolling down inclined planes. His experiments with these simple
devices allowed him to grasp the essential aspects of motion, something
that had eluded the greatest minds of previous ages. Prior to Galileo’s
work many respected philosophers (including Aristotle) described mo-
tion in ways that modern scientists consider to be either meaningless
or just plain silly. For example, they said, “Water flows downhill be-
cause it is trying to return to its natural place, which is the ocean.”
Or they said, “Smoke rises because smoke is air and the natural place
for air is up in the sky.” They explained the motion of the stars and
planets with the preposterous statement that, “The celestial bodies are
perfect, therefore they move about the Earth in perfect circles.”
A crucial element in this erroneous theory of motion was the idea
that an object will move to its “natural place” and then remain at
rest in that place. This was called, of course, “natural motion.” For
something to move away from its natural place, it was necessary to give
it a push or a pull. This kind of motion, requiring a force, they called
“violent motion.”
Galileo’s most significant insight was that objects do not tend to
come to rest, but rather they tend to keep on moving. In fact they tend
to keep on moving at a constant speed in a straight line. (We call this
uniform motion.) The property of material bodies to maintain their
state of motion is called inertia. As you recall from your introductory
course in physics, the law of inertia states that:
A body will remain in uniform motion
as long as no net external force acts on it.
This basic principle has come to be known as Newton’s first law.
We will return to it later.
Although his equipment was crude, Galileo’s experiments showed
him that time is an essential component of motion. Mechanical clocks
had not yet been invented, but Galileo devised a number of ingenious
ways to estimate time intervals. He counted his pulse or watched the
32 2. KINEMATICS

oscillations of a pendulum or weighed the amount of water dripping

out of a container.
Time is a very difficult concept to comprehend. Everyone has an
intuitive, qualitative understanding of time. If you think about it, you
will realize that your personal definition of time has some connection
to the motion of material bodies. The length of a day is related to the
rotation of the Earth, and a year is the Earth’s orbital period. Most
people describe an hour and a minute in terms of the motion of the
hands on a clock. Scientists define the second in terms of the vibrations
of certain molecules. Is it possible to define time without reference to
motion? If all objects in the universe were perfectly still, would time
exist? Newton avoided these philosophical questions by assuming time
to be an absolute parameter that is continuously changing at a constant
rate. Einstein gave an even simpler definition when he said, “Time is
what you measure with a clock.” We shall adopt Newton’s point of
view at present, then come back for a deeper look at the question of
time when we consider Einstein’s theory of relativity in Chapter 19.2
As mentioned above, kinematics is the study of the relationships
between position, velocity, acceleration, and time. Galileo and his dis-
ciples grappled with these concepts. Fortunately, we can use many
powerful mathematical tools that were not available to Galileo. For ex-
ample, the calculus was invented by Newton and Leibniz after Galileo’s
death, and vector analysis was not developed until about 150 years ago.

2.3. Basic Concepts in Kinematics

Let me remind you that the general relations between position (r),
velocity (v), and acceleration (a) are given by
dr
v= = ṙ,
dt
dv
a= = v̇ = r̈.
dt
One-dimensional motion refers to any kind of motion for which the
position can be described in terms of a single parameter. For example,
the position of a car driving down a curving road can be specified
by its distance from the starting point. Similarly, the position of a
bead sliding along a wire can be specified by the distance to the bead
from some given point on the wire. A single number is enough to
2Some physicists believe that time, in the Newtonian sense of one event happen-
ing before another event, does not apply in the realm of quantum mechanics. For
an interesting discussion of this problem see the article by Charles Seife, “Quantum
Physics: Spooky Twins Survive Einsteinian Torture,” Science, 294, 1265, (2001).
 2.3. BASIC CONCEPTS IN KINEMATICS 33

completely specify the position of the bead, no matter how the wire
may be twisted or curled. However, the linear distance may not be the
most convenient parameter to use. For example, the position of a bead
sliding on a circular hoop might best be described in terms of an angle.
The simplest example of one-dimensional motion is a particle mov-
ing along a straight line, such as the x-axis. The acceleration is
dv
a= .
dt
Here I wrote a and v rather than a and v because if the motion is one-
dimensional there is no need to use vectors. Multiplying both sides by
dt gives the following very simple differential equation:
dv = adt.
To solve, you integrate both sides:
Z vf Z tf
dv = adt,
v0 t0

where the limits indicate that the time runs from t0 to tf (that is,
“initial” time to “final” time), and the initial and final values of the
velocity are v0 and vf .
The quantity t0 represents the time when the particle was at the
initial position x0 . Since the starting time for any process is arbitrary
(it just depends on when you push the button on the stopwatch) it is
nearly always set equal to zero (t0 = 0). R
Now you can go no further. You cannot evaluate the integral adt
because you do not have an expression for the acceleration as a function
of time. But if you do know the acceleration as a function of time,
a = a(t), then you can carry out the integration. When you do so,
you will obtain an expression for the velocity as a function of time,3
v = v(t). Once you have determined the velocity, you can continue the
analysis and write

dx = v(t)dt,
which follows immediately from the definition of velocity.
Integrating once again, you will obtain an expression for the posi-
tion as a function of time,
x = x(t),
or, more explicitly,
x = x(t, v0 , x0 ).
3To be precise, this will give you an expression for the velocity in terms of the
time, the initial time and the initial velocity, v = v(t, t0 , v0 ).
34 2. KINEMATICS

2.3.1. Motion in One Dimension with Constant Accelera-

tion. As discussed in Section 1.1.1, if the acceleration is constant, we
can integrate dv = adt to obtain
v(t) = v0 + at, (2.1)
and
1
x(t) = x0 + v0 t + at2 . (2.2)
2
I am sure you are very familiar with Equations (2.1) and (2.2) from
your introductory physics course. In fact, you probably memorized
them. However, it is more important at this stage in your physics
career to understand the process for obtaining these equations.

Exercise 2.1. Starting with x = v0 t + 12 at2 and v = v0 + at, obtain

the relations
1
x = (v + v0 )t, (2.3)
2
and
2ax = v 2 − v02 . (2.4)
Are these expressions valid if the acceleration is not constant?

I must emphasize that the formulas obtained above (Equations 2.1

through 2.4) are only valid for constant acceleration. If the acceleration
is a function of time or velocity or position, then you must go through
the entire procedure to obtain the correct expressions for v(t) and x(t).
2.3.2. Projectile Motion (Motion in a Plane). A projectile is
any object that is launched with some initial velocity and then moves
under the action of the gravitational force which imparts to it a con-
stant downward acceleration.
The horizontal motion is motion at constant velocity, and therefore
x = x0 + v0x t,
and the vertical motion is motion at constant acceleration, and there-
fore
1
y = y0 + v0y t − gt2 ,
2
where g = 9.8 m/s.
It is helpful to remember that at the top of the trajectory the ver-
tical velocity (vy ) is zero. This gives us a useful way to determine
 2.3. BASIC CONCEPTS IN KINEMATICS 35

the time for the projectile to reach the top of its trajectory. Since
vy = v0y + ay t we have
0 = v0y − gttop
so
ttop = v0y /g.
Worked Example 2.1. Consider the projectile motion of a
cannon ball. Assume the initial speed is v0 and the cannon is aimed
at an angle θ above the horizontal. Determine the range.
Solution: The range (R) is the horizontal distance traveled
by the projectile along a flat horizontal plane; it is the value of
x when y = 0. The initial velocity components are v0x = v0 cos θ
and v0y = v0 sin θ. Since the total time of flight is twice the time
required to reach the top of the trajectory, the range is given by
x(t = 2ttop ). It is easy to obtain a formula for the range in terms
of the initial velocity as follows:
vy = v0y − gt.
But vy = 0 when t = ttop so 0 = v0y − gttop = v0 sin θ − gttop .
Consequently,
ttop = (v0 sin θ) /g.
Therefore,
R = v0x (2ttop ) = 2v0 cos θ (v0 sin θ) /g,
v02
R=2 sin θ cos θ. (2.5)
g

Our analysis of projectile motion made two assumptions: (1) The

acceleration of gravity is constant, and (2) There are no horizontal
forces acting on the projectile. These assumptions mean that we are
assuming a flat, non-rotating Earth, and that there is no air resis-
tance acting on the body. These seemingly preposterous assumptions
(a flat, motionless, airless Earth!) illustrate an important technique
in physics. When faced with a difficult problem, we solve a similar,
simpler problem. Frequently, the solution to the simple problem is
sufficiently accurate. The solution of the simple problem is often re-
ferred to as the “zeroth order” approximation to the solution of the
real problem. Later in this course you will learn techniques for includ-
ing the factors that were ignored in the zeroth order approximation to
get results that are nearer and nearer to the exact solution. You will
appreciate more fully how this technique works when you study projec-
tile motion on a rotating Earth in Chapter 13. As you can imagine, a
36 2. KINEMATICS

working physicist is continually faced with very complicated problems.

It is part of a physicist’s education to learn what is fundamentally im-
portant and what can be left until later as a refinement (or “higher
order approximation”). For example, you can estimate the trajectory
of a thrown baseball with reasonable accuracy by assuming a flat, non-
rotating Earth, but if you want to calculate the trajectory of a rocket,
those factors will have to be included.

Exercise 2.2. Determine the elevation angle for a projectile such

that the range will be maximized. (Hint: the maximum value of a
function is determined by setting its derivative to zero.) Answer: 45◦ .
Exercise 2.3. An airplane traveling at 900 km/hr at 5,000 m
altitude is directly over the target when it drops a bomb. How far
from the target does the bomb hit? Answer: ' 8 km.

In working problems involving projectile motion in a uniform grav-

itational field, keep in mind that the horizontal motion has constant
velocity (vx = constant) and the vertical motion has constant accel-
eration (ay = constant). Projectile motion is a combination of two
independent one-dimensional motions. These are related to one an-
other through the time. If you eliminate the time from the equations,
you obtain an equation for the trajectory of the projectile as shown in
the following example.

Worked Example 2.2. Prove that the path of a projectile in

a uniform gravitational field is a parabola
Solution: The equation of a parabola whose axis is parallel to
the y axis in Cartesian coordinates is
y = ax2 + bx + c.
In section 2.3.2 we showed that for a projectile
1
y = − gt2 + v0y t + y0 ,
2
x = v0x t + x0 .
Therefore,
x − x0
t= ,
v0x
 2.3. BASIC CONCEPTS IN KINEMATICS 37

and consequently
g (x − x0 )2 x − x0
y=− 2
+ v0y + y0 ,
2 v0x v0x
g v0y v0y
y = − 2 (x2 − 2xx0 + x20 ) + x− x0 ,
2v0x v0x v0x
g g v0y gx2 v0y
y = − 2 x2 + [( 2x0 + )]x + (− 20 − x0 + y0 ).
2v0x 2v0x v0x 2v0x v0x
This has the required form as can more easily be appreciated if we
set x0 = 0 and y0 = 0 to obtain
g 2 v0y
y=− x + x.
2v0x v0x

Although projectile problems tend to be straightforward, sometimes

it is necessary to use a more subtle approach, as illustrated in the
following example.
Worked Example 2.3. A cannon that fires shells with a muz-
zle speed v0 is mounted on the top of a cliff a height h above a level
plain. Show that the angle at which thepcannon should be aimed
to give maximum range is θ = sin−1 (v0 / 2(v02 + gh)).
Solution: I am including this problem to show that a seem-
ingly simple projectile problem can require considerable ingenuity
to obtain a solution. The obvious approach would be to attempt
to solve the problem by using the fact that if t is the time of flight
(tf ), then the value of x is the range (R) and the value of y is
−h (where we assume that the coordinate origin (0, 0) is at the
cannon). That is
R = (v0 cos θ) (tf ),
1
−h = (v0 sin θ) (tf ) − g(tf )2 .
2
You can solve the first equation for tf , obtaining tf = R/(v0 cos θ).
Plugging into the second equation leads to
 2
gR 1
− h = R tan θ − 2
. (2.6)
2v0 cos2 θ
Now, you could solve this quadratic equation to obtain an expres-
sion for R as a function of θ. Then taking the derivative of R with
respect to θ and setting it equal to zero gives an expression for θ
38 2. KINEMATICS

that maximizes R. However, the resultant expression is so compli-

cated that you will find it extremely difficult to solve for θ. (You
might want to give it a try.)
However, the problem can be solved fairly simply by using a
clever trick. Instead of solving for R and taking its derivative
respect to θ, note that
dR dR dh
= .
dθ dh dθ
This expression will be zero if dhdθ
= 0. So going back to the expres-
sion above for −h, equation (2.6), we have
  2 
dh d gR 1
− = R tan θ − = 0,
dθ dθ 2v02 cos2 θ
gR
0 = 1 − 2 tan θ,
v0
2
v0
R= .
g tan θ
Plugging this into equation (2.6) yields
2
v02 v02 v02 v02

g
−h = − 2 = − .
g 2v0 cos2 θ g tan θ g 2g sin2 θ
Therefore
v02 v02
= + h,
2g sin2 θ g
v0
sin θ = p 2 .
2(v0 + gh)

2.3.3. Rotation About a Fixed Axis. Rotational motion about

a fixed axis (as discussed in Section 1.5) is one dimensional motion
described by the variables θ, ω and α where
dθ
ω= ,
dt
dω d2 θ
α= = 2.
dt dt
Worked Example 2.4. A physics student is carefully observ-
ing a wheel while it is being spin balanced. The wheel speeds up for
10 seconds and subsequently rotates at a constant angular speed
of 50 rad/sec. Going home, the student decides to write an expres-
sion for the acceleration of the wheel that would account for this
 2.3. BASIC CONCEPTS IN KINEMATICS 39

behavior, and comes up with

α = bte−ct (rad/s2 ).
Obtain an expression for the angular velocity (ω) as a function of
time. What are reasonable values for b and c?
Solution: Using the definition α = dωdt
,
Z ω(t) Z t
dω = αdt
ω0 t=0
Z t
ω(t) − ω0 = bte−ct dt
0
Note that ω0 = 0. Integrating by parts,
 t
1 −ct
ω(t) = b 2 e (−ct − 1)
c 0
b
= − 2 (e−ct )(ct + 1) − e0 (1)
 
c
b 
= 2 1 − e−ct (ct + 1) ,

c
Note that at time t = 0 the angular speed is zero. It increases
rapidly for a time, and then asymptotically approaches a maximum
value of b/c2 . If the time to spin up to this value is 10 seconds, then
at t = 10 the term e−ct ' 0, suggesting a value of c near unity. Let
c = 1. The final angular speed of the wheel is 50 radians/second so
b/c2 = 50, and b ' 50. It is instructive to plot α vs t and ω vs t.

2.3.4. The Relation between Linear and Rotational Mo-

tion. Imagine a child has tied a string to a rock and is whirling it
around in a circle. If the stone makes one revolution every T seconds,
it has an angular speed given by
2π
ω= .
T
It also has a linear speed that is tangent to the circular path (as is
obvious if the child lets go of the string). If the speed is constant, the
linear (or tangential) speed is given by
2πr
vT = ,
T
where r is the radius of the circular path. Clearly, the linear and
angular speeds are related by
vT = ωr.
40 2. KINEMATICS

We shall usually represent the angular velocity by a vector ω whose

magnitude is the angular speed ω and whose direction is perpendicular
to the plane of the motion. (If you curl the fingers of your right hand
in the direction of the motion, your thumb will point in the direction
of the vector ω. See Figure 2.1.)
Thus, the tangential velocity vT is related to the angular velocity
ω by

vT = ω × r. (2.7)

Figure 2.1. The direction of the vector angular velocity

is given by a right hand rule. The fingers of the right
hand point in the direction of the motion and the thumb
points in the direction of the vector ω.

It is interesting to note that rotational motion introduces two differ-

ent kinds of acceleration. To appreciate this, consider a rotating wheel.
If the wheel speeds up or slows down, it has an angular acceleration
given by α = dω/dt. Now recall that linear acceleration is defined as
a change in the linear velocity vector. Therefore, a car moving in a
circle at constant speed is accelerating. This acceleration is due to the
change in the direction of the velocity vector. Similarly, a point on
a rotating body is accelerating even if the body rotates at a constant
angular speed. Figure 2.2 illustrates the change in the velocity vector
of a point on the rim of a wheel that is rotating about a fixed axis with
constant angular speed ω.
 2.3. BASIC CONCEPTS IN KINEMATICS 41

v(t+Δt)

Δv v(t)
Δv
v(t) v(t+Δt) Δθ
Δθ

Figure 2.2. Illustration of centripetal acceleration.

The sketch on the left represent the motion of a par-
ticle, and the sketch on the right is the hodograph, a plot
of the velocity vectors.

By definition4,
dv v(t + ∆t) − v(t) ∆v
a= = lim = lim .
dt ∆t→0 ∆t ∆t→0 ∆t

Tip-to-tail addition of the vector v(t + ∆t) and the vector −v(t)
shows that ∆v is a vector pointing towards the center of the circle.
To help you see this, on the left side of Figure 2.2, I drew the vector
∆v at a midpoint between the vectors v(t) and v(t + ∆t). You can
appreciate that the vector ∆v is pointing approximately towards the
center of the circle. As ∆t → 0, the vector ∆v points more and more
closely towards the center.
The same concept is shown in a different way on the right hand side
of Figure 2.2. If the angular velocity is constant then the speed (that
is, the magnitude of the velocity vector) is a constant equal to v, but
the direction of the vector changes with time. The tip of the velocity
vector traces out a circle, as indicated on the right hand side of Figure
2.2. (A plot showing the path of the tip of the velocity vector is called a
hodograph.) For circular motion at constant speed, the hodograph is a
circle of radius v. The vector ∆v is a chord of this circle. Transferring
the vector ∆v back to the left hand side of the figure shows that it
points toward the center of the circle.
Consequently, a particle moving at a constant speed in a circular
path is accelerating towards the center of the circle. We can determine
the magnitude of this acceleration by considering the hodograph again.
In the limit of infinitesimally small times, ∆θ → dθ, and ∆v → dv,
4In the rest of this section, the velocity v should be expressed as vT but I left
off the subscript to keep the notation simple. This is common practice and one
must distinguish between tangential and linear velocity from the context
42 2. KINEMATICS

and the chord approaches the subtended arc. From the basic relation
ds = rdθ, but applied to the hodograph, we see that in the limit, the
magnitude of ∆v is dv = vdθ. Therefore,

dv vdθ
a= = .
dt dt
But dθ = ds/r, so

v ds v2
a= = .
r dt r
That is, if a particle is moving in a circle at constant speed, it is
accelerated towards the center with a “centripetal” acceleration given
by
v2
ac = .
r
Since v = rω, this can be written in the equivalent form
ac = ω 2 r.
Note that a point in a rotating body can experience three different
types of acceleration:
(1) linear acceleration (if the body as a whole is accelerating),
(2) tangential acceleration (aT = rα; if the angular velocity of
the body is changing),
(3) centripetal acceleration (ac = v 2 /r = ω 2 r; due to the rota-
tion of the body).

Exercise 2.4. (a)Determine the centripetal acceleration of the Earth

due to its orbital motion about the Sun. Assume the orbit is circular.
(b) Determine the centripetal acceleration of a point on the equator
of Earth due to the rotational motion of the Earth. Answers: (a)
5.95 × 10−3 m/s2 , (b) 3.37 × 10−2 m/s2 .
Exercise 2.5. Show that the angles denoted ∆θ on the right and
left hand plots of Figure 2.2 are the same.
 2.4. THE POSITION OF A PARTICLE ON A PLANE 43

2.4. The Position of a Particle on a Plane

The position of a point P on a flat surface can be specified in a
variety of ways. One usually arbitrarily selects an origin and draws a
pair of perpendicular reference lines through it. (These are denoted x
and y in Figure 2.3.)

y P
5

3 r
2

1
q
x
1 2 3 4 5

Figure 2.3. The position of the point P in rectangu-

lar coordinates is (4, 5) and in polar coordinates it is
(6.4, 0.9) where the angular measure is given in radians.

The most useful way to describe the position of P is by a vector

(r) drawn from the origin to P, as in Figure 2.3. The magnitude of r is
denoted by r and its direction is given by θ, the angle the vector makes
with the x axis.
To describe the vector r we can either use its components (namely,
x and y, the Cartesian coordinates), or its magnitude and direction (r
and θ, the polar coordinates). These are equivalent descriptions of r,
so (x, y) and (r, θ) are related. A glance at Figure 2.3 shows that these
relations are
x = r cos θ,
y = r sin θ.
The inverse relations (r and θ in terms of x and y) are
p
r = x2 + y 2 ,
θ = tan−1 (y/x).
Equations such as these that relate two sets of coordinates are called
“transformation equations.”
In your study of mechanics, you will only be using three different
coordinate systems, namely: Cartesian, cylindrical, and spherical coor-
dinates. There are, however, many other coordinate systems, some be-
ing quite specialized and appropriate for only a few types of problems.
44 2. KINEMATICS

In early editions of his book “Mathematical Methods for Physicists,”

George Arfken5 included a long discussion describing fourteen differ-
ent coordinate systems. Just to give you a flavor of these, consider
the “bipolar coordinates” η and ζ that are related to the Cartesian
coordinates, x and y by the following transformation equations:

a sinh η a sin ζ
x= and y = .
cosh η − cos ζ cosh η − cos ζ
You will be happy to know that we will not be using this particular
set of coordinates! Fortunately, most coordinate transformations are
much simpler than this one.

Exercise 2.6. A particle moves along the x axis at a constant speed

of 3 m/s. Determine dr/dt and dθ/dt. Answer: (3 m/s, zero)
Exercise 2.7. A particle travels in a circle of radius 3 m at a
constant speed of 5 m/s. Determine dr/dt and dθ/dt. Answer: (zero,
5/3 rad/s)
Exercise 2.8. In bipolar coordinates a particle is at η = 0.5, ζ
= 0.5. Assuming a = 1 (meter) determine its position in plane polar
coordinates r, θ. Answer: (2.832 m, 0.744 rad)

2.5. Unit Vectors

Unit vectors are the building blocks of kinematics. They are vectors
that point in the increasing direction of the coordinates and have a
length of one unit. (Unit vectors are unitless and have unit length.)
Figure 2.4 illustrates the three unit vectors ı̂, ̂, k̂ associated with the
Cartesian coordinate system (x, y, z).
The coordinate system in Figure 2.4 is a right-handed coordinate
system. This means that if you point the fingers of your right hand
along the x-axis and bend them towards the y-axis, your thumb will
be pointing in the direction of the z-axis.
5“Mathematical Methods for Physicists,” 2nd ed., George Arfken, Academic
Press, New York, 1970. The discussion was dropped in later editions of the book
because high speed computers allow one to solve most physics problems using sim-
ple coordinate systems. Many of the more exotic coordinate systems that were
applicable only to one or two specific problems have fallen into disuse.
 2.5. UNIT VECTORS 45

z
^k
^i y
^j
x

Figure 2.4. The unit vectors ı̂, ̂, k̂ in the Cartesian

coordinate system. The coordinate system is right-
handed.

Note that the Cartesian unit vectors are mutually perpendicular or

orthogonal.
In your vector analysis course you learned the definition of the cross
product (or vector product) of two vectors. Recall that if
c=a×b
then the magnitude of the vector c (denoted |c| or c ) is
| c |=| a | | b | sin(a, b), (2.8)
where sin(a, b) is the sine of the angle between a and b, and the direc-
tion of c is given by the right hand rule.
The definition of the dot product (or scalar product) of two vectors
is:
a · b = | a | | b | cos(a, b). (2.9)
The cross (or vector) product and the dot (or scalar) product are the
two most common ways of multiplying vectors. Later in this course you
will learn another type of vector multiplication which involves mathe-
matical objects called dyadics.
Consider the dot product of the unit vectors. It is quite obvious
from Equation (2.9) that

ı̂ · ı̂ = 1, ̂ · ̂ = 1, k̂ · k̂ = 1,
ı̂ · ̂ = 0, ı̂ · k̂ = 0, ̂ · k̂ = 0.
These relations describe the orthogonality of the unit vectors. Similarly,
from the definition of the cross product, Equation (2.8), we find that

ı̂ × ı̂ = 0, ̂ × ̂ = 0, k̂ × k̂ = 0,
46 2. KINEMATICS

and
ı̂ × ̂ = k̂, ̂ × ı̂= −k̂
̂ × k̂ = ı̂, k̂ × ̂= −ı̂
k̂ × ı̂ = ̂, ı̂ × k̂= −̂
The last six relations are easily remembered with the help of the
mnemonic of Figure 2.5.

^ ^
i i

^
+ ^
j
^
k
- ^
j
k

Figure 2.5. A mnemonic for the cross product of unit

vectors. The sketches indicate, for example, that ı̂ × ̂
= + k̂, and ̂ × ı̂ = - k̂.

Exercise 2.9. Two vectors are given by A = 2ı̂ +3̂ +4k̂ and
B = 3ı̂ − 2 ̂ Determine A + B, A − B, A · B and A × B.
Exercise 2.10. Calculate the angle between the vector V=2ı̂+4̂+6k̂
and the vector W=2ı̂+2̂+2k̂. Answer: 22.21◦
Exercise 2.11. The direction angles of a vector are the angles
between the vector and the x, y, and z axes. Determine the direction
angles of the vector r = 6ı̂+5̂−2k̂. Answer: α = 41.0◦ , β = 51.67◦ ,
γ = 104.3◦ .
Exercise 2.12. A vector forms the diagonal of a cube of side a.
Express it in terms of the unit vectors.
Exercise 2.13. Show that the cross product of two vectors A and
B can
 be written as
 the following determinant:
 ı̂ ̂ k̂ 
 
 Ax Ay Az 
 
 Bx By Bz 
Exercise 2.14. Let A and B be the sides of a parallelogram. Show
that the area of the parallelogram is A×B.
 2.6. KINEMATICS IN TWO DIMENSIONS 47

2.6. Kinematics in Two Dimensions

2.6.1. Cartesian Coordinates. In two dimensions, the position
of a point can be specified by two numbers. When using Cartesian
coordinates we usually set z = 0 and define the plane of motion to be
the x-y plane. (As long as the motion is confined to a flat plane, we
can let the plane be the z = 0 surface with no loss of generality.) In
Cartesian coordinates the position of a particle is given by
r =xı̂ + y̂.
The definition of velocity is v = ṙ = dr/dt, so
d
v= (xı̂ + y̂).
dt
The Cartesian unit vectors (ı̂, ̂, k̂) are constant in magnitude and
direction so their time derivatives are zero. Consequently, writing ẋ for
dx/dt and ẏ for dy/dt, the velocity is given by
v = ẋı̂ + ẏ̂.
Using the definition of acceleration, a = v̇ we obtain,
a =ẍı̂ + ÿ̂.

2.6.2. Plane Polar Coordinates. Let us now go through the

same analysis as above but for the plane polar coordinates (r, θ). That
is, let us derive expressions for position, velocity and acceleration in
plane polar coordinates. Although the analysis is complicated, it is
very important for you to understand exactly what I am doing. The
ideas presented in this section are the basis for the rest of this chapter.
The first thing we need to do is to define unit vectors for the polar
coordinates. In Figure 2.6 the point P is located at (r, θ) where r is
the length of the vector r and is equal to the distance from the origin
to the point P. The angle between r and the x-axis is denoted θ. At
the tip of vector r you can see the two unit vectors r̂ and θ̂. These
have unit length. The direction of r̂ is the direction of increasing r.
The direction of θ̂ is the direction of increasing θ. This means that an
increase in θ will cause the vector r to rotate in the direction of θ̂.
As long as their magnitudes and directions are unchanged, vectors
are allowed to “slide.” We can slide r̂ and θ̂ to the origin and represent
them on a unit circle as shown in Figure 2.7.
48 2. KINEMATICS

y ^ ^r
q
P
r

^
j q
^ x
i

Figure 2.6. Definition of the plane polar unit vectors

r̂ and θ̂. The unit vector r̂ points in the direction of
increasing r, and θ̂ points in the direction of increasing
θ.

The unit vectors r̂ and θ̂ are not constant because they are not
associated with a set of fixed axes. Rather, they are associated with
a vector r that can change with time. If r changes in magnitude, this
does not affect r̂ or θ̂ but if r changes in direction, then the direction
of the unit vectors r̂ and θ̂ will change. This means that r̂ and θ̂ are
functions of θ. In mathematical form we express this as
r̂ = r̂(θ),
and
θ̂ = θ̂(θ).
The position of a point in plane polar coordinates is given quite
simply by
r = rr̂.

^
j
^ ^
q r
q ^
i

Figure 2.7. The unit vectors r̂ and θ̂ and the unit cir-
cle. All the unit vectors originate at the center of the
circle and all have magnitude unity, so they end on the
circumference of the circle.
 2.6. KINEMATICS IN TWO DIMENSIONS 49

If you take the derivative of this equation with respect to time, you
obtain an expression for the velocity in polar coordinates.
dr d dr dr̂
v= = (rr̂) = r̂ + r .
dt dt dt dt
The last term comes from the product rule of differentiation and reflects
the fact that the unit vector r̂ may be changing with time. Although
the magnitude of r̂ is always unity, it may change in direction. Since r̂
is a function of θ, from the calculus the differential of r̂ is
dr̂
dr̂ = dθ.
dθ
So,
dr̂ dr̂ dθ dr̂
= = θ̇.
dt dθ dt dθ
But what is dr̂/dθ? Looking at Figure 2.7 you see that r̂ and θ̂ can be
resolved into Cartesian components as follows:
r̂ = 1 cos θı̂ + 1 sin θ̂,
where I explicitly included the number 1 to emphasize that the radius of
the unit circle is 1. You obtain a similar expression for θ̂ by inspection
of Figure 2.7. Dropping the “1” we write:
r̂ = cos θı̂ + sin θ̂, (2.10)
θ̂ = − sin θı̂ + cos θ̂. (2.11)
Therefore,
dr̂ d
= (cos θı̂ + sin θ̂) = − sin θı̂ + cos θ̂ = θ̂.
dθ dθ
Furthermore,
dθ̂
= − cos θı̂ − sin θ̂ = −r̂.
dθ
You may wish to memorize the following relations, as you will be using
them frequently in this and other courses.

dr̂ dθ̂
= θ̂ and = −r̂. (2.12)
dθ dθ
Having determined dr̂/dθ and dθ̂/dθ we can obtain dr̂/dt and dθ̂/dt
as follows:
dr̂ dr̂ dθ̂
= = θ̇θ̂,
dt dθ dt
and
dθ̂ dθ̂ dθ
= = −θ̇r̂.
dt dθ dt
50 2. KINEMATICS

As noted previously, in polar coordinates the velocity is given by

d d dr dr̂
v = r = (rr̂) = r̂ + r ,
dt dt dt dt
so
v =ṙr̂ + rθ̇θ̂. (2.13)
This equation for the velocity can be differentiated to obtain the ex-
pression for acceleration in polar coordinates:
a = (r̈ − rθ̇2 )r̂ + (rθ̈ + 2ṙθ̇)θ̂. (2.14)
We shall frequently refer to the expressions for velocity and acceleration
in terms of polar coordinates, so you should either memorize Equations
2.12, 2.13 and 2.14 or mark this page so you can find them easily.

Worked Example 2.5. A turntable rotates at a constant an-

gular speed ω. An ant crawls directly toward the rim along a radial
line at a constant speed b. You observe the ant from above. From
your point of view, the ant is moving in a spiral. Write an expres-
sion for the velocity and acceleration of the ant in polar coordinates.
Solution: Given ṙ = b and θ̇ = ω. Therefore, r = bt + r0 and
θ = ωt + θ0 . Note that θ̈ = 0 and r̈ = 0.
r = rr̂ = (bt + r0 )r̂
d dr̂
v = [(bt + r0 )r̂] = br̂ + (bt + r0 )
dt dt
v = br̂ + (bt + r0 )ω θ̂,
and
d
a= [br̂ + (bt + r0 )ω θ̂]
dt
dr̂ dθ̂
= b + bω θ̂ + (bt + r0 )ω
dt dt
= bω θ̂ − (bt + r0 )ω 2 r̂
= −(bt + r0 )ω 2 r̂ + 2bω θ̂.

Exercise 2.15. Derive Equation (2.14) from (2.13).

Exercise 2.16. Given r = a(ı̂ sin ωt+̂cos ωt), where ω is a con-
stant. (a) Determine the magnitude of the velocity. (b) Prove the
velocity is perpendicular to r. (c) Describe the motion. Answer: (a)
aω.
 2.7. KINEMATICS IN THREE DIMENSIONS 51

Exercise 2.17. The position of a certain particle is described by

r = 4t (meters), θ = 0.2t (radians). Determine its velocity as a func-
tion of time in (a) polar coordinates and (b) Cartesian coordinates. An-
swers: (a) v=4r̂+0.8tθ̂. (b) v = (4 cos 0.2t−0.8t sin 0.2t)ı̂+(4 sin 0.2t+
0.8t cos 0.2t)̂.
Exercise 2.18. You are analyzing the motion of a red dot painted
on the rim of a wheel of radius r = 5.0 cm. It returns to the same
point every 2 seconds. At t = 0 the particle is at (r, θ) = (5, 0).
Express the position and velocity vectors in Cartesian and polar co-
ordinates. Answers: r = 5r̂ = (5 cos πt)ı̂+(5 sin πt)̂, v = 5π θ̂ =
(−5π sin πt)ı̂+(5π cos πt)̂.

2.7. Kinematics in Three Dimensions

We now generalize to motion in three dimensions. We shall describe
three different coordinate systems: Cartesian coordinates, cylindrical
coordinates, and spherical coordinates. These are the most commonly
used three-dimensional coordinate systems in physics.
You will find that the cylindrical and spherical coordinates rely
heavily on the plane polar coordinates described in the previous section.

2.7.1. Cartesian Coordinates. The position of a particle in a

three-dimensional Cartesian coordinate system is given by
r =xı̂+y̂ + z k̂.
In an inertial (nonaccelerating) coordinate system the three unit
vectors (ı̂, ̂, k̂) are constant in magnitude and direction so their time
derivatives are zero. This makes it easy to obtain expressions for the ve-
locity and acceleration in terms of Cartesian coordinates. The velocity
is the time derivative of position.
dr
v= = ẋı̂ + ẏ̂ + ż k̂.
dt
Similarly, the acceleration is obtained by taking the time derivative of
the velocity.
dv
a= = ẍı̂+ÿ̂+z̈ k̂.
dt
52 2. KINEMATICS

2.7.2. Cylindrical Coordinates. To describe cylindrical coordi-

nates, visualize a particle located at some arbitrary point in a three
dimensional Cartesian coordinate system, as in Figure 2.8. Draw a
perpendicular from the x-y plane (that is, the plane z = 0) up to the
position of the particle. This perpendicular has magnitude z and di-
rection k̂ so we denote it by the vector z k̂. The point at which the
perpendicular intersects the z = 0 plane can be specified by its plane
polar coordinates. (But when using cylindrical coordinates we generally
use the symbols ρ and φ rather than r and θ.) The set of coordinates,
ρ, φ, z are called cylindrical coordinates. Figure 2.8 gives a graphical
representation of this coordinate system. Note that the vector from
the origin to the point is still denoted r and also note that φ is mea-
sured from the x-axis to the projection of r onto the x-y plane, which
is denoted ρ.

z
P
^
zk
r
y
x φ ρ

Figure 2.8. The position of point P can be specified

either in terms of the Cartesian coordinates x, y, z or the
cylindrical coordinates ρ, φ, z.

The next task is to determine unit vectors for the cylindrical coordi-
nate system. This is quite simple because the new coordinate system is
made up of old coordinates whose properties you have already learned.
You can appreciate from Figure 2.9 that ρ̂ and φ̂ are the same as the
plane polar unit vectors r̂ and θ̂, and the unit vector in the z direction
is just k̂ as in the Cartesian coordinate system.
Although k̂ is constant in magnitude and direction, ρ̂ and φ̂ will
change in direction as the angle φ varies. That is, ρ̂ and φ̂ are functions
of φ. We write:
ρ̂ = ρ̂(φ),
and
φ̂ = φ̂(φ).
The set ρ̂, φ̂, k̂ form an orthogonal set. The orthogonality of these
unit vectors leads the following set of easily proved properties:
 2.7. KINEMATICS IN THREE DIMENSIONS 53

ρ̂ · ρ̂ = 1, φ̂ · φ̂ = 1, k̂ · k̂ = 1
ρ̂ · φ̂ = 0, ρ̂ · k̂ = 0, φ̂ · k̂ =0
ρ̂ × φ̂ = k̂, φ̂ × k̂ = ρ̂, k̂ × ρ̂ = φ̂
The relationships between cylindrical coordinates and Cartesian co-
ordinates (the transformation equations) are:
x = ρ cos φ y = ρ sin φ z = z.
The inverse relationships are:
−1 y
p  
ρ= x +y 2 2 φ = tan z = z.
x
The unit vectors ρ̂ and φ̂ can be expressed in terms of ı̂ and ̂ as
follows:

ρ̂ = cos φı̂ + sin φ̂, (2.15)

φ̂ = − sin φı̂ + cos φ̂. (2.16)
The position of a particle in cylindrical coordinates is specified by

r = ρρ̂ + z k̂. (2.17)

The velocity is obtained by taking the time derivative of the position
vector:

dρ̂ dφ
v = ṙ = d
dt
(ρρ̂ + z k̂) = ρ̇ρ̂ + ρ dρ̂
dt
+ ż k̂ =ρ̇ρ̂ + ρ( dφ dt
) + ż k̂,
or
v =ρ̇ρ̂ + ρφ̇φ̂ + ż k̂. (2.18)
Similarly, the acceleration is given by
a = r̈ =(ρ̈ − ρφ̇2 )ρ̂ + (ρφ̈ + 2ρ̇φ̇)φ̂ + z̈ k̂. (2.19)

r
y
^ ^
ρ k
x φ
φ
^
ρ

Figure 2.9. Definition of the cylindrical unit vectors ρ̂,

φ̂, and k̂.
54 2. KINEMATICS

(You should be able to carry out the intermediate steps in obtaining

this expression for a. Don’t just nod your head and think, “Of course
I can do it.” Pick up your pencil and work it out! The exercise will
make the concepts stick in your mind.)

Worked Example 2.6. A bead slides on a wire bent into a

helix. The position of the bead as a function of time is given by
ρ = a, φ = ωt, z = bt2 , where a, ω, b are constants. Determine
an expression for the velocity and acceleration of the bead as a
function of time.
Solution: Recall that in three dimensions using cylindrical
coordinates, we have
r = ρ̂ + z k̂,
v = ρ̇ρ̂ + ρφ̇φ̂ + ż k̂,
a = (ρ̈ − ρφ̇2 )ρ̂ + (ρφ̈ + 2ρ̇φ̇)φ̂ + z̈ k̂.
For this particular problem
ρ = a ∴ ρ̇ = 0 ⇒ ρ̈ = 0
φ = ωt ∴ φ̇ = ω ⇒ φ̈ = 0
z = bt2 ∴ ż = 2bt ∴ z̈ = 2b
Consequently,
v = ρ̇ρ̂ + ρφ̇φ̂ + ż k̂ =0 + aω φ̂ + 2btk̂
= aω φ̂ + 2btk̂
and
a = (ρ̈ − ρφ̇2 )ρ̂ + (ρφ̈ + 2ρ̇φ̇)φ̂ + z̈ k̂
= 0 − aω 2 ρ̂ + (0 + 0)φ̂ + 2bk̂

= −aω 2 ρ̂ + 2bk̂

Exercise 2.19. Carry out the steps to obtain Equation (2.19).

Exercise 2.20. The position of a particle is given by ρ = 3t2 ,
φ = 2t and z = 12t, (distances in meters, angles in radians). What is
the acceleration at time t = 2 seconds? Answer: a = (−42ρ̂ + 48φ̂)
m/s2 .
 2.7. KINEMATICS IN THREE DIMENSIONS 55

2.7.3. Spherical Coordinates. The last set of coordinates we

will consider are called spherical coordinates. In these coordinates the
position of a particle is specified by the vector r which has length r and
direction (or orientation) specified by two angles, θ and φ, as shown
Figure 2.10. These angles are defined as follows: The “polar” angle θ
is measured from the z-axis to the vector r. The “azimuthal” angle φ
is measured from the x-axis to the projection of r onto the x-y plane.
The component of r along the z-axis is r cos θ. The component of r in
the x-y plane (denoted ρ) has a length of r sin θ. The x component of
r is, therefore, r sin θ cos φ and the y component is r sin θ sin φ.

z
^
r
z =r cos θ ^
φ
θr
^ ^
θ
k y =r sinθ sinφ
x = r sinθ cosϕ ^
ρ y
φ
ρ=rsinθ
x
Figure 2.10. Spherical coordinates and the unit vec-
tors r̂, θ̂, φ̂. Note that the Cartesian components of the
vector r are x = r sin θ cos φ, y = r sin θ sin φ and
z = r cos θ. The unit vectors r̂, θ̂, φ̂ point in the direc-
tion of increasing r,θ, φ, respectively. Also shown are the
cylindrical unit vectors ρ̂ and k̂.

What are the appropriate unit vectors? The unit vector r̂ is in the
direction of increasing r. The unit vector θ̂ is in the direction that the
tip of r will move if θ increases. In Figure 2.11 I sketched a quarter
circle to illustrate the motion of r as θ changes. The unit vector θ̂ is
tangent to this circle and has the direction of increasing θ. The unit
vector φ̂ lies in the x-y plane and gives the direction that the tip of
ρ̂ will move if φ increases. By sliding φ̂ down to the x-y plane, you
can see that φ̂ in spherical coordinates is the same as the φ̂ of the
cylindrical coordinates.
Although it may not be immediately obvious, the spherical unit
vectors r̂, θ̂, φ̂ form an orthogonal set. Figure 2.10 shows these three
mutually perpendicular unit vectors at the tip of the vector r. This is
56 2. KINEMATICS

φ
ρ

Figure 2.11. The unit vectors r̂, θ̂, and φ̂

probably the easiest way to remember them. In that figure, both r̂

and θ̂ lie in the shaded plane that contains r and the z-axis, and φ̂ is
perpendicular to that plane.
The relationships (transformation equations) between the spherical
coordinates and the Cartesian coordinates are:
x = r sin θ cos φ, (2.20)
y = r sin θ sin φ,
z = r cos θ.
The inverse transformation equations are:
1
r = (x2 + y 2 + z 2 ) 2 ,
p !
x 2 + y2
θ = tan−1 ,
z
y
φ = tan−1 .
x
Equations (2.20) are extremely useful and you might want to mem-
orize them. (However, they are easy to remember if you can visualize
Figure 2.10.) You will use these equations throughout this course as
well as in many upper division and graduate physics courses.
Figure 2.12 shows a unit circle that contains both r and the z-
axis. Its intersection with the x-y plane gives ρ̂, the unit vector for the
cylindrical coordinates. From the figure,
r̂ = cos θk̂ + sin θρ̂. (2.21)

Recalling that ρ and φ are the same as the plane polar coordinates
(formerly denoted r and θ), and using Equation (2.25),
ρ̂ = cos φı̂ + sin φ̂.
Consequently,
r̂ = sin θ cos φı̂ + sin θ sin φ̂ + cos θk̂. (2.22)
 2.7. KINEMATICS IN THREE DIMENSIONS 57

Figure 2.12. The unit vectors r̂, θ̂, φ̂ for spherical co-
ordinates and the unit circle. Also shown are k̂ and ρ̂.
In the sketch on the right, θ̂ has been translated from
the tip of r̂ to the origin.

We can also express θ̂ and φ̂ in terms of the Cartesian unit vectors

(ı̂,̂, k̂). From the unit circle on the right in Figure 2.12, it is clear that
θ̂ = cos θρ̂ − sin θk̂.
Substituting for ρ̂ we obtain
θ̂ = cos θ cos φı̂ + cos θ sin φ̂ − sin θk̂. (2.23)
Finally, φ̂ is the same as for the cylindrical coordinates, so
φ̂ = − sin φı̂ + cos φ̂. (2.24)
Now that we have determined r̂, θ̂, φ̂, we can evaluate their time
derivatives. Expressions for the velocity and acceleration are somewhat
more difficult to obtain in terms of spherical coordinates than in cylin-
drical coordinates because the unit vectors r̂ and θ̂ are functions of
both θ and φ. (It is clear from Figure 2.10 that a change in either θ or
φ will cause a change in the direction of r̂ and θ̂.) On the other hand,
φ̂ is a function only of φ. Thus we write, in the usual mathematical
notation,
r̂ = r̂(θ, φ),
θ̂ = θ̂(θ, φ),
φ̂ = φ̂(φ).
Recall that if a quantity, (say f ), is a function of several variables such
as x, y and z, then f = f (x, y, z), and by the rules of calculus, the
58 2. KINEMATICS

differential quantity df is given by:

∂f ∂f ∂f
df = dx + dy + dz.
∂x ∂y ∂z
(This simple relationship is used over and over again in physics. If you
have already taken a course in thermodynamics, you are aware how
important it is.)
Going back to the discussion of the unit vectors, you can appreciate
that if r̂ = r̂(θ, φ), then
∂r̂ ∂r̂
dr̂ = dθ + dφ,
∂θ ∂φ
and consequently
dr̂ ∂r̂ dθ ∂r̂ dφ
= + ,
dt ∂θ dt ∂φ dt
or
dr̂ ∂r̂ ∂r̂
= θ̇ + φ̇. (2.25)
dt ∂θ ∂φ
Therefore,
∂r̂
= cos θ cos φı̂ + cos θ sin φ̂ − sin θk̂. (2.26)
∂θ
But comparison with Equation (2.23) shows us that the right hand side
of this equation is simply θ̂.
Therefore,
∂r̂
= θ̂.
∂θ
We also need to determine ∂r̂/∂φ. Expressing r̂ in terms of the Carte-
sian unit vectors (Equation 2.22) and taking the derivative with respect
to φ we obtain
∂r̂
= − sin θ sin φı̂ + sin θ cos φ̂
∂φ
= sin θ(− sin φı̂ + cos φ̂),
or
∂r̂
= sin θφ̂.
∂φ
So finally, the expression for dr̂/dt (Equation 2.25) can be written

dr̂
= θ̇θ̂ + sin θφ̇φ̂.
dt
 2.7. KINEMATICS IN THREE DIMENSIONS 59

Similarly, we obtain the following two relations (see Problem 2.36):

dθ̂ ∂ θ̂ ∂ θ̂
= θ̇ + φ̇ = −θ̇r̂ + cos θφ̇φ̂, (2.27)
dt ∂θ dφ
dφ̂ ∂ φ̂ ∂ φ̂
= θ̇ + φ̇ = 0 − φ̇ρ̂ = −φ̇ρ̂. (2.28)
dt ∂θ ∂φ
Using Figure 2.12 you can show that,
ρ̂ = sin θr̂ + cos θθ̂,
so we can also write Equation (2.28) as
dφ̂
= −φ̇ sin θr̂ − φ̇ cos θθ̂.
dt
Having obtained the time derivatives of the three unit vectors r̂, θ̂, φ̂
it is straightforward to determine expressions for the velocity and ac-
celeration in spherical polar coordinates. Since
dr d dr dr̂
v= = (rr̂) = r̂ + r , (2.29)
dt dt dt dt
the velocity is
v = ṙr̂ + rθ̇θ̂ + r sin θφ̇φ̂. (2.30)
Taking the time derivative of both sides and using Equations (??) and
(??) leads to the following expression for the acceleration in spherical
polar coordinates.
a = (r̈ − rθ̇2 − rφ̇2 sin2 θ)r̂ + (rθ̈ + 2ṙθ̇ − rφ̇2 sin θ cos θ)θ̂
(2.31)
+(rφ̈ sin θ + 2ṙφ̇ sin θ + 2rθ̇φ̇ cos θ)φ̂.

Worked Example 2.7. Prove that the unit vectors in spheri-

cal coordinates are orthogonal. You may assume the orthogonality
of the Cartesian unit vectors.
Solution: The unit vectors for the spherical coordinates can
be expressed in terms of the Cartesian unit vectors by
r̂ = sin θ cos φı̂ + sin θ sin φ̂ + cos θk̂
θ̂ = cos θ cos φı̂ + cos θ sin φ̂ − sin θk̂
φ̂ = − sin φı̂ + cos φ̂
60 2. KINEMATICS

If two vectors are orthogonal (mutually perpendicular) their dot

product is zero. So
   
r̂ · θ̂= sin θ cos φı̂ + sin θ sin φ̂ + cos θk̂ · cos θ cos φı̂ + cos θ sin φ̂ − sin θk̂
= sin θ cos φ cos θ cos φ + sin θ sin φ cos θ sin φ − cos θ sin θ
= (sin θ cos θ)(cos2 φ + sin2 φ) − cos θ sin θ
= sin θ cos θ − cos θ sin θ = 0
∴ r̂ and θ̂ are perpendicular.
Similarly
 
r̂ · φ̂= sin θ cos φı̂ + sin θ sin φ̂ + cos θk̂ · (− sin φı̂ + cos φ̂)
= − sin θ cos φ sin φ + sin θ sin φ cos φ = 0
∴ r̂ and φ̂ are perpendicular.
Finally
 
θ̂ · φ̂= cos θ cos φı̂ + cos θ sin φ̂ − sin θk̂ · (− sin φı̂ + cos φ̂)
= − cos θ cos φ sin φ + cos θ sin φ cos φ = 0
∴ θ̂ and φ̂ are perpendicular.

Exercise 2.21. A particle is at (3,4,5) meters in Cartesian coordi-

nates. Determine r, θ, φ. Answer: (7.07, π/4, 0.927)

2.8. Summary
In this chapter you learned how to express the position, velocity,
and acceleration of a particle in two dimensions in Cartesian coordi-
nates and plane polar coordinates, and in three dimensions in Cartesian
coordinates, cylindrical coordinates and spherical coordinates.
In three dimensions the position, velocity, and acceleration in Carte-
sian coordinates are given by the simple relations:
r = xı̂ + y̂ + z k̂,
v = ẋı̂ + ẏ̂ + ż k̂,
a = ẍı̂ + ÿ̂ + z̈ k̂.
 2.9. PROBLEMS 61

These relationships for two dimensions in plane polar coordinates are:

r = rr̂,
v = ṙr̂ + rθ̇θ̂,
a = (r̈ − rθ̇2 )r̂ + (rθ̈ + 2ṙθ̇)θ̂.
In three dimensions using cylindrical coordinates, the relationships are:
r = ρρ̂ + z k̂,
v = ρ̇ρ̂ + ρφ̇φ̂ + ż k̂,
a = (ρ̈ − ρφ̇2 )ρ̂ + (ρφ̈ + 2ρ̇φ̇)φ̂ + z̈ k̂.
Finally, in spherical coordinates the position, velocity and acceleration
are given by:
r = rr̂,
v = ṙr̂ + rθ̇θ̂ + r sin θφ̇φ̂.
a = (r̈ − rθ̇2 − rφ̇2 sin2 θ)r̂ + (rθ̈ + 2ṙθ̇ − rφ̇2 sin θ cos θ)θ̂
+ (rφ̈ sin θ + 2ṙφ̇ sin θ + 2rθ̇φ̇ cos θ)φ̂.

2.9. Problems
Problem 2.1. A particle initially at rest undergoes an acceleration
given by a = 3e−0.5t m/s2 . Determine the terminal velocity.
Problem 2.2. The nitrogen atom in an ammonia molecule can be
assumed to oscillate in simple harmonic motion, so that its position
at any time t is given by z = A cos ωt where A and ω are constants.
Obtain expressions for its velocity and acceleration as functions of time.
Problem 2.3. This is the story of Freddy Physics who got a ticket
for going through a red light. He explained to the judge that if he
slammed on the brakes when he saw the signal light turn yellow, he
would not be able to stop before the intersection, and if he continued
at a constant speed the light would turn red before he reached the
intersection. In this problem you will determine how far from the light
Freddy was when the light turned yellow. Suppose he is traveling at
the speed limit, v0 , that the light remains yellow for a time t0 and
that Freddy’s reaction time is τ . His car decelerates at a rate a. (a)
Determine the distance d1 he will travel in time t0 if he continues at the
speed limit (b) Determine the distance d2 required to stop. (c) Show
that he is bound to go through the red light if t0 < τ + v0 /2|a|
62 2. KINEMATICS

Problem 2.4. Assume a particle is given an acceleration a = 10e−2t

m/s2 . The velocity at time t = 0 is zero. (a) Obtain expressions for the
velocity and position of the particle as functions of time. (b) Determine
the velocity as t → ∞.
Problem 2.5. An object has an acceleration that is inversely pro-
portional to the velocity squared: a = 9/v 2 m/s2 . Assume the object
is initially at rest at the origin. Determine its position at time t = 10
sec.
Problem 2.6. In some circumstances the acceleration is a function
of the velocity. (An example is the motion of an object in a resistive
medium such as air.) Assume a = −kv 2 where k is a constant. Deter-
mine the velocity and position as a function of time. What is the value
of the velocity if the time is very large?
Problem 2.7. Mary is playing outfield and John is playing third
base, 30.0 meters away. Mary can throw a baseball at 20.0 m/s. Ignor-
ing air resistance, determine the two launch angles she can use to get
the ball to John.
Problem 2.8. A cannon launches a shell at 250 m/s at an elevation
of 40◦ . Assuming level ground, determine the vertical and horizontal
components of the velocity one second before the shell hits the ground.
Problem 2.9. When great athletes or ballet dancers leap into the
air, they appear to “hang” at the top of the trajectory. (The reason
for this is that the vertical speed of a projectile is smallest near the top
of its path and greatest near the ground.) Show that half of the total
time for the jump is spent in the top one quarter of the trajectory.
Problem 2.10. The British Navy sends a gunboat to quell a revolt
on a British colony. This colony happens to be on an island exactly
3 km wide with a 1.5 km high mountain running down its center, as
shown in Figure 1.10. The gunboat is anchored offshore of the western
side of the island, 2.5 km from the center. On the eastern shore a
distance of 1.5 km from the center, the rebels have a cannon that fires
shells at 200 m/s. The gunboat’s cannon fires shells at 300 m/s. Find
the angle at which the British should aim their cannon. Show that the
rebels cannot sink the British ship. See Figure (2.13). Answer: 77.08o .

Problem 2.11. A man is at the base of a hill that slopes upward

at a constant angle of 10 ◦ . He aims a rifle up the hill at an angle of
30 ◦ above the horizontal. If the initial speed of the bullet is 300 m/s,
determine how far up the hill the bullet will hit. You may neglect the
fact that the initial vertical position of the bullet is not zero.
 2.9. PROBLEMS 63

300 m/s 1.5 km

200 m/s

2.5 km 1.5 km

Figure 2.13. The British gunboat and the rebel cannon

of Problem 2.10. Do the shells of each one reach the
other?

Problem 2.12. A projectile is launched at an angle θ0 with initial

speed v0 . (a) Obtain an expression for y as a function of x, y = y(x).
(b) Obtain an expression for the speed and direction of the projectile
as a function of its position.
Problem 2.13. David is practicing with his sling before going out
to meet Goliath. The sling is 1.3 meters long and David whirls it in
a horizontal circle over his head at 3 rev/sec. The circular path of
the sling is 2.0 meters above the ground. How far does the stone fly
horizontally before hitting the ground?
Problem 2.14. In projectile problems one usually assumes the
gravitational acceleration is directed straight down. Now assume the
gravitational acceleration is a constant equal to 9.8 m/s2 but pointing
at an angle 30◦ below the horizontal. Obtain an expression for the
range of a projectile launched at 20◦ to the horizontal with an initial
velocity of 300 m/s.
Problem 2.15. Determine the distance traveled along its parabolic
path when a projectile moves from the origin to the top of its trajectory.
It is launched with initial velocity v0 at an angle θ.
Problem 2.16. James Bond want to impress Miss Moneypenny.
They go out to an empty field and Bond fires his pistol at an elevation
angle θ1 , then an instant later, he fires it again at a lower angle θ2 .
The bullets collide in midair. Show that the time interval between the
shots is
2v sin(θ1 − θ2 )
∆t = .
g cos θ1 + cos θ2
64 2. KINEMATICS

Problem 2.17. A projectile fired from y = 0 has reached the top

of its trajectory. Its coordinates at this point are (xt , yt ). Obtain a
formula for the magnitude and direction of the initial velocity in terms
of xt , yt .
Problem 2.18. A particle is moving in a circular path of radius
3 meters with angular speed ω = 10 rad/sec. Place the origin of the
Cartesian coordinate system at the center of the circle.
(a) Obtain expressions for its Cartesian coordinates (x, y) as func-
tions of time.
(b) Express r and v in Cartesian coordinates.
(c) Determine the speed from your expression for v.
Problem 2.19. Assume the orbit of Earth is perfectly circular.
Using tabulated values for the radius of the Earth’s orbit and the ap-
propriate masses, determine the ratios of the centripetal accelerations
of the Earth and the Sun. From this, determine the radius of the Sun’s
orbit around the center of mass of the system.
Problem 2.20. The position (x, y) of a particle (in meters) is given
by
x = 3 sin 5t
y = 3 cos 5t.
(a) Plot the path of the particle. (b) Determine the tangential accel-
eration and (c) the centripetal acceleration of the particle.
Problem 2.21. Consider a “surface skimming” satellite which is in
a circular orbit whose radius is equal to the Earth’s radius. (a) Obtain
an expression for the period of such a satellite in terms of the accelera-
tion and radius. (b) Look up the radius of the Earth and evaluate the
acceleration, speed and period of a surface skimming satellite.
Problem 2.22. A spinning top is rotating at 20 rev/sec about
its axis. The axis itself is oriented at 15◦ from the vertical and is
precessing at 0.5 rev/sec about the vertical. What is the total vector
angular velocity ω? (Express in Cartesian coordinates.)
Problem 2.23. A satellite is in a circular orbit around the Earth.
As shown in Problem
p 2.21 the period, radius, and acceleration are
related by τ = 2π r/a. Furthermore, it is known from Newton’s law
of universal gravitation that the acceleration decreases with distance
from the center of Earth according to
1
a ∝ 2.
r
 2.9. PROBLEMS 65

If the orbital radius increases by a small amount ∆r, the period will
increase by ∆τ. Show that
∆τ 3 ∆r
= .
τ 2 r
Problem 2.24. Prove the triangle inequality which states that for
two vectors A and B, |A + B| ≤ |A| + |B| .
Problem 2.25. Show that A · (B × C) is the volume of a paral-
lelepiped whose edges can be represented by the vectors A, B, C.
Problem 2.26. Evaluate (A × B) · C if A = 3ı̂+2̂+4k̂, B = 3ı̂+
2k̂ and C = 4k̂. Also show that you get the same result by evaluating
A · (B × C).
Problem 2.27. Show that
 
 Ax Ay Az 
 
(A × B) · C =  Bx By Bz  .
 Cx Cy Cz 
Problem 2.28. Prove the “BAC minus CAB” rule:
A × (B × C) = B(A · C) − C(A · B).
Problem 2.29. Derive the law of cosines for triangles, i.e., show
that if the sides of a triangle have lengths a, b, c, and the angle between
a and b is θ, then
c2 = a2 + b2 − 2ab cos θ.
Problem 2.30. The position of a particle is given by r = 2t (me-
ters), θ = 5t (radians). (a) Plot the path of the particle. (b) Obtain
an expression for the speed as a function of time.
Problem 2.31. Suppose the position of a particle as a function of
time is given by r = 1 + sin t and θ = 1 − e−t . Obtain v and a in terms
of r̂ and θ̂.
Problem 2.32. A particle moves in a plane such that r = 1 − cos θ
m/s and θ̇ = 4 rad/s. Determine v and a.
Problem 2.33. A particle moves in a plane in such a way√ that
r = 2 + sin t (m) and the magnitude of the velocity is v = 2 cos t
(m/s). Find a formula for θ = θ(t).
Problem 2.34. For plane polar coordinates, determine
d3 r
.
dt3
66 2. KINEMATICS

Problem 2.35. Let êν be a unit vector parallel to the velocity

vector v. Therefore, the acceleration is given by
   
dv dêν
a= êν + v .
dt dt
(a) Show that (dêν /dt) is a vector perpendicular to êν . (b) Show that
if r is the radius of curvature of the path of the motion, then (dêν /dt)
= v/r. (c) Finally, show that if ên is a unit vector directed towards the
center of curvature on the concave side of the path, that we can write
·v2
ên .
a =at êν + an ên = vêν +
r
Problem 2.36. Derive Equations (2.27) and (2.28).
Problem 2.37. A bug is crawling on the curved surface of a cylin-
der of radius 25 cm. A scientist determines that the bug’s coordinates
vary according to φ = 4.0t and z = 0.3t2 . Write an expression for the
bug’s velocity in cylindrical coordinates and evaluate its speed at time
t = 2 seconds.
Problem 2.38. A particle of mass m moves on the curved sur-
face of a cylinder of radius R. Express the acceleration in cylindrical
coordinates.
Problem 2.39. In cylindrical coordinates, the divergence of a vec-
tor F is given by
1 ∂(ρFρ ) 1 ∂Fφ ∂Fz
∇·F= + + .
ρ ∂ρ ρ ∂φ ∂z
Evaluate ∇ · F for
√
F(ρ, φ, z) = ρρ̂ + z sin φφ̂ + ρz k̂.
Problem 2.40. An object initially at (0,0,1) meters in Cartesian
coordinates, moves in the y direction at a constant speed v. (a) Obtain
an expression for its position as a function of time in spherical coordi-
nates. (b) Determine ṙ and θ̇ as functions of time. (c) Show that the
speed, expressed in spherical coordinates, is constant.
Problem 2.41. A satellite moves with constant speed about the
Earth in a circular orbit of radius a inclined at 90◦ to the equator, so
that the satellite goes over the poles. A system of Cartesian coordinates
fixed in the Earth has its origin at the center of the Earth, the z-axis
passing through the North Pole, and the x-axis in the equatorial plane
and passing through the Greenwich meridian. Obtain expressions for
 2.9. PROBLEMS 67

the position of the satellite (as functions of time) in Cartesian coordi-

nates. The angular speed of the satellite is ω(rad/s) and the rotation
rate of the Earth is Ω (rad/s).
Problem 2.42. Show that the element of arc length in spherical
coordinates is

1/2
ds = dr2 + r2 dθ2 + r2 sin2 θdφ2 .
Problem 2.43. Derive the equations of transformation between
cylindrical and spherical coordinates.
Problem 2.44. Show that the unit vectors of the spherical coor-
dinate system can be expressed in terms of Cartesian coordinates as
follows:
xı̂+y̂ + z k̂
r̂=
(x2 + y 2 + z 2 )1/2
z(xı̂+y̂)− (x2 + y 2 ) k̂
θ̂=
(x2 + y 2 )1/2 (x2 + y 2 + z 2 )1/2
−yı̂+x̂
φ̂=
(x2 + y 2 )1/2
Problem 2.45. An object is moving in such a way that its position
in cylindrical coordinates is given by ρ = a − bt, φ = 21 kt2 , z = 0. (a)
Describe the path. (b) Find the speed as a function of time.

Problem 2.46. A proton with velocity v = aρ̂ + bk̂ enters a uni-

form magnetic field oriented in the +z direction, so that B =B0 k̂. (a)
Describe the subsequent motion of the proton. (b) Otain an expression
for its position as a function of time in cylindrical coordinates.
Problem 2.47. An object moves at constant speed. Prove that
any non-zero acceleration must be perpendicular to the velocity.
Problem 2.48. (a) Show that the component of the acceleration
in the direction of the velocity can always be expressed as
v·a
aq = .
v
(b) Find the component of the acceleration perpendicular to the veloc-
ity.
Computational Projects
68 2. KINEMATICS

Computational Project 2.1. A certain (imaginary) planet is

in orbit about the Sun. It follows an elliptical path that is given by
4
r= ,
1 + 0.2 cos θ
where r is measured in Astronomical Units (AU). Plot the orbit of the
planet in polar coordinates.
Computational Project 2.2. Two cannons are located at
(essentially) the same point. The first cannon fires a projectile at an
initial speed of v1 = 250 m/s at an elevation angle of θ1 = 64.1◦ . After
a time interval ∆t, the second cannon fires its projectile at v2 = 300
m/s and θ2 = 47.3◦ . Determine the time interval ∆t such that the two
projectiles collide in midair. Check your answer against the analytical
result
2v1 v2 sin(θ1 − θ2 )
∆t = .
g v1 cos θ1 + v2 cos θ2
Computational Project 2.3. A volcanic eruption throws a
boulder vertically into the air with an initial speed of 50 m/s. (a)
Write a computer program to plot the position and velocity of the
boulder as a function of time. (b) Now assume the boulder is ejected
at an angle of ten degrees to the vertical. Plot the two components of
position and velocity as functions of time and also plot the trajectory
of the boulder (on an x, y plot). Ignore the effect of air resistance.
Computational Project 2.4. In Worked Example 2.3 we ob-
tained an expression for the launch angle that would give the greatest
range for a cannon mounted on the top of a hill. (a) Validate that
expression by varying the launch angle θ. (b) Determine how the opti-
mum launch varies as the height of the hill is increased from zero.
Computational Project 2.5. Worked Example 2.6 describes a
bead sliding on a helical wire. Plot the trajectory of the bead.
Computational Project 2.6. In a shotput competition an ath-
lete throws a heavy metal ball as far as possible. A champion athlete
can throw the “put” a distance of 20 meters. Assume the ball is thrown
with an initial speed of 13.66 m/s at an angle of 45◦ . If it is thrown
from a height of one meter above the ground, the ball’s initial position
is (0,1). Using the relations
x = x0 + v0x t,
y = y0 + v0y t − (1/2)gt2 ,
obtain x(t) and y(t). Plot the trajectory for y > 0. Also plot trajectories
for angles ranging from 20◦ to 70◦ .
Chapter 3
Newton’s Laws: Determining
the Motion

This chapter (and much of the rest of this book) deals with dy-
namics, that is, the relation between the forces acting on a body and
its motion. A force is an interaction between a body (or particle) and
its environment, usually described as a push or a pull in a specified
direction. In this book you will encounter a number of familiar forces,
such as the gravitational force and forces exerted by springs as well as
a few less familiar ones.
On a very basic level, all forces are manifestations of the “funda-
mental forces,” namely, the gravitational force, the nuclear (or strong)
force, and the electroweak force. The electroweak force is often thought
of as two different forces, the electromagnetic force and the weak nu-
clear force. The electromagnetic force, in turn, is often thought of as
the electric force and the magnetic force.1 All the known forces in
nature are ultimately related to these fundamental forces. For exam-
ple, the force exerted by your muscles can be traced back to electrical
forces.
Dynamics is neatly summarized by Newton’s three laws. You stud-
ied Newton’s laws in your introductory mechanics course. By this time,
you surely know the laws by heart and you know how to use them to
solve reasonably complicated physics problems. However, it is impor-
tant for you as a physicist to have a thorough understanding of these
fundamental statements about the nature of the physical universe. In

1To learn more about the fundamental forces you might read the interesting
article by Charles Seife, “Can the Laws of Physics be Unified?” Science, 309, 82
(2005).
69
70 3. NEWTON’S LAWS: DETERMINING THE MOTION

this chapter I will discuss some aspects of Newton’s Laws that you
may not have considered before. I hope this will give you a greater
appreciation for the scope and significance of the laws.

3.1. Isaac Newton(Optional Historical Note)

In some people’s minds, Sir Isaac Newton was the greatest physicist
who ever lived. Others might give that honor to Albert Einstein, but
no one would deny the tremendous insight and genius of Isaac Newton.
Newton was born in 1642, the year Galileo died.2 Newton’s father
died before Isaac was born, and his mother moved to her brother’s
farm, where the boy grew up. When the time came, his uncle sent
him to Cambridge. While he was studying at Cambridge, a plague
swept through England, and the university authorities sent the students
home for a year. Isaac went back to the farm where he spent his time
thinking about the properties of the physical universe and doing simple
but very clever experiments. After the year was up, the plague had run
its course, and Newton returned to Cambridge. We can only imagine
the scene when he and his major professor met:

Professor: Hello Isaac. Welcome back. Did you spend

your time fruitfully while the University was closed?
Newton: Yes sir, I believe I did.
Professor: Very good, Isaac. Precisely what did you
do?
Newton: Well, sir, I proved the binomial theorem,
invented calculus, designed and built a reflecting tele-
scope, derived the law of universal gravitation, devel-
oped a theory of optics, and determined three fun-
damental laws of nature governing the motion of any
physical object.

According to legend, the professor quit his job and left the position
to Newton.
Newton did not formulate his laws out of the blue. He was very fa-
miliar with the work of his predecessors, especially Galileo and Kepler.
(As he stated, “If I have seen further than others, it is by standing on

2Newton was born on 25 December 1642, according to the Julian calendar

then in use in England. In 1753, England adopted the Gregorian calendar making
Newton’s birth date 4 January 1643.
 3.2. THE LAW OF INERTIA 71

the shoulders of giants.”) In fact, Newton’s first law, the law of iner-
tia, was formulated and demonstrated by Galileo. It is possible that
Galileo got the idea from Rene Descartes. However, the second and
third laws were first formulated by Isaac Newton.3
Although Newton discovered the law of universal gravitation when
he was very young, he did not publish it until many years later because
his early calculations did not give the correct value for the period of
the Moon. Urged by his friend, the architect Sir Christopher Wren,
Newton carried out the calculations more carefully, obtained the right
answer, and convinced himself that his law was correct.
Newton invented the reflecting telescope and carried out a large
number of experimental investigations, particularly in optics. He was
also interested in alchemy and theology. But his greatest contribution
was in formulating physics as an exact mathematical science. His ge-
nius was immediately recognized and he soon became one of the most
famous men in Europe. After Newton, people realized that natural
processes take place in a manner that can be analyzed and predicted.
People began to think of the universe as a “clockwork” mechanism in
which the future development of any system could (in principle) be
determined from a knowledge of its present state. The poet Alexander
Pope, eulogized Newton with the couplet,
Nature and Nature’s Laws lay hid in Night,
God said, Let Newton be! and All was Light.
Personally, Newton was not easy to deal with. It was said of him
that, “He suffered people poorly, and fools not at all.” Since his position
at Cambridge was a semi-clerical post, he was not allowed to marry. It
appears that the life of a bachelor suited him well. He died in 1727 at
the ripe old age of eighty four.4

3.2. The Law of Inertia

The law of inertia, or Newton’s first law,5 can be stated as follows:
3Newton expressed the laws of motion in his famous book, Philosophia Naturalis
Principia Mathematica published in 1687. A fairly recent English version is, The
Principia: Mathematical Principles of Natural Philosophy translated by I. Bernard
Cohen and Anne Whitman, University of California Press, Berkeley, 1999.
4A recent very readable biography is Isaac Newton by James Gleick, Pantheon
Books, New York, 2003.
5This law was first expounded by Galileo. Newton gave Galileo full credit for
it, but over the years we have grown accustomed to calling it “Newton’s First Law.”
If you happen to be a purist, you might want to know Newton’s precise words.
He said, “Every body perseveres in its state of rest, or of uniform motion in a right
72 3. NEWTON’S LAWS: DETERMINING THE MOTION

A body in motion will remain in uniform mo-

tion and a body at rest will remain at rest
unless acted upon by a net external force.
You have certainly thought about this law and you realize that it
says that in the absence of external forces a body at rest will remain
at rest. This is a very profound concept. Like many profound con-
cepts, after you have accepted it, it appears completely obvious. It is
important to note that internal forces cannot change the motion of a
system. If you sit on a chair and pull up on the seat, you remain at
rest no matter how hard you pull. You cannot lift yourself into the air.
The reason is, of course, that your pull is an internal force.
You might object that an exploding bomb is an example of internal
forces causing a system to accelerate. After the explosion, pieces of
the bomb are flying in all directions. This is true. But the center of
mass of the bomb remains at rest. The position of an extended body
is described by the position of its center of mass and if you trace the
trajectories of all the pieces, you will find that the center of mass did
not accelerate.
Perhaps the most amazing thing about the law of inertia is the idea
that a moving body will continue to move even though no external
force is applied to it. This was hard for people in Newton’s time to
understand. It is still hard for some people to understand. They say,
“When I push a trunk across the floor, as soon as I quit pushing, the
trunk quits moving.” Of course, you and I know the trunk quits moving
because there is a force acting on it - the force of friction. Furthermore,
you are familiar with games such as air hockey in which the puck slides
across the table at a (nearly) constant velocity because the frictional
force has (very nearly) been eliminated. You have also been exposed to
countless TV scenes of astronauts floating in space, so you do not have
the conceptual difficulties your great-grandfather had in accepting that
an object, not subjected to any force, will move forever in a straight
line at a constant velocity.
Consider a moving body with no external forces acting on it. Ac-
cording to the first law, this body will move at a constant speed in a
straight line. If you think about it for a moment, you will realize that
this will only be true in certain specific reference frames. For example,
line, unless it is compelled to change that state by forces impressed thereon.” (The
Principia by Isaac Newton, translated by Andrew Motte from the Latin. Pub. D.
Adee, 1848.)
I urge you to express Newton’s laws in your own words. In fact, sometimes I
state them one way and sometimes another. The crucial thing is not the words we
use but the concept expressed by those words.
 3.2. THE LAW OF INERTIA 73

if you are riding in a bus and someone places a tennis ball in the aisle,
it will remain at rest in the aisle as long as the bus is moving at a con-
stant velocity. But if the bus accelerates the ball will roll. If the bus
slows down, the ball rolls forward. If the bus turns left, the ball rolls
to the right. If I were somebody who did not know very much physics,
I might conclude that the ball does not obey Newton’s first law. You
realize, of course, that I am describing the motion of the ball relative
to the bus and the bus is, as I said, accelerating. So you would refute
my statement by pointing out that relative to the ground the ball did
obey the first law. It did maintain a state of uniform motion in the
reference frame of the Earth. I might then point out that the Earth is
also an accelerating reference frame because it is rotating. How would
you respond?
If we follow such arguments to their logical conclusion we will decide
that the first law is only valid in a non-accelerating reference frame.
(This is also true of Newton’s other two laws.) A non-accelerating
frame is called an inertial reference frame. Does such a reference frame
actually exist? Newton stated that his laws were valid in a reference
frame that was “at rest with respect to the fixed stars.” But we know
the stars are all in motion! It is probably best to treat the concept
of an inertial reference frame as a useful idealization. For many prob-
lems, the Earth can be treated as if it were at rest. This approximation
breaks down when the rotation of the Earth must be considered. Then
we usually assume that the inertial reference frame is a nonrotating
reference frame with origin at the center of the Earth. That approx-
imation breaks down if we need to include the orbital motion of the
Earth around the Sun. If necessary, we can take the origin of coordi-
nates to be at the center of the Sun, or even the center of the galaxy.

The physical property called “inertia” is associated with the fact

that a moving body tends to preserve its state of motion. The ex-
pression, “preserving its state of motion” means the body has constant
velocity. A locomotive has more inertia than a ping-pong ball. The
word inertia thus appears to be a synonym for mass. Since it requires a
great force to change the motion of an object with great mass, one often
hears the expression, “Mass is a measure of the inertia of a body.” But
the tendency of a body to maintain its state of motion also depends on
its velocity. It is easier to deflect a slowly moving five gram ping-pong
ball than a speeding five gram bullet. In this case, inertia appears to
be a synonym for momentum.
It is very difficult to define fundamental quantities such as mass,
distance, time, charge, and so on. Similarly, inertia is usually described
74 3. NEWTON’S LAWS: DETERMINING THE MOTION

by the somewhat vague expression that it is the tendency of a body to

maintain its state of motion. (We do not have a formula for inertia!)
On the other hand, the law of inertia (Newton’s First Law) is perfectly
well defined.

3.3. Newton’s Second Law and the Equation of Motion

Newton’s second law can be stated in the form
The rate of change of the momentum of an
isolated body is equal to the net external
force applied to it.
In equation form this is written:

dp
= F, (3.1)
dt
where F is the net or total force acting on the body. (You may prefer
to write it as Fnet or as ΣF.) Now momentum is defined as p =mv, so
if the mass is constant, Newton’s second law takes on the familiar form

F = ma.
Keep in mind that Newton’s second law is applicable only in an inertial
reference frame.
Newton’s second law expressed in the form a = F/m is sometimes
referred to as the “equation of motion.” The reason is that if the forces
are known, we can determine the acceleration of the body, and once
the acceleration has been determined we can integrate to obtain the
“motion,” that is, the velocity and position as functions of time. Ac-
cording to classical mechanics, a knowledge of the position and velocity
of all the particles in a system allow one to determine the forces and
hence to predict the future development of the system.6 Therefore, an
6This is a wonderful fact of nature. As stated by Landau and Lifshitz on the
very first page of their excellent advanced mechanics book, “If all the coordinates
and velocities are simultaneously specified, it is known from experience that the
state of the system is completely determined and that its subsequent motion can,
in principle, be calculated.” (Emphasis added.) They are suggesting that there is
no fundamental reason why the motion of a system depends only on position, ve-
locity and time; we must accept it as an experimental fact. (Reference: Mechanics,
Course in Theoretical Physics, Volume 1, L. D. Landau and E. M. Lifshitz, Perg-
amon Press, New York, 1976. Page 1.) Interestingly, in electrodynamics there is a
quantity called the “Abraham-Lorentz force” that is usually expressed as a function
of acceleration. This force leads to some strange behavior that has been referred to
as “philosophically repugnant” by David Griffiths (Introduction to Electrodynamics,
3rd Ed, Prentice Hall, 1999, page 467).
 3.3. NEWTON’S SECOND LAW AND THE EQUATION OF MOTION 75

expression for the acceleration in terms of the velocity, position and

time is called an equation of motion. Newton’s second law is one way
to obtain the equation of motion. In this course you will also learn
other ways to obtain the equation of motion.
Much of our work in this course will involve obtaining and solving
the equation of motion for a variety of different physical systems. The
equation of motion a = a(x, v, t) = F/m can be obtained directly from
an analytical expression for F. As you will learn in Chapter 4, the
equation of motion can be obtained without explicitly using Newton’s
second law. However, for now, we will stick to the second law.
Although we shall not be particularly concerned with philosoph-
ical questions, it should be mentioned that over the years since the
publication of Newton’s great work, “Principia,” there has been much
debate and speculation on the meaning of the second law. Clearly it is
a statement of how a body reacts when a force is applied to it. If the
mass is constant, the second law tells us that the acceleration of the
body is proportional to the force and inversely proportional to the mass
(a = F/m). From this point of view, mass becomes the proportionality
constant in the relationship a ∝ F. Some thinkers have gone so far as
to state that the second law is just a definition of mass. But if mass
is nothing more than a measure of inertia, one cannot explain the Law
of Universal Gravitation (F = −G (mM/r2 ) r̂) which states that the
gravitational force between two bodies of masses m and M is propor-
tional to the product of the masses of the bodies. The gravitational
law implies that mass is the source of the gravitational force. Perhaps
the “mass” in the gravitational force law is not the same as the “mass”
in Newton’s second law. Perhaps we are using the same word for two
different things. In that case, the question could be resolved by using
different symbols and different names for the two things. Thus, for
example, we could define the quantities mG and mI as:
mG = gravitational mass = the property by which
a body exerts a gravitational force on other bod-
ies, and Newton’s gravitational law would be ex-
pressed as F = −G (mG MG /r2 ) r̂
and
mI = inertial mass = the property by which a body
resists a change in its state of motion, so that New-
ton’s second law would be expressed as F = mI a.
However, many very careful experiments (some of the most famous
were by Baron Eötvos in Hungary over a hundred years ago), showed
76 3. NEWTON’S LAWS: DETERMINING THE MOTION

that mG and mI are equal to within one part in 20 million. This leads
one to conclude that gravitational mass and inertial mass are indeed
the same thing. The fact that inertial mass and gravitational mass are
equal is called “The Principle of Equivalence.” Albert Einstein used
this equivalence as a basic postulate of his theory of General Relativity
(1915).7
Newton’s second law is a cornerstone of classical physics because
it can be used to calculate the acceleration of a body, given the force
acting on it. Once the acceleration of a body is known, the laws of
kinematics determine its velocity and position at any later time. This
means that if you know the net force acting on a body, you can calculate
its position at any future time. The ability to predict the motion is the
power of the second law.
Recall again that dynamics is the study of how a force affects the
motion of a body. As you might suspect, dynamics usually involves
accelerating bodies. However, in some situations there may be forces
acting on a body but nevertheless the acceleration is zero. Consider,
for example, a body acted upon by two equal and opposing forces. The
effects of these forces cancel out and the body does not accelerate. Zero
acceleration and zero velocity is an important special case in dynamics
and is called statics. Statics is of particular interest to civil engineers
who want to make sure the structures they design, such as bridges and
skyscrapers, will have zero acceleration. Statics was treated briefly in
Section 1.5.2 and will be dealt with in greater detail in Chapter 14.
The principle of superposition states that if two or more forces act
on a particle, the net effect is that due to a single force equal to the
vector sum of all the forces. You will be exposed to the principle of
superposition in other areas of physics. For example, the net electric
field at a point is the vector sum of all the electric fields acting at that
point.
In our discussion of Newton’s laws we considered interactions be-
tween particles. When we apply Newton’s second law to an extended
body the acceleration a refers to the acceleration of the center of mass
of the body. A net external force F applied at any point of an extended

7Ifyou are interested in some of the philosophical implications of mass, force,

inertia and Newton’s laws, you might enjoy the three articles by Franck Wilczek
entitled, “Whence the Force of F = ma?” These articles were published in Physics
Today in the issues of December 2004, July 2005, and October 2005.
The article “Drop Test”by Adrian Cho (Science, 6 March, 2015, page 1096)
describes three experiments being performed to measure the equivalence of inertial
mass and gravitational mass to better than one part in ten trillion.
 3.3. NEWTON’S SECOND LAW AND THE EQUATION OF MOTION 77

rigid body will cause the center of mass of the body to accelerate ac-
cording to a = F/m. This is often not at all intuitive. For example,
consider a tricycle with the pedals in a vertical position as shown in
Figure 3.1. If you pull forward on the top pedal, the bicycle will ob-
viously move forward. But what happens if you pull forward on the
bottom pedal? It might seem that this would propel the tricycle back-
ward, but Newton’s second law tells us that if we pull forward, the
tricycle moves forward. (If you do not believe me, try it! You will be
surprised by the motion of the pedal itself. This is not at all the same
as what would happen if you were riding the tricycle. In that case, the
force your foot exerts on the pedal is an internal force.)

Figure 3.1. What direction does the tricycle move

when you pull on the bottom pedal as shown?

Exercise 3.1. A box of mass 5 kg is placed on an inclined plane

of angle 35◦ . A force of 10 N parallel to the plane in a direction up the
plane is applied to the box. The coefficient of sliding friction is 0.07.
Determine the acceleration of the box. Answer: -3.05 m/s2 (down the
plane).

Exercise 3.2. A crate of mass 50 kg is sitting on the flat bed of

a truck. The truck accelerates at 0.2 m/s2 . The coefficient of static
friction is 0.15. Does the crate slide? Answer: No.

Exercise 3.3. A spool of thread is lying on its side on a table and

is free to roll. You pull on the free end of the string. Does the spool
move towards you? Does it matter whether the string comes from the
bottom side of the spool or the top side? (Try it!)
78 3. NEWTON’S LAWS: DETERMINING THE MOTION

3.4. Newton’s Third Law: Action Equals Reaction

Using Isaac Newton’s terminology of “action” and “reaction”, we
can formulate his third law as follows:
To every action there is always opposed an equal reaction.
In different words, if one body exerts a force on a second body, the
second body exerts an equal but oppositely directed force of the same
kind on the first body. In applying the third law, it is important to
remember that the two forces involved act on different bodies.
The third law states that the force one body exerts on another
body (the “action”) is equal and opposite to the force the second body
exerts on the first (the “reaction”). The law does not, however, state
that these forces must lie along the same line. Thus, in Figure 3.2 we
see that in both case (a) and case (b) the forces are equal and opposite,
but in case (b) the forces do not act along the same line. When the
action-reaction forces lie along the same line we say the third law is
obeyed in its strong form. Otherwise, we say it is obeyed in the weak
form.

(a) Strong Form (b) Weak Form

Figure 3.2. Illustrating the strong form and the weak

form of Newton’s third law. The arrows represent the
forces acting on the particles. In both cases the forces
are equal and opposite, but in the strong form the forces
act along the line joining the particles.

Newton’s third law is intimately related to the law of conservation of

linear momentum. Consider, for example, an isolated system consisting
of two particles that exert equal and opposite forces on each other:
F1 = −F2 .
By the second law, the forces can be expressed as changes in the mo-
mentum of the particles:
dp1 dp2
=− .
dt dt
 3.4. NEWTON’S THIRD LAW: ACTION EQUALS REACTION 79

Therefore,
d
(p1 + p2 ) = 0.
dt
That is, the total momentum of an isolated system is constant.
Is the third law always obeyed? This is a question that physicists
have debated for many years. We can state unequivocally that the third
law is always obeyed in purely mechanical systems, but the situation
is not quite as clear when we consider the electromagnetic interaction
between charged particles, as illustrated by the following example.

Worked Example 3.1. Consider the action-reaction pair of

forces for two charged particles interacting through the magnetic
force. See Figure 3.3. Is the third law obeyed?
Solution: Figure 3.3 shows two moving charged particles. The
problem is to determine if the forces they exert on each other are
equal and opposite.
You remember from your introductory physics course in elec-
tricity and magnetism that the force on a charge q1 moving at
velocity v1 in a magnetic field B12 is
F = q1 v1 ×B12 .
The magnetic field B12 acting on q1 is due to the motion of charge
q2 and is given bya
B12 = (q2 v2 × r)(µ0 /4πr3 )
where r is the vector from q2 to q1 .
Now it is easy to see that B12 6= 0, but B21 = 0 because v1 ⊥ r.
Therefore, the force on q1 is non-zero, but the force on q2 is zero.
We might therefore conclude that the third law has failed! This
result is actually cited in many physics texts as evidence that the
third law is not universal and does not apply to electromagnetic
forces.
But wait! Our conclusion was reached too hastily! Upon further
(and much deeper) analysis one finds that the two moving charged
particles set up electromagnetic fields that have momentum and
energy associated with them. When one takes into consideration
the forces acting on the particles due to the rate of change of elec-
tromagnetic momentum, then the third law is found to hold. Fur-
thermore, if the charges are part of electric currents in two nearby
circuits, it is easy to show that the forces exerted by the circuits
on one another are equal and opposite. Consequently, it is safe to
80 3. NEWTON’S LAWS: DETERMINING THE MOTION

conclude that the third law has the same general range of validity
as the first two.b
aThis relation is based on the Biot-Savart law which, strictly speaking, is not
valid for point charges. However, the correct expressions reduce to the given
formula as long as the velocity of the particle is much less than the speed of
light. (D. J. Griffiths, Introduction to Electrodynamics, 3rd Ed., Prentice Hall,
1999, p. 439.
bRoald K. Wangsness, Electromagnetic Fields, 2nd Ed., Wiley and Sons, New
York, 1986, pages 219 and 359. For a different point of view, see “Classical
Dynamics” by Jerry Marion and Stephen Thornton, 3rd Ed., Harcourt, Brace,
Jovanovich, 1988, page 45.

q1 q2
r12
+ +
v1
v
2

Figure 3.3. Two moving charged bodies give rise to an

apparent counter example to Newton’s third law. For
this arrangement v1 × r12 = 0 whereas v2 × r21 6= 0.

The point of the preceding example is that you, as a physicist, must

be very careful in applying the third law. As an illustration, let me
give you a puzzle. You may have heard it before. It is the story of a
donkey hitched to a cart. The donkey pulls on the cart. The cart pulls
back on the donkey. By Newton’s third law, action equals reaction, and
these two forces are equal and opposite. Thus it would appear that the
forces cancel out. How, then, is it possible for the cart to move?
Give up? Well, the answer is that the forces do not cancel because
they are acting on different bodies! Suppose a (single) body is acted
upon by two forces of equal magnitude but opposite directions. You
can then say that these forces cancel. But when you are adding up
the forces that are acting on a particular body, you certainly cannot
include forces that are acting on a different body! You would never
say the force the Sun exerts on the Earth is cancelled by the force the
Earth exerts on the Sun. These forces are indeed equal and opposite,
 3.5. IS ROTATION ABSOLUTE OR RELATIVE? 81

but they act on different bodies. Similarly, for the cart and donkey
problem, the donkey exerts a force on the cart and the cart exerts an
equal and opposite force on the donkey. Considering only the cart, the
forces on the cart are the force exerted by the donkey and frictional
forces (the road on the wheels, etc.). If the force exerted by the donkey
is greater than the frictional force, the cart accelerates forward. You
should also think about the forces acting on the donkey and explain
why the donkey is accelerating forward and not backward. Note that
the only forces that can make an object accelerate are external forces.
You cannot pick yourself up by pulling on your shoestrings!
In determining action-reaction pairs, it is useful to remember that
they are always the same kind of force. For example, a book on a
tabletop is pulled downwards by the gravitational attraction of the
Earth. It is easy to make the error that the reaction is the upward
normal force of the table. But this is not the same kind of force. The
reaction is the upward gravitational pull of the book on the Earth.
What is the reaction force to the normal force exerted by the table?

3.5. Is Rotation Absolute or Relative?

Linear motion is relative. A person in a moving bus is at rest
relative to the bus, but has non-zero velocity relative to the ground.
The velocity of an object depends on (is relative to) the reference frame
in which it is viewed.
Galileo suggested that a person in the hold of ship sailing in per-
fectly smooth water would not be able to determine whether or not the
ship was moving. According to Newton (and later Einstein) it is impos-
sible to distinguish between one inertial reference frame and another.
If you were to place a TV camera inside a closed box, there would be no
physical phenomenon you could observe and no experiment that you
could carry out that would tell you whether the box was at rest, or
moving at a constant velocity.
Is rotational motion also relative? Newton pondered this question
and decided that rotation is absolute. He described an experiment in
which he suspended a bucket full of water from a long rope. He turned
the bucket, twisting the rope. Upon releasing the bucket, he observed
the water level while the rope unwound and the bucket rotated. He
noted that initially, when the bucket and water were at rest, the surface
of the water was flat. Next, the bucket began to rotate, but the water
was still at rest (for a while) and the surface of the water was still flat.
Finally, the water took on the rotation of the bucket and the surface
of the water became concave. (I am sure none of this surprises you.)
82 3. NEWTON’S LAWS: DETERMINING THE MOTION

Now in the initial and final stages, the water was not moving relative
to the bucket, but the surface was flat in the initial stage and concave
in the final stage. Therefore, the behavior of the surface was not due to
the motion of the water relative to the bucket. Also, when the bucket
was rotating but the water was not rotating, the surface remained
flat. Newton concluded that the concavity of the surface was due to
the absolute rotation of the water and not its motion relative to the
bucket.
If you placed the bucket of water in a closed box and set the box
rotating, a TV camera inside the box would show the surface of the
water to be concave. Thus there is a physical phenomenon that can be
used to determine whether or not a reference frame is rotating. There
is no other reference frame (Newton decided) in which the surface of
the water is flat, so rotation is not relative.8
Ernst Mach was fascinated by this simple experiment. However,
he believed that all motion was relative, including rotational motion.
Mach claimed that rotation was relative to all the mass in the universe.
He reasoned that a bucket of water at rest would have a flat surface,
just as Newton observed. But, he asked, would the surface be curved
if the bucket were spun in a totally empty universe? In other words,
what would the bucket be spinning relative to? From Mach’s point
of view, it is the rest of the universe that causes the surface to be
curved, and the same effect would be obtained by spinning the entire
universe around a stationary bucket. You cannot determine whether
the water is rotating and the universe is at rest, or the water is at
rest and the whole universe is rotating around it! Einstein’s theory of
general relativity does not include this “total relativity” of Mach, but
later in his life Einstein tried to analyze the consequences of imposing
it on his theory.
Although the question of whether or not rotation is relative or abso-
lute is still being debated,9 we shall go along with Newton and assume
that rotation is absolute. This point of view is supported by a re-
cent in-depth study10 of Mach’s principle that shows there are many
effects that cannot be explained by a rotating universe and a stationary
8Newton stated that centrifugal force is the feature that distinguishes absolute
motion from relative motion.
9An interesting article on this subject is “Total Relativity: Mach 2004” by
Frank Wilczek, Physics Today, April 2004, page 10. Newton’s bucket is also used
as a starting point for the very readable book, The Fabric of the Cosmos: Space,
Time and the Texture of Reality by Brian Greene, Knopf, New York, 2004.
10H. Hartman and C. Nissim-Sabat, “On Mach’s critique of Newton and Coper-
nicus,” Am. J. Physics, 71, 1163 (2003).
 3.6. DETERMINING THE MOTION 83

bucket. For example, Mach’s principle cannot explain how two buck-
ets, rotating in opposite directions, can both have concave surfaces.
Finally, we might note that in Quantum Mechanics we find that rota-
tional motion is quantized whereas linear motion is not. A body can
have any linear velocity whatever, but a body can only have certain
discrete values for rotational motion.

3.6. Determining the Motion

Recall that “determining the motion” means obtaining an equation
for the position of the body as a function of time. This is done by
solving the equation of motion (which could simply be Newton’s second
law). The equation of motion is a second order differential equation
for the position, so solving for the motion involves carrying out two
integrations. In this section you will learn how to obtain expressions
for the velocity and position of a particle subjected to several different
kinds of force. Specifically, we will consider the following situations:
Force is constant: F = const
Force is a function of time: F = F (t)
Force is a function of velocity: F = F (v)
Force is a function of position: F = F (x)
Unless otherwise noted (and to keep things simple) the motion will
be one dimensional and the object that is moving will be a particle of
constant mass. The generalization to two or three dimensions is quite
straightforward.
Constant Force
If the force is constant, from F = ma we appreciate that the accel-
eration is constant. The equation of motion is is
ẍ = F/m = constant = a,
and integrating dv = adt yields
v(t) = v0 + at. (3.2)
Integrating again we obtain
1
x(t) = x0 + v0 t + at2 . (3.3)
2
Equation (3.3) giving x = x(t) is the solution of the equation of
motion. Thus, we have “determined the motion.”
You learned Equations (3.2) and (3.3) in your introductory physics
course. Perhaps you did not fully appreciate at that time that these
84 3. NEWTON’S LAWS: DETERMINING THE MOTION

equations are valid only for a constant force. The rest of this chapter
deals with forces that are not constant.

Exercise 3.4. It is observed that the position of an object of mass

3 kg is given by x(t) = 3t + 6t2 meters. Determine the force acting on
it. Answer: 36 N.

Force as a Function of Time

A somewhat more complicated situation arises when the force F can
be expressed as a function of time, i.e., F = F (t). Then the equation
of motion is

F (t)
a= .
m
or
dv 1
= F (t).
dt m
Separating variables and integrating:
Z v
1 t
Z
dv = F (t)dt.
v0 m 0
In writing this last equation I was careless and used the same symbol (v)
for a variable of integration and a limit of integration. (This is the kind
of thing that drives mathematicians crazy!) You can usually get away
with this sloppy notation, but sometimes it can get you into trouble.
So I will rewrite the equation above and use double primes for the
variables of integration and single primes for the limits of integration.
Z v0 =v(t0 ) Z 0
00 1 t
dv = F (t00 )dt00 .
v0 m 0

Integrating and rearranging a bit gives

Z 0
0 1 t
v(t ) = v0 + F (t00 )dt00 . (3.4)
m 0
To evaluate the integral in Equation (3.4) you need an explicit ex-
pression for the force as a function of time. Suppose that you do have
such an expression. Then integrating once again you obtain the posi-
tion as a function of time. Since v(t0 ) is given by Equation (3.4) you
 3.6. DETERMINING THE MOTION 85

can substitute and integrate to obtain

Z x(t0 ) Z t" Z 0 #
1 t
dx = v0 + F (t00 )dt00 dt0 .
x0 0 m 0

So,
Z t "Z t0
#
1
x(t) = x0 + v0 t + F (t00 )dt00 dt0 .
m 0 0
Although this expression looks rather forbidding, the procedure for
1
R
finding x = x(t) is not at all difficult: you simply integrate Rm F (t)dt
to get the velocity as a function of time and then integrate v(t)dt to
obtain the position as a function of time (and don’t worry about all
those primes and double primes).

Worked Example 3.2. A particle of mass m is acted upon

by a force F (t) = Ae−bt . The particle is initially at rest at x = 0.
Determine the velocity and position of the particle at time t = 1/b.
Solution: From F = ma
dv
a = F/m = (A/m)e−bt =
dt
so
Z t t
−bt A −bt  A  −bt 
v − v0 = (A/m)e dt = − e  = − e −1
0 mb 0 mb
A
where v0 = 0. For t = 1/b the velocity is v = mb (1 − 1e )
To determine the position we must use the general expression
for the velocity, not its final value. That is,
Z t=1/b Z t=1/b
A  −bt 
x − x0 = vdt = − e − 1 dt
0 0 mb
 1/b
A −bt A
= 2
e + t
mb mb 0
A −1 A 1 A
= e + −
mb2 mb b mb2
A 1
= ,
mb2 e
where x0 = 0.

Exercise 3.5. A force F = 3t2 N acts on a particle for 3 s after

which the particle moves freely. The particle is initially at the origin
86 3. NEWTON’S LAWS: DETERMINING THE MOTION

with zero velocity. Determine its position at time t=5 s. Assume the
mass of the particle is 0.1 kg. Answer: 742.5 m.
Exercise 3.6. A certain electromagnetic wave has its electric field
oriented along the z-axis. The magnitude of the field varies in time
according to E = E0 cos ωt. Initially, an electron is at rest at the
origin. Determine the motion of the electron. (Recall that the force
exerted on a charged body in an electric field is given by F = qE.)
Answer: x = −(qE0 /ω 2 m)(cos ωt − 1).

Force as a Function of Velocity

In nature many forces depend on the velocity of the body. For
example, a particle moving through a fluid, such as a marble falling
through water or a baseball thrown through the air, will experience a
resistive force. This force can usually be assumed to be proportional to
the velocity for small speeds and proportional to the velocity squared
for higher speeds. (Of course, this is just an approximation; the actual
dependence of the resistive force on the velocity is more complicated.
See, for example, the fluid mechanics books by Batchelor11 or by Lan-
dau and Lifshitz.12) Another velocity dependent force is the Lorentz
force. This is the force on a charged particle moving with velocity v in
a region of space where there is an electric field E and a magnetic field
B. You probably remember from your introductory course in electric-
ity and magnetism that the Lorentz force is given by F = q(E+v × B).
(The magnetic force was considered in Worked Example 3.1 in connec-
tion with Newton’s third law. It presents us with a rather complicated
situation because it depends on the direction of the velocity as well as
on its magnitude. In this section we will only consider forces that act
in a constant direction.)
Consider the one dimensional problem of determining the motion of
a particle if the force F is a function of the magnitude of the velocity.
That is,
F = F (v).
The equation of motion is
dv F (v)
= .
dt m
11G. K. Batchelor, An Introduction to Fluid Dynamics, Cambridge University
Press, 1970.
12L. D. Landau, and E. M. Lifshitz, Fluid Mechanics, Vol 6 of Course of The-
oretical Physics, Pergamon Press, Oxford, 1959.
 3.6. DETERMINING THE MOTION 87

Separating variables and integrating:

Z v(t)
1 t
Z
dv
= dt.
v0 F (v) m 0
So,
Z v(t)
dv
t=m .
v0 F (v)
Given an explicit expression for F (v) you can carry out the integration
and obtain an equation for t involving v and v0 (as well as some other
parameters such as the mass and any other constants that appear in
the expression for the force). You have now obtained an expression of
the form t = t(v). This can usually be inverted to yield the desired
form, v = v(t). To be specific, after inverting you will have
v = v(t, v0 , α)
where α is the set of constant parameters mentioned above. You then
use the definition
dx
v= ,
dt
and integrate again to obtain the position as a function of time. That
is,
Z x(t) Z t
dx = v(t, v0 , α)dt,
x0 0
or Z t
x(t) = x0 + v(t, v0 , α)dt.
0
This procedure is a bit complicated so I will give you an example.
Please go through the steps carefully.

Worked Example 3.3. Imagine you are paddling a canoe.

When you appproach the dock you quit paddling and let the re-
sistance of the water bring you to a stop. Assume this force is
proportional to the first power of the speed. Determine the mo-
tion.
Solution: By Newton’s second law, F = m dv dt
. According to
the assumption, F = −bv where b is the constant of proportionality
and the negative sign indicates that it is a retarding force. Then,
−bv = m dvdt
, or dv
v
= − mb dt. Hence,
Z v(t)
b t
Z
dv
=− dt,
v0 v m 0
88 3. NEWTON’S LAWS: DETERMINING THE MOTION

and
v(t) bt
ln v|v(t)
v0 = ln =− ,
v0 m
or
v(t) = v0 e−bt/m .
Thus we have obtained an expression for the velocity at any given
time; half of our job is finished. Integrating again we will obtain
the position as a function of time. Starting with the definition of
velocity, v = dx
dt
, we write
dx
= v(t) = v0 e−bt/m .
dt
Consequently,
Z x(t) Z t
dx = v0 e−bt/m dt.
x0 0
Integrating,
h vm it v0 m −bt/m
0 −bt/m
x(t) − x0 = − e =− (e − e0 ),
b 0 b
or
v0 m
x(t) = x0 + (1 − e−bt/m ).
b
It is interesting to consider the limiting value of our result by letting
t → ∞.13 This leads to the rather un-intuitive result for the velocity of
the canoe because it indicates that the velocity of the canoe [v0 e−bt/m ]
does not go to zero until t → ∞. The canoe is in motion for an infinite
time! How far do you suppose it will travel in this infinite amount
of time? To answer this question consider the expression for position
and set t = ∞ to obtain x(t = ∞) = x0 + v0 b/m. Note that this is
a finite distance. So, even though it takes an infinitely long time for
the canoe to stop, it only goes a finite distance. Does this result make
sense? Yes, it does. The retarding force gets smaller and smaller as
the velocity decreases. Perhaps you might complain that the problem
is poorly posed and it should ask for the time required for the velocity
to fall below some particular value. For example, once the canoe is
moving at some very small velocity, such as one inch per hour, then for
all intents and purposes it is stopped.
If an object is moving in air at less than about 20 m/s (≈ 45 mph),
it is safe to assume the retarding force is proportional to the first power
of the speed. For objects moving at higher speeds (but less than the

13It
is a good idea to subject your answers to a “sanity check” by seeing how
they behave in limiting cases such as t = 0 and t = ∞.
 3.6. DETERMINING THE MOTION 89

speed of sound) it is more realistic to assume the resistive force is

proportional to the speed squared. That is, F = −Dv 2 . The equation
of motion can then be written in the form
dv
m = −Dv 2 .
dt
The proportionality constant D depends on the size and shape of the
body and the density of the fluid through which it is moving. A rea-
sonable formula for calculating D is
1
D = CD Aρ,
2
where the “drag coefficient” CD is a unitless parameter of the order
unity which depends on the shape of the body. In practical applications
the value CD = 0.2 is often used. The quantity ρ is the air density, and
A is the cross-sectional area of the object.
If you apply this law to a body falling through air under the action of
the force of gravity, the resistive force is upwards and the gravitational
force is downwards, so the equation of motion becomes
dv
m = +Dv 2 − mg.
dt
In applying this last equation you will have to be very careful with
the signs. If the body is rising, then both gravity and the force of air
resistance are acting downwards.
Worked Example 3.4. A particle of mass m is acted upon by
the gravitational force (−mg) and a retarding force given by bv 2 .
It is dropped from rest at an initial height x0 . Obtain expressions
for its velocity and position as a function of time.
Solution In problems such as this one, it is often convenient to
express relations in terms of the “terminal velocity” which is the
speed when the two forces are equal and opposite and the object
is no longer accelerating. In this problem the net force is
F = mg − bv 2
where we let “down” be positive. The terminal velocity is obtained
by setting F = 0, so p
vT = mg/b.
Now we solve for the velocity as a function of time using
dv F b
a= = = g − v2
dt m m
90 3. NEWTON’S LAWS: DETERMINING THE MOTION

or
dv b gm b
= ( − v 2 ) = (vT2 − v 2 )
dt m b m
Therefore,
Z v(t)
b t
Z
dv
2 2
= dt
v0 =0 vT − v m 0
which yields
1 v b
tanh−1 = t
vT vT m
and consequently,
r
vT bt bg
v(t) = vT tanh = vT tanh t.
m m
The position is given by
Z t Z t r
bg
x(t) − x0 = − vdt = vT tanh tdt,
0 0 m
where the minus sign is due to the decrease in height of the object
with time. A udu substitution leads to
Z √ gb
m u= m t
x(t) − x0 = − tanh udu
b u=0
and finally
m gb
x(t) = x0 − log cosh t.
b m

Exercise 3.7. A body is dropped from rest from a hot air balloon.
Make the unrealistic assumption that the resistive force is proportional
at all times to the velocity, i.e., fR = −bv. (a) What is its terminal
velocity vT ? (b)How much time is required for the body to reach a
velocity of 0.9 vT ? Answer: (b) 2.3m/b.
Exercise 3.8. A body is falling under the effects of gravity and a
retarding force proportional to the
p square of the speed. Determine its
terminal velocity. Answer: vT = gm/b

Force as a Function of Position

The most important forces you will encounter in this course are
forces that depend on position. These forces are most easily treated
using conservation of energy, and you will do that later. However, for
 3.6. DETERMINING THE MOTION 91

the sake of completeness, let us obtain the motion by a straightforward

integration of the equation of motion.
If F = F (x), the second law is
d2 x
m = F (x).
dt2
This differential equation is not hard to solve if you first separate the
variables. Use the chain rule to write
d2 x dv dv dx dv
2
= = = v. (3.5)
dt dt dx dt dx
Therefore,
dv 1
v = F (x),
dx m
or
1
vdv = F (x)dx.
m
Having separated the variables, you can now integrate to obtain
1 x
Z
1 2 v
2
v |v0 = F (x)dx. (3.6)
m x0
You will need an explicit expression for F (x) to carry out the integral
on the right. A bit of algebra will then yield an expression for the
velocity as a function of position
v = v(x). (3.7)
But what you really want is an expression for the position as a function
of time, x = x(t). Writing dx/dt for v in Equation (3.7) and rearranging
generates a differential equation involving x and t, namely,
dx
= dt.
v(x)
Therefore, Z x Z t
dx
= dt.
x0 v(x) 0
That is, Z x
dx
t= .
x0 v(x)
Assuming this integral can be carried out, you will obtain an expression
of the form t = t(x). Finally, this must be inverted to yield x = x(t),
a mathematical procedure that may involve a significant amount of
algebra.
Simple Harmonic Motion: A particle that oscillates back and
forth periodically is undergoing simple harmonic motion, (SHM).
92 3. NEWTON’S LAWS: DETERMINING THE MOTION

This particularly important type of motion is generated by a force that

is a function of position. Specifically, the force that leads to SHM is a
force that is always directed toward a fixed point and is proportional
to the distance from the body to the fixed point.

k
m

Figure 3.4. A mass m connected to a massless spring

of constant k on a frictionless surface.

Worked Example 3.5. A prime example of simple harmonic

motion is the motion of a block of mass m attached to a spring of
force constant k, and sliding on a frictionless horizontal surface, as
illustrated in Figure 3.4. Obtain and solve the equation of motion.
Interpret the expression for position as a function of time.
Solution: The force exerted by the spring is F = −kx where
x is the amount the spring has been stretched. This is equal to
the displacement of the block from the equilibrium position. The
equation of motion is
mẍ = −kx. (3.8)
Going through the steps from Equation (3.5) to (3.6) you obtain
2 x
Z
2 2
v = v0 + (−kx)dx.
m x0
k
∴ v 2 = v02 − (x2 − x20 )
m
or r
k
v = v02 − (x2 − x20 ).
m
2 k 2
Note that v0 + m x0 is a constant whose value depends on the initial
conditions. Denoting this constant by C you can write
r
dx k
v= = C − x2 .
dt m
 3.6. DETERMINING THE MOTION 93

Separate variables again and integrate to obtain the motion.

Z x Z t
dx
q = dt,
k 2
x0 C−m x 0

or r
x
k t
Z Z
dx
p = dt.
x0 mC/k − x2 m 0
The integral on the left can be found in any table of integrals,
yielding x r t
x  k 
sin−1 p  = t .

mC/k  m 
x0 0
Therefore
r
−1 x −1 x0 k
sin p − sin p = t.
mC/k mC/k m
The second term on the left is just a constant; call it β. Then
r
−1 x k
sin p = t + β,
mC/k m
or r r !
mC k
x= sin t+β .
k m
This is usually written in the easily remembered form
x = A sin (ωt + β) . (3.9)
The constant r r
mC mv02 + kx20
A= =
k k
is the amplitude. The quantity
p
ω = k/m
is the (angular) frequency of the oscillation. The constant β is
called the “phase constant” and is related to the position of the
oscillator at time t = 0. By adding π/2 to β one can express x(t)
in terms of the cosine rather than the sine.

Exercise 3.9. Verify Equation (3.9) by direct substitution into the

equation of motion.
94 3. NEWTON’S LAWS: DETERMINING THE MOTION

3.7. Numerical Method to Determine the Motion (Optional)

3.7.1. Closed Form Solutions. It is usually fairly easy to deter-
mine the forces acting on a body and to write the equation of motion.
However, it is not always possible to solve the equation of motion in
closed form. That is, it is not always possible to write down an ana-
lytical solution in which the position is expressed as a function of time
in terms of simple functions. For example, x = 3t2 , and x = 4 cos 5t
are analytical solutions. However, there are ways to describe the mo-
tion that do not involve analytical expressions. For example, you might
measure the position of an object in the laboratory and draw up a table
giving its position at various times. This would give you a collection
of numbers that you could use to determine the position of the body
at any given time, but you would not have a solution in closed form.
Similarly, a graph of x vs t is a visual representation of position as a
function of time, but once again it is not in closed form. A closed form
(or analytical solution) is a relationship between x and t in terms of
elementary mathematical functions, namely powers, roots, logarithms,
exponentials, and trigonometric functions.
If you have the equation of motion for a system and you know the
initial conditions, you can always write a computer program that will
yield a “numerical solution” giving the position and velocity of the
body at any desired time. With the advent of large, fast computers,
numerical solutions have become very common in physics. In fact, there
is a whole branch of physics called “computational physics” in which
you use a computer to solve physics problems by applying numerical
methods.14 Many important problems cannot be solved analytically; in
recent years more and more work has gone into developing sophisticated
techniques for obtaining numerical solutions using computers.
You may be wondering why we bother to study the techniques for
obtaining analytical solutions instead of just writing a computer pro-
gram to solve the problem. There are many reasons. For one thing,
numerical techniques for solving physics problems are often based on
the analytic solution of a similar (usually simpler) problem. Further-
more, numerical techniques are nearly always verified by determining
whether or not they can reproduce a known analytic solution. Knowing
analytical techniques for solving the equations of motion is extremely
14An excellent computational physics textbook is “Numerical Methods for
Physics, 2nd Edition, ” by Alejandro L. Garcia, CreateSpace Publishing 2015.
 3.7. NUMERICAL METHOD TO DETERMINE THE MOTION (OPTIONAL) 95

helpful in obtaining and interpreting numerical solutions. It might also

be mentioned that numerical solutions have a number of disadvantages.
A numerical solution, as the name implies, is just a number and not a
formula. Numerical solutions yield a specific answer to the problem at
hand and very rarely lead to general formulas.
In this book, I emphasize obtaining analytical solutions for prob-
lems that have closed form solutions. You will find that the techniques
developed here will be used frequently in your other physics courses
and in your professional career. If you end up working as a Computa-
tional Physicist you will often apply the material you are learning in
this course.
You have seen that, in principle, the equation of motion can be
solved analytically for a number of different types of forces, specifically
for forces that are functions of time, velocity or position.
All of the procedures described above basically boil down to in-
tegrating the equation of motion twice, leading to expressions of the
form:

v = v(t, v0 , x0 ),
x = x(t, v0 , x0 ).

These give the position and velocity as functions of time and the ini-
tial conditions, v0 and x0 . If you are lucky, these expressions are in
closed form, that is, in terms of familiar functions such as logarithms,
exponentials, trigonometric functions, etc. (You are comfortable with
these so-called “elementary functions,” but if you were confronted with
an expression involving the gamma function or an elliptic integral, you
might not feel so happy.)
If v(t) and x(t) are given in closed form, you have a great deal
of knowledge about the motion. First of all, you can determine the
velocity and position at any time by simply plugging the time into
the equation. Furthermore, you can simply look at the equation and
get a very good idea about the motion. For example, if you obtain
the solution x = A cos ωt you immediately know that the motion is
oscillatory with amplitude A and angular frequency ω. Generally, it is
easy to manipulate the solution to obtain more information about the
system. For example, squaring the expression for the velocity to get v 2
immediately leads to an equation for the kinetic energy as a function
of time. If the motion is two-dimensional and you obtain solutions x(t)
and y(t) you can easily combine them to get an equation for the path
(or orbit) of the body.
96 3. NEWTON’S LAWS: DETERMINING THE MOTION

A closed form analytic solution of the equation of motion contains

all the essential physical information about the system. (In this sense,
it is similar to the quantum mechanical wave function from which one
can extract all knowledge of the system.) However, it often happens
that such a solution is not available. Perhaps we are not able to solve
the equation of motion due to a lack of mathematical knowledge, or
perhaps the equation of motion is actually insolvable. In such a case,
we are forced to use a different technique, namely, we are forced to
evaluate a numerical solution.
Since numerical solutions are seldom as useful as analytical solu-
tions, you should make sure that you cannot find an analytical solution
before you start writing a computer program. An analytical solution
can usually be found in much less time than it takes to write, debug,
and run a computer program. With this warning out of the way, I
will now describe an elementary technique for obtaining a numerical
solution to the equation of motion.15
A numerical solution of a differential equation essentially involves
replacing the differential equation with a difference equation. For ex-
ample, suppose you wanted to solve
dx
= f (x, y),
dy
You would replace this expression, in which dx and dy are infinitesimal
quantities, with the expression
∆x
= f (x, y).
∆y
Here ∆x and ∆y are small quantities but are not infinitesimal. To
generate a solution recall the definition of derivative:
dx x(y + ∆y) − x(y)
= lim .
dy ∆y→0 ∆y
15Physicists frequently use numerical techniques for solving differential equa-
tions which are difficult or impossible to solve analytically. This is especially true if
they are analyzing fluid flow. Fluid dynamics is, of course, a branch of Newtonian
mechanics and the flow of a fluid is controlled by Newton’s laws. However, the
forces involved are very complicated. Some are velocity dependent quantities that
depend on the pressure exerted by other parts of the fluid. Other forces include
viscous drags and the force of gravity. When one writes Newton’s second law for a
fluid, it may have a large number of terms in it. Often, the only practical way to
solve this equation and thus determine the motion of the fluid is to use numerical
techniques and a high speed computer. (The second law, when applied to fluid
flow, is called the Navier-Stokes equation.)
 3.7. NUMERICAL METHOD TO DETERMINE THE MOTION (OPTIONAL) 97

Replace dx/dy = f (x, y) by the approximate relation

x(y + ∆y) − x(y)

' f (x, y),
∆y

and rearrange to get

x(y + ∆y) ' x(y) + f (x, y)∆y.

If you know x and y and can calculate f (x, y) you can use this rela-
tionship to determine x for a slightly larger value of y, namely, y + ∆y.
Note that the relationship is not exact. However, the approximation
becomes better and better as ∆y becomes smaller and smaller.
Let us apply this technique to the equation of motion. We want
to solve for x(t) numerically, given an initial value x0 and an explicit
expression for the acceleration, a = a(x, v, t). The procedure involves
carrying out the following sequence of operations:
1. Select ∆t to be a small time step.
2. Set x = x0 and t = t0 .
3. Determine a = a(x, v, t).
4. Obtain new time from t = t + ∆t.
5. Obtain new velocity from v = v + a · ∆t.
6. Obtain new position from x = x + v · ∆t.
7. Go to Step 3.
You go through this procedure repeatedly. At each step you obtain
new values for x, v, and t.
Consider a simple but important numerical technique called the
Euler method. This is a technique for obtaining the numerical solution
of an equation of the form

d2 x
= f (x, v, t).
dt2
If the function on the right-hand side is the force divided by the mass,
this is just Newton’s second law.
Write the second-order differential equation above as two first-order
equations. These are,

dx
= v(x, t),
dt
dv
= f (x, v, t).
dt
98 3. NEWTON’S LAWS: DETERMINING THE MOTION

Solve as follows: let x1 and v1 be the initial values of x and v. Let the
initial time be t1 and let τ be a small time step. Then
x2 = x1 + v1 τ
v2 = v1 + f (x1 , v1 , t1 )τ
The next step is to replace x1 by x2 and v1 by v2 and let t1 be replaced
by t2 = t1 + τ. Then evaluate
x3 = x2 + v2 τ,
v3 = v2 + f (x2 , v2 , t2 )τ.
You can appreciate that, in general, the technique consists in replacing
xn by xn+1 and vn by vn+1 and repeating as many times as desired.
This is the essence of the Euler method.
Some years ago, Alan Cromer16 noted that the Euler method, as
expressed above, is unstable. That is, the answer diverges further and
further from the correct solution. However, Cromer discovered that the
equations could be made stable by a simple but crucial change, namely
by calculating v first and using the new value of v to determine x. The
so-called “Euler-Cromer” algorithm is
v2 = v1 + f (x1 , v1 , t1 ) · τ,
x2 = x1 + v2 τ.
This simple scheme allows you to numerically determine the velocity
and position of a particle at any future time if you know its acceleration
f (x, v, t) at a prior time.
When carrying out a numerical integration, you should check fre-
quently to determine if the solution is stable. For example, you might
check at each time step to make sure the energy or angular momentum
is conserved.

3.8. Summary
A number of important concepts were presented in this chapter. Ba-
sically, you learned (or reviewed) Newton’s three laws and you learned
how to determine the motion for a mechanical system (assuming you
know the force).
Newton’s three laws are:
(1) The law of Inertia. (A body tends to preserve its state of
motion.)
16Alan Cromer, “Stable solutions using the Euler approximation,” American
Journal of Physics, 49, 455 (1981). In this paper Cromer explains why one algo-
rithm is stable and the other is not.
 3.9. PROBLEMS 99

(2) The second law, F = dp dt

, can often be written as F = ma. It
gives the relation between the net external force acting on a
body and its acceleration.
(3) Action equals reaction. (Two bodies always exert equal and
opposite forces on each other.)
If the mass is constant, Newton’s second law can be expressed as
F
a= .
m
We call such a relation an “equation of motion” because it can be
integrated to determine the velocity v and the position x as functions
of time. Obtaining an expression for x = x(t) is called “determining
the motion.”
To determine the motion you need to integrate the equation of
motion twice, thus:
Z t
v(t) = adt
0
Z t
x(t) = v(t)dt
0
We have described the techniques for determining the motion for four
different kinds of forces. These forces are: (1) constant force, (2) force
a function of time, (3) force a function of velocity, and (4) force a
function of position. Make sure you understand the proper procedure
for each of these cases.
Finally, you were exposed to the method of determining the mo-
tion using numerical techniques on a computer. The “Euler-Cromer”
algorithm is a simple and useful way to obtain x = x(t).

3.9. Problems
Problem 3.1. A particle of mass 3 kg is acted upon by the two
forces (in newtons)
F1 = 6ı̂,
and
F2 = 3ı̂ + 3̂.
Determine the magnitude and direction of the acceleration.
Problem 3.2. A pail is filled with oil to a depth of 10 cm. A
steel marble of mass 0.2 kg is released from rest 50 cm above the top
surface. Assuming the oil exerts a constant resistive force of 2.4 N on
the marble, determine the speed with which it reaches the bottom of
the pail. Answer: 3.06 m/s
100 3. NEWTON’S LAWS: DETERMINING THE MOTION

Problem 3.3. An automobile starts from rest and accelerates at

a constant rate for 1 km, covering that distance in 20 seconds. Ignore
air resistance. Determine the coefficient of static friction between the
tires and the pavement.
Problem 3.4. A sailor on board a ship of mass 2 × 106 kg moving
at 0.2 m/s throws a heavy rope to a man standing on the wharf who
drops the noose around a bollard. After the rope is taut, the ship
is brought to rest, stretching the rope 0.5 m. Find the average pull
sustained by the rope.
Problem 3.5. A particle of mass m is observed to have a veloc-
ity given by v = A cos αx where A and α are constants. Obtain an
expression for the force F = F (x).
Problem 3.6. (a) Two stars of masses M1 and M2 attract one
another with the gravitational force F = −GM1 M2 /r2 . Determine the
ratio of their accelerations. (b) Assume both stars move in circles about
their common center of mass. Determine the radii of these circles in
terms of M1 , M2 , and r. (Actually the stars will move in elliptical orbits
with the center of mass at their common focal point.)
Problem 3.7. You are searching for an inertial reference frame.
Determine the accelerations of the following reference frames. (a) A
reference frame fixed to the surface of Earth at latitude 37◦ . (b) A
reference frame with origin at the center of Earth with one axis always
pointing towards the Sun. (c) A reference frame with origin at the
center of the Sun with one axis always pointing towards the center of
the galaxy. (The Sun is about 3/5 of the distance from the center of
our galaxy to the edge. It takes some 200 million years to “orbit” the
galaxy. The diameter of the Milky Way is about 105 ly.)
Problem 3.8. Atwood’s Machine consists of two masses m1 and
m2 tied together by a light string that passes over a smooth pulley.
Assume the pulley is massless (so it has zero moment of inertia). (a)
Obtain an equation for the acceleration of the masses and the tension
in the string. (b) Obtain an expression for the acceleration if the pulley
is not smooth and has moment of inertia I and radius R.
Problem 3.9. One normally learns Newton’s three laws as state-
ments about the nature of the physical universe. However, some people
prefer to consider only the third law as a law of nature and the first
two as definitions. Assume that you subscribe to this point of view.
Write a short essay (one or two paragraphs) explaining why you are
interpreting Newton’s laws in this manner.
 3.9. PROBLEMS 101

Problem 3.10. The gravitational force between two masses is

F = −G (M1 M2 /r2 ) and the electrostatic force between two oppo-
site charges is F = −k (Q1 Q2 /r2 ) . (a) Show that all massive objects
attracted to the Earth will have the same acceleration independent of
mass. (b) Show that two charged objects attracted to a point charge Q
will not have the same acceleration unless both have the same charge
to mass ratio.
Problem 3.11. A friend of yours, hearing that the universe is
expanding at an increasing rate, theorizes that there is a universal re-
pulsive force between any two masses, as well as the usual gravitational
force. According to your friend, this hypothetical repulsive force de-
creases with the inverse of R rather than the inverse square of R and
is given by
M1 M2
F = G0
R
where G0 = 6.67 × 10−31 Nm/kg2 . Determine the separation between
two bodies when the repulsive force would become dominant. Could
this explain the expansion of the universe? Would a galaxy the size
of the Milky Way be stable under such a force? (The diameter of the
Milky Way is about 100,000 ly.)
Problem 3.12. A certain satellite in a circular orbit around Earth
has a period T. Suppose the gravitational force were not exactly an
inverse square force, but rather depended on distance as 1/rα where
α = 2 +  where  is a small number. If the value of  were 10−4
how would this affect the period of the satellite? Would this be a
measurable quantity? (You may assume that in SI units the universal
gravitational constant has the same numerical value, although the units
would have to be different. Let the distance from the center of the
perfectly spherical Earth to the satellite be 7000 km.)
Problem 3.13. Suppose that gravitational mass and inertial mass
are different but that they are proportional to one another. Would
heavy objects and light objects then fall at the same rate? As an
explicit example, assume that inertial mass is twice as great as gravita-
tional mass. Determine the acceleration of a freely falling object near
the surface of the Earth.
Problem 3.14. Consider the following three systems: (1) A book
is sitting on a table which is at rest on the surface of Earth. (2) A
rocket is taking off from the surface of Earth. A large cloud of burned
fuel forms a contrail behind it. (3) A donkey that is hitched to a
cart is trotting down a road. (a) Identify the action-reaction pairs in
102 3. NEWTON’S LAWS: DETERMINING THE MOTION

these three cases. (b) What is the force that is causing the rocket to
accelerate? (c) What is the force on the donkey that keeps it in motion?
Problem 3.15. The Aristotelians noticed that if you push a box
across the room, it moves with a constant speed, and if you push it
harder the speed is greater, but as soon as you quit pushing, the box
stops. This suggests that the equation of motion should be F = mv.
Furthermore, the Aristotelians claimed that heavier objects fall to the
ground faster than light objects. (Thus, an object four times heavier
than a light object will fall to the ground in one fourth the time.)
(a) Show that these assumptions lead to the absurd conclusion that
the Earth exerts the same gravitational force on all bodies, regardless
of their mass. (b) Describe a simple experiment to show that the
Earth does not exert the same gravitational attraction on all bodies,
regardless of their mass.
Problem 3.16. Two carts of masses M and 2M have springs at-
tached to either end as shown in Figure 3.5. Cart 1 has mass M and
cart 2 has mass 2M. They are on a frictionless segment of track of
length L with barriers at the ends. The two carts are brought together
at the center of the track, the springs in contact are compressed, and
the carts are released from rest. Cart 1 moves left, hits the barrier at
x = 0 and bounces back. Cart 2 moves right, hits the barrier at x = L
and bounces back. The collisions are completely elastic. Where do the
carts meet? The dimensions of the carts are much smaller than the
length of track, so the carts can be treated as particles. (b) Where do
the carts meet if cart 2 is four times more massive than cart 1?

1 2
0 L/2 L

Figure 3.5. Two carts with attached springs on a fric-

tionless track. If the carts are squeezed together at the
midpoint, where will they meet? The mass of cart 2 is
twice the mass of cart 1. (See Problem 3.16)

Problem 3.17. The velocity of a particle of mass m varies with

distance x as
v(x) = v0 + bx
 3.9. PROBLEMS 103

where b and v0 are constants. (a) Find the force acting on the particle,
F (x). (b) Determine x(t). (c) Obtain an expression for the force as a
function of time, F (t).
Problem 3.18. A certain object is losing mass at a rate κ (kg/sec).
It is acted upon by a constant force F. (a) Determine its acceleration
and its position as a function of time. Assume that at t = 0, the mass
is m0 , the velocity is zero and the object is at x = 0.
 (b)
 Show that
1 F
as κ → 0, the expression for x(t) reduces to x = 2 m0 t2 , that is,
x = 21 at2.
Problem 3.19. An object of mass 2 kg is thrown upwards with an
initial velocity of 4 m/s. It is acted upon by the constant gravitational
force of the Earth and a mysterious force of 2 newtons that always
opposes the motion of the object. (This is like the force of air resistance,
except we are assuming the force is constant.) Determine the position
of the object at time t = 0.5 s. (Be careful; by this time the object is
falling downwards and the mysterious force is acting upwards!)
Problem 3.20. A builder plans to use a chain of length 1.5 meters
and linear density 7 kg/m to suspend a 220 kg mass from an overhead
beam. Determine the forces at either end and at the middle of the
chain. Obtain an equation for the force on a link at any point of the
chain. If the maximum tension that can be supported by a link is 2200
N, where will the chain break?
Problem 3.21. A bathroom scale is placed in an elevator. A man
weighing 180 lbs stands on the scale and observes that as the elevator
ascends, the scale reads 200 lbs while the elevator is accelerating and
140 lbs when it is slowing to a stop. What are the acceleration and the
deceleration of the elevator? (g=32.2 ft/s2 )
Problem 3.22. A bullet initially traveling at 400 m/s passes through
a 2 cm thick slab of wood and emerges at speed vf . When fired into a
thick block of the same wood, the bullet is found to penetrate 10 cm.
Determine vf .
Problem 3.23. A physics student designed an apparatus for her
experimental physics class. The apparatus consisted of a block m =
0.25 kg mass attached to a compressed spring of constant k mounted on
a cart of mass M = 2 kg, as shown in Figure 3.6. The car is pulled by a
string that passes over a pulley to a 10 kg hanging weight. The student
used photogate timers to determine the acceleration of the mass on the
spring when it was 15 cm from its equilibrium position. She found this
104 3. NEWTON’S LAWS: DETERMINING THE MOTION

acceleration to be -2 m/s2 relative to the lab bench. (a) What is the

spring constant? (b) What is the acceleration of the block relative to
the cart? (You may assume there is no friction in the system and the
spring is massless.)

m
M=2kg

10

Figure 3.6. A block on an accelerated cart.

Problem 3.24. Johannes Kepler determined that the period of a

planet squared is proportional to the cube of the semi-major axis of
the orbit. Let us assume planets move in circular orbits, in which case
the semi-major axis is equal to the radius of the orbit. Using this fact,
deduce that the gravitational force obeys an inverse square law. (Note:
You cannot use Newton’s law of universal gravitation!) Also show that
the velocity of a planet decreases proportionally
√ with the inverse square
root of its distance from the sun (v ∝ 1/ r).
Problem 3.25. A singly charged lithium ion is accelerated by an
electric field E =10ı̂ volts/meter applied between two plates ten cen-
timeters apart. After passing through a small hole in the negative
plate, the ion enters a region of space where there is a magnetic field
B =5̂̂ millitesla. The ion then moves in a circular path. What is the
radius of this circle?(You may assume the mass of the Li ion is 7 amu.)
Problem 3.26. A particle of mass 0.1 kg is acted upon by a force
given by
 2
t−C
F =A+B
2
where A = 1 N, B = −1 N/s2 and C = 2 s. Determine the speed and
position of the particle at time t = 4 s, if it is initially at the origin
with a velocity +6 m/s.
Problem 3.27. A particle of mass m, initially at rest, is subjected
to a force F (t) = Ae−αt cos(βt). Determine v(t) and x(t).
 3.9. PROBLEMS 105

Problem 3.28. An object with initial velocity v0 is slowed by a

force given by F = −Aebv where A and b are constants. Obtain an
expression for the velocity as a function of time.
Problem 3.29. An airplane accelerates down the runway. Assume
the jet engines exert a constant force on the airplane and assume air
resistance is proportional to the velocity squared. Obtain an expression
for the velocity as a function of time. What is the terminal velocity?
Problem 3.30. A falling body is subjected to the gravitational
force (acting down) and the retarding force of air resistance (acting
Dv 2 where D is a constant.
up). Suppose the retarding force is given byp
Show that in the limit of small t (t << m/Dg) the velocity and
position are given by
1
v ' −gt, x ' − gt2 ,
2
p
and in the limit of large t (t >> m/Dg)
p m p
v ' − mg/D, x' ln 2 − mg/Dt.
D
Hint: Expand the hyperbolic functions.
Problem 3.31. You throw a ball vertically upward with an initial
velocity v0 . Assume that air resistance generates a force given by Dv 2
opposing the motion. (a) Determine the terminal velocity (note that
this involves the falling motion of the ball). (b) Determine the height
attained by the ball. (c) Determine the speed of the ball when it returns
to your hand (in general this will not be the terminal velocity).
Problem 3.32. A race car of mass 800 kg reaches a velocity of
200 kilometers per hour when suddenly the engine quits operating and
the brakes go out! The only thing slowing the car down is air resis-
tance (rolling friction is assumed negligible). Assume the drag force is
proportional to the square of the velocity (F = −Dv 2 ). After the car
coasts 500 meters it is moving at 5 km/h. What is the drag coefficient
D?
Problem 3.33. A skydiver has a mass of 100 kg (including the
parachute). When the parachute is open, the retarding force is pro-
portional to the velocity to the first power. Furthermore, it is known
experimentally that the parachute exerts a retarding force per unit area
of 90 N/m2 when the speed is 3 m/s. (a) It is desired that the skydiver
not exceed a safe speed of 1.5 m/s upon reaching the ground. What is
the required area of the parachute? (b) Assuming an initial speed of
106 3. NEWTON’S LAWS: DETERMINING THE MOTION

zero, how much time does it take for the skydiver to reach a speed of
90% of the terminal velocity?
Problem 3.34. A bullet of mass m is fired vertically upward with
an initial velocity v0 . Assume the resistive force of the air is given by
mkv where m is the mass, v is the velocity and k is a constant. How
much time is required for the bullet to reach the top of its path?
Problem 3.35. A physics student on the roof of a tall building
drops a ball of diameter 10 cm and mass 0.3 kg. The air density is 1.2
kg/m3 . (a) What is the terminal speed? (b) Obtain an expression for
the speed of the ball as a function of how far it has fallen. (c) If the
roof is 100 meters above the ground, what is the speed of the ball when
it hits the sidewalk?
Problem 3.36. An object is dropped from an airplane that is flying
at an altitude of 6000 m at a speed of 1000 km/hr. Determine the
horizontal distance traveled by the object by the time it hits the ground.
For the sake of making the calculations somewhat simpler, you may
assume that the resistive force of the air is proportional to the first
power of the velocity and that the terminal velocity of the object is 98
m/s. (Justify neglecting small terms.)
Problem 3.37. An object of mass m in a uniform gravitational
field is dropped from some initial height. Assume that the force of air
resistance is proportional to the first power of the velocity (∝ bv). Show
that the distance y it has fallen in time t is
mg  m −bt/m

y= t+ e −1 .
b b
Problem 3.38. Assume the force of air resistance is given by Fair =
2
Dv where v is the velocity and D is a constant equal to 0.01 kg/m.
An object of mass 2 kg initially at rest, is dropped from a height of
1000 m. The gravitational force can be assumed constant. Determine
how far the object has fallen in 20 s.
Problem 3.39. An object of mass 10 kg is dropped from a high
place. Assume air resistance is proportional to the velocity squared
(= Dv 2 ) where D = 0.01 Ns2 /m2 . (a) Evaluate the terminal velocity.
(b) Determine the time required for the ball to reach 0.9vT .
Problem 3.40. A bead on a long straight wire is repelled from the
origin x = 0 by a force proportional to its distance to that point. The
bead is initially at rest at position xo . Determine its motion.
Problem 3.41. A particle of mass m kg is at rest at x0 . It is
subjected to a force of magnitude F = k/x where k = 4 Nm. the
 3.9. PROBLEMS 107

force is directed along the positive x-axis. Determine the speed of the
particle when it passes through the point x = 2x0 .
Problem 3.42. A block of mass m on a frictionless horizontal
surface is attached to two springs in series. The springs have force
constants k1 and k2 . Determine
p pthe angular frequency of oscillation of
the block. Answer: ω = 1/m (k1 k2 )/(k1 + k2 ).
Problem 3.43. An asteroid, initially at rest, falls from a great
distance (x0 ) towards a star. You are asked to determine the time
required for it to hit the star. The star can be assumed to be at rest
at the origin at all times. Evaluate the time required for the asteroid
to reach x = 0, that is, assume the star is a point object.
Problem 3.44. A mass m is attached to a spring of constant k.
Determine the motion if it is initially at the unstretched
R position and
is given an impulse J. (Impulse is defined by J = Fdt.)
Problem 3.45. (a) Obtain a general relation for the motion of a
particle of mass m acted upon by a repulsive force F = +kx. Assume
the particle is initially at the position x = x0 . Note that the solution
will have the form x = Aeβt + Be−βt . (b) It may surprise you to know
that for this force there is a possible solution in which the particle
moves to the origin and remains at rest there. In this case, what is the
initial value of the velocity?
Problem 3.46. An asteroid of mass m is initially very far from
Earth and has zero velocity relative to Earth. It falls under the action
of the gravitational force, F = −GmM/z 2 where M is the mass of
Earth. Determine the speed of the asteroid as a function of z, its
distance from the center of Earth.
Problem 3.47. An enterprising scientist drills a hole straight through
the Earth, from North Pole to South Pole, and drops an object of mass
m in the hole at the North Pole. How long does it take for the object
to emerge from the hole at the South Pole? You may assume that there
is no air in the hole. Hint: The gravitational force inside a uniform
sphere is given by F = −GM 0 m/r2 where M 0 is the mass contained
within a sphere of radius r and r is the distance from the center of the
sphere to the object m.
Computational Projects
Computational Project 3.1. Write a program to plot the ve-
locity and position of the canoe of Worked Example 3.3. Assume the
mass of the canoe plus person is 200 kg. Use b = 150 kg/s as a first
108 3. NEWTON’S LAWS: DETERMINING THE MOTION

estimate and get plots for various values of b. Use a reasonable choice
for the initial velocity.
Computational Project 3.2. Demonstrate that the Euler-Cromer
method is stable but the Euler method is not. Do this by assuming
a particle moves in one dimension (x) under the influence of a force
F = −kx. This will lead to simple harmonic motion. Compare the
total energy as a function of time as obtained using the Euler method
and the Euler-Cromer method. Also, on the same plot, present the to-
tal energy as a function of time as obtained from the exact (analytical)
expression. You will appreciate that the Euler method does not con-
serve energy, the total energy obtained by the Euler-Cromer method
oscillates around a constant value, and the analytical expression gives
a constant value for the energy at all times.
Computational Project 3.3. A projectile is fired perpendicu-
lar to the Earth’s surface with an initial velocity of 600 m/s. Assume a
constant gravitational force acts on the projectile so its acceleration is
9.8 m/s2 downwards. Write a program to determine the position of the
particle as a function of time. Next, consider the same problem, but
now include the effect of air resistance, assuming it is a retarding force
that can be expressed as fair = 2.5 × 10−4 v 2 where v is the velocity of
the projectile.
Chapter 4
The Lagrangian Method

In this chapter you will learn a different approach for solving physics
problems. This approach is associated with Lagrange,1 although many
other physicists and mathematicians contributed to it, including Leib-
niz and Euler.
We shall first introduce a function called the Lagrangian, and then
show how it can be used to determine the equations of motion.
A mechanical system is fully described by the Lagrangian. This
means that if you know the Lagrangian for a system, you can deter-
mine the equation of motion, the momentum, and all other relevant
mechanical quantities.
So far, we have been using Newton’s second law to determine the
equation of motion. The Lagrangian method is a very useful alterna-
tive technique for determing this equation, but before discussing the
Lagrangian, it is helpful to review how Newton’s second law leads to
an expression for the equation of motion.

4.1. The Equation of Motion by Inspection

We have considered the equation of motion to be an equation for
the acceleration as a function of position, velocity and time. That is,
a = a(x, v, t),
or
ẍ = ẍ(x, ẋ, t).

1JosephLouis Lagrange was a French physicist and mathematician who lived

from 1736 to 1813.
109
110 4. THE LAGRANGIAN METHOD

In the preceding chapter you learned how to solve the equation of

motion for a variety of different force laws. But how can you determine
the equation of motion?
The equation of motion for a system can often be determined from
Newton’s second law in the form F = ma. For example, the equation of
motion for a simple pendulum is easily obtained by drawing the forces
acting on the bob and resolving them into appropriate components, as
illustrated in Figure 4.1. A glance at the figure will remind you that the
gravitational force can be resolved into a component along the string
(mg cos θ) and a component (mg sin θ) tangent to the circular path of
the bob. The position of the bob can be specified by the arc length s,
measured from the lowest point in the motion. Consequently, F = ma
becomes
−mg sin θ = ms̈.
This equation involves two variables, s and θ. But s = lθ, where l is
the length of the string. So the equation of motion can be written in
terms of the single parameter θ as:
g
θ̈ = − sin θ. (4.1)
l

l
θ
T

s mgcosθ
mg
mgsinθ
Figure 4.1. Forces acting on a simple pendulum. The
length of the string is l.

I went through this trivial example to remind you that it is easy

to determine the equation of motion for a simple system, such as a
pendulum.
But now consider the problem of determining the equation of mo-
tion for the double pendulum shown in Figure 4.2. For simplicity,
assume the double pendulum is constrained to oscillate in the plane of
the page so the positions of the bobs can be described by the two angles
θ1 and θ2 . Observe that the motion of m2 depends on the motion of
m1 so the tension in the lower string is not equal to m2 g cos θ2 . (The
 4.2. THE LAGRANGIAN 111

reason is that the point of support of the lower pendulum is accelerat-

ing so m2 does not behave like a pendulum in an inertial system.) It
would be very difficult to obtain the accelerations of the two masses by
applying Newton’s second law. A different approach is needed.
I will now describe a very powerful but fairly simple technique called
the Lagrangian approach. This will allow you to obtain the equations of
motion for complicated problems such as the double pendulum. First,
however, I would like to tell you a bit more about the Lagrangian.

Exercise 4.1. Draw a force diagram for each of the masses of the
double planar pendulum. Consider the motion of mass m2 . What is
wrong with writing m2 s̈2 = −m2 g sin θ2 ?

y
x
l
1
θ1
m1

l2
θ m2
2

Figure 4.2. A double planar pendulum.

4.2. The Lagrangian

By studying Lagrangian dynamics you will develop a greater un-
derstanding of the framework of mechanics and an appreciation of how
the physical universe behaves. Lagrangian dynamics goes to the very
core of the science of mechanics. In fact, as you will see in Section 4.7,
the Lagrangian approach can be used to derive Newton’s laws.
If you take a graduate courses in mechanics, you will find yourself
deeply immersed in Lagrangian methods. Here I will only use it as
a technique - an easy way to obtain the equations of motion for a
complicated system. Nevertheless, later in this chapter I will present
an elementary derivation of the technique, showing how Lagrange’s
112 4. THE LAGRANGIAN METHOD

equations are obtained using the calculus of variations. You will learn
that Lagrangian dynamics is based on a profound statement about
the nature of the physical world called Hamilton’s principle. Later (in
Chapter 8) you will see how the Lagrangian is related to symmetries
in physical systems, and to the conservation laws of physics.
The Lagrangian method, as described here, is applicable to systems
in which all the forces are conservative. Consequently, we will not
consider systems with frictional forces. Although there are ways of
incorporating dissipative forces into the Lagrangian formulation, they
involve advanced methods and we do not consider them in this book.
So if friction is present, you must go back to Newton’s second law.
The study of Lagrangian dynamics begins with the definition of a
physical quantity called the Lagrangian. It is denoted by L and is de-
fined as the difference between the kinetic energy (T ) and the potential
energy (V ):

L = T − V. (4.2)

As an application, I will evaluate the Lagrangian for the simple

pendulum of Figure 4.1. Assume the origin of coordinates is at the
point of support. In Cartesian coordinates the potential energy is V =
mgy. The kinetic energy is T = 12 m(ẋ2 + ẏ 2 ). Therefore the Lagrangian
for the simple pendulum is

1
L = T − V = m(ẋ2 + ẏ 2 ) − mgy. (4.3)
2

There are two coordinates in this expression for L, namely x and y. As

you will see shortly, this leads to two equations of motion. Normally it
is desirable to have as few equations of motion as possible. Therefore,
it is desirable to use a coordinate system (such as r, θ) that will min-
imize the number of equations of motion. Physicists frequently need
to transform from one set of coordinates to another. This is done by
using transformation equations. Consider transforming from the Carte-
sian coordinates, x, y, z, to a different set of coordinates, say q1 , q2 , q3 .
The transformation equations are a set of equations having the general
form

x = x(q1 , q2 , q3 , t)
y = y(q1 , q2 , q3 , t)
z = z(q1 , q2 , q3 , t).
 4.2. THE LAGRANGIAN 113

The q’s are called “generalized coordinates.” In practice they will often
be Cartesian, cylindrical, or spherical coordinates, but they are not
limited to these familiar coordinates.2
In the example of the pendulum, the transformation equations from
x and y to θ are
x = l sin θ,
y = −l cos θ,
where the coordinate origin is at the point of suspension and θ is mea-
sured from the negative y-axis (this is not the way we usually define
θ!)
Differentiating these equations with respect to time,
ẋ = lθ̇ cos θ,
ẏ = lθ̇ sin θ.
Plugging these expressions into the Lagrangian for the pendulum (Equa-
tion 4.3) leads to
1 1
L = m(lθ̇ cos θ)2 + m(lθ̇ sin θ)2 + mgl cos θ
2 2
1 2 2
= ml θ̇ (cos2 θ + sin2 θ) + mgl cos θ
2
1 2 2
= ml θ̇ + mgl cos θ. (4.4)
2
Equation (4.4) gives the same information as Equation (4.3) that was
previously obtained for the Lagrangian in terms of x and y, except that
now L depends only on θ.
All I have done so far is to write the Lagrangian in Cartesian co-
ordinates and then transform to polar coordinates. In doing so, the
Lagrangian went from being a function of two variables to being a
function of one variable. This procedure illustrates a very important
point: the Lagrangian should be expressed in terms of the least possible
number of coordinates.
Let me give you a simple “cookbook” procedure for obtaining the
Lagrangian in terms of a set of generalized coordinates q1 , q2 , q3 .
1. If at all possible, write the Lagrangian L = T −V in terms
of Cartesian coordinates. In Cartesian coordinates the translational
2The generalized coordinates are normally a “minimal set” of coordinates. For
example, in Cartesian coordinates the simple pendulum requires two coordinates
(x and y), but in polar coordinates only one coordinate (θ) is required. So θ is the
appropriate generalized coordinate for the pendulum problem.
114 4. THE LAGRANGIAN METHOD

kinetic energy is
1
T = m ẋ2 + ẏ 2 + ż 2 .


2
2. Write the transformation equations:
x = x(q1 , q2 , q3 , t),
y = y(q1 , q2 , q3 , t),
z = z(q1 , q2 , q3 , t).
3. Take derivatives to obtain expressions for ẋ, ẏ, ż in terms
of q’s, q̇’s and t.
4. Write L = T − V in terms of the q’s, q̇’s and t.
You might wonder why I am emphasizing that you should first write
the Lagrangian in Cartesian coordinates. The answer is simple. The
translational kinetic energy in Cartesian coordinates always has the
simple form T = 21 m (ẋ2 + ẏ 2 + ż 2 ) , whereas it can be very compli-
cated in other coordinate systems. Furthermore, the potential energy
is often (but not always) easier to express in Cartesian coordinates.
A very simple (but important) system is a block of mass m on a
frictionless surface and connected to a spring of constant k. See Figure
4.3. It is easy to appreciate that the Lagrangian for this system is
1 1
L = mẋ2 − kx2 . (4.5)
2 2

k
m

Figure 4.3. A mass m on a frictionless surface acted

upon by a spring of constant k.

Exercise 4.2. A rock of mass m is dropped from a height z0 . Write

the Lagrangian. Answer: L = 12 mż 2 − mgz.

Worked Example 4.1. Determine the Lagrangian of the dou-

ble planar pendulum.
 4.2. THE LAGRANGIAN 115

Solution: See Figure 4.2. In Cartesian coordinates the kinetic

and potential energies are:
1 1
T = m1 (ẋ21 + ẏ12 ) + m2 (ẋ22 + ẏ22 ),
2 2
V = m1 gy1 + m2 gy2 ,
where (x1 , y1 ) are the coordinates of the bob of mass m1 and (x2 , y2 )
are the coordinates of the other bob. The origin of coordinates is
at the point of suspension. The transformation equations from the
Cartesian coordinates to the angles θ1 and θ2 are:
x1 = l1 sin θ1
x2 = l1 sin θ1 + l2 sin θ2
y1 = −l1 cos θ1
y2 = −l1 cos θ1 − l2 cos θ2
The velocity components are obtained by taking derivatives of
position with respect to time:
ẋ1 = l1 θ̇1 cos θ1
ẋ2 = l1 θ̇1 cos θ1 + l2 θ̇2 cos θ2
ẏ1 = l1 θ̇1 sin θ1
ẏ2 = l1 θ̇1 sin θ1 + l2 θ̇2 sin θ2
Therefore,
ẋ21 + ẏ12 = l12 θ̇12 (cos2 θ1 + sin2 θ1 ) = l12 θ̇12 ,
and
 2
ẋ22 + ẏ22 2
= (l1 θ̇1 cos θ1 + l2 θ̇2 cos θ2 ) + l1 θ̇1 sin θ1 + l2 θ̇2 sin θ2

= l12 θ˙12 + l22 θ˙22 + 2l1 l2 cos(θ1 − θ2 ).

Consequently
1 1
L = T − V = (m1 + m2 )l12 θ̇12 + m2 (l22 θ̇22 + 2l1 l2 θ̇1 θ̇2 cos{θ1 − θ2 })
2 2
+m1 gl1 cos θ1 + m2 g(l1 cos θ1 + l2 cos θ2 ). (4.6)
I hope you appreciate that obtaining the Lagrangian is a straight-
forward procedure although it may be somewhat tedious. This ex-
ample also illustrates the value of using Cartesian coordinates as
your starting point.
116 4. THE LAGRANGIAN METHOD

y
x
s

Figure 4.4. A disk of mass m and radius R rolls down

a perfectly rough plane.

Worked Example 4.2. Determine the Lagrangian for a disk

of mass m and radius R rolling down an inclined plane of angle
α. See Figure 4.4. The plane is perfectly rough so the disk rolls
without slipping.
Solution: Since the disk is rolling, its total kinetic energy is
rotational kinetic energy plus translational kinetic energy. That is,
T = Trot + Ttrans . The rotational kinetic energy is not, of course,
expressed in Cartesian coordinates; it is Trot = 12 I θ̇2 . (The moment
of inertia is I and the angle through which the disk has rotated is
θ.) The moment of inertia of a disk is 12 mR2 . Therefore,
 
1 1 1
Trot = mR θ̇2 = mR2 θ̇2 .
2
2 2 4
As always, Ttrans = 12 m(ẋ2 + ẏ 2 ). Let s be the distance measured
down the plane. It is convenient to use s as the generalized coor-
dinate. The transformation equations are
x = s cos α,
y = −s sin α.
The translational velocity components are
ẋ = ṡ cos α,
ẏ = −ṡ sin α.
Consequently the translational kinetic energy is
1 1
Ttrans = m(ẋ2 + ẏ 2 ) = mṡ2 .
2 2
 4.3. LAGRANGE’S EQUATIONS 117

I expressed the translational kinetic energy in terms of ṡ and the

rotational kinetic energy in terms of θ̇. But these two quantities are
related. As the disk rolls down the plane, the relationship between
the angle θ through which it has rotated, and the distance s it has
moved down the plane is
s = Rθ.
(A relationship between coordinates, such as this one, is called
a “constraint.” A constraint allows us to express one coordinate
in terms of the others and to describe the problem with one less
coordinate.) Therefore, ṡ = Rθ̇. Consequently, Trot = (1/4)mṡ2 ,
and the total kinetic energy can be written in terms of the single
parameter ṡ :
1 1 3
T = Ttrans + Trot = mṡ2 + mṡ2 = mṡ2 .
2 4 4
The potential energy is
V = mgy = −mgs sin α.
The Lagrangian is defined as L = T − V, so,
3
L = mṡ2 + mgs sin α. (4.7)
4

Exercise 4.3. A pendulum is made of a bob of mass m but the

string is replaced by a spring of constant k. Write the Lagrangian in
terms of the length of the spring and the angle it makes with the
vertical. Let the unstretched length of the spring be l0 . Answer:
1 1 1
L = ml˙2 + ml2 θ̇2 + mgl cos θ − k(l − l0 )2 .
2 2 2

4.3. Lagrange’s Equations

As you have seen, the Lagrangian for a system can be obtained
fairly easily by using the definition L = T − V. The algebra may get a
bit complicated at times, but basically all you need to do is express the
translational kinetic energy and the potential energy in terms of Carte-
sian coordinates (if possible) and then transform to the appropriate set
of “generalized” coordinates.
By this time you are probably wondering what the Lagrangian is
used for. What good is it? The simplest answer is that the Lagrangian
118 4. THE LAGRANGIAN METHOD

allows us to determine the equations of motion for a mechanical system.

In fact, the Lagrangian technique is usually the easiest way to find the
equations of motion for a complicated system. (As I mentioned earlier,
the Lagrangian method is much more that just a useful technique, but
for the present let’s take a purely utilitarian approach and consider the
Lagrangian as a tool for generating the equations of motion.)
Suppose the Lagrangian is expressed as a function of a single gen-
eralized coordinate q, the corresponding generalized velocity q̇, and the
time t. That is, L = L(q, q̇, t). Then the equation of motion is given by:
 
d ∂L ∂L
− = 0. (4.8)
dt ∂ q̇ ∂q
This is called Lagrange’s equation. You should memorize it. Note
particularly that the first partial derivative is taken with respect to the
generalized velocity (q̇) and the second partial derivative is taken with
respect to the generalized position (q).
If the Lagrangian is a function of n coordinates, q1 , q2 , ..., qn , there
are n equations of motion, each having the form of Equation (4.8).
That is, the equations of motion are
 
d ∂L ∂L
− = 0 ; i = 1, 2, ..., n. (4.9)
dt ∂ q̇i ∂qi
Consider, for example, the mass on the spring of Figure 4.3. As
noted in Equation 4.5, the Lagrangian is L = 12 mẋ2 − 12 kx2 . For this
system the generalized coordinate q is just the Cartesian coordinate x.
There are no other coordinates. Therefore, there is only one Lagrange
equation, namely,  
d ∂L ∂L
− = 0.
dt ∂ ẋ ∂x
But
∂L
= mẋ,
∂ ẋ
and
∂L
= −kx.
∂x
Consequently, Lagrange’s equation is
d
(mẋ) + kx = 0,
dt
or
mẍ + kx = 0.
Of course, you get the same result immediately by applying Newton’s
second law.
 4.3. LAGRANGE’S EQUATIONS 119

To illustrate the process for a slightly more complicated system,

the following worked example asks for the equation of motion for a
simple pendulum and the next one asks for the equations of motion for
a double pendulum.

Worked Example 4.3. Determine the equation of motion for

a simple pendulum.
Solution: The Lagrangian for a pendulum is L = 21 ml2 θ̇2 +
mgl cos θ. (See Equation 4.4.) In this expression, the single coor-
dinate is θ and Lagrange’s equation (Eq. 4.8) becomes
 
d ∂L ∂L
− = 0.
dt ∂ θ̇ ∂θ
But the partial of L with respect to the angular velocity θ̇ is
 
∂L ∂ 1 2 2
= ml θ̇ + mgl cos θ = ml2 θ̇,
∂ θ̇ ∂ θ̇ 2
and the partial of L with respect to the angular position θ is
 
∂L ∂ 1 2 2
= ml θ̇ + mgl cos θ = −mgl sin θ.
∂θ ∂θ 2
Therefore, Lagrange’s equation is
d  2 2
ml θ̇ + mgl sin θ = 0,
dt
or,
g
θ̈ = − sin θ.
l
This is, of course, exactly the same equation of motion obtained
earlier by elementary methods (Equation 4.1).

Worked Example 4.4. Determine the equations of motion

for the double planar pendulum illustrated in Figure 4.2.
Solution: Recall that the double pendulum was too compli-
cated to analyze using Newton’s second law. The Lagrangian for
the double pendulum is given by Equation (4.6) which is repeated
120 4. THE LAGRANGIAN METHOD

here:
1 1
L = (m1 + m2 )l12 θ̇12 + m2 [l22 θ̇22 + 2l1 l2 θ̇1 θ̇2 cos(θ1 − θ2 )]
2 2
+m1 gl1 cos θ1 + m2 g(l1 cos θ1 + l2 cos θ2 ).
This problem involves two coordinates, θ1 and θ2 , and therefore
there are two Lagrange equations, namely:
 
d ∂L ∂L
− = 0, (4.10)
dt ∂ θ̇1 ∂θ1
 
d ∂L ∂L
− = 0.
dt ∂ θ̇2 ∂θ2
To facilitate writing them, first evaluate the derivatives:
∂L
= −m2 l1 l2 θ̇1 θ̇2 sin(θ1 − θ2 ) − m1 gl1 sin θ1 − m2 gl1 sin θ1 ,
∂θ1
∂L
= m2 l1 l2 θ̇1 θ̇2 sin(θ1 − θ2 ) − m2 gl2 sin θ2 ,
∂θ2
∂L
= (m1 + m2 )l12 θ̇1 + m2 l1 l2 θ̇2 cos(θ1 − θ2 ),
∂ θ̇1
∂L
= m2 l22 θ̇2 + m2 l1 l2 θ̇1 cos(θ1 − θ2 ).
∂ θ̇2
Plugging into Equations (4.10) yields two coupled equations of mo-
tion:
d h 2 2
i
m1 l1 θ̇1 + m2 l1 θ̇1 + m2 l1 l2 θ̇2 cos(θ1 − θ2 )
dt
+m2 l1 l2 θ̇1 θ̇2 sin(θ1 − θ2 ) + (m1 + m2 )gl1 sin θ1 = 0,
and
d  2

m2 l2 θ̇2 + m2 l1 l2 θ̇1 cos(θ1 − θ2 )
dt
−m2 l1 l2 θ̇1 θ̇2 sin(θ1 − θ2 ) + m2 gl2 sin θ2 = 0.
These equations can be further simplified and written in the
form
h i
2 2
(m1 + m2 )l1 θ̈1 + m2 l1 l2 θ̈2 cos(θ1 − θ2 ) + θ̇2 sin(θ1 − θ2 )
+(m1 + m2 )gl1 sin θ1 = 0,
and
h i
m2 l22 θ̈2 +m2 l1 l2 θ̈1 cos(θ1 − θ2 ) − θ̇12 sin(θ1 − θ2 ) +m2 gl2 sin θ2 = 0.
 4.4. DEGREES OF FREEDOM 121

(I’m sure you now appreciate why I did not attempt to solve
the double pendulum problem by using F = ma!)

Exercise 4.4. Find the equation of motion for a disk rolling down
a perfectly rough inclined plane of angle α. (The Lagrangian was ob-
tained in Worked Example 4.2.) Answer: s̈ = 32 g sin α.
Exercise 4.5. Find the equation of motion for a sphere rolling
down a perfectly rough inclined plane of angle α. (Use the Lagrangian
technique.) The moment of inertia of a sphere is 52 mR2 . Answer: s̈ =
5
7
g sin α.
Exercise 4.6. Using the Lagrangian technique, determine the equa-
tion of motion for a body of mass m falling in a constant gravitational
field.
Exercise 4.7. Find the Lagrangian and the equation of motion for
two astronomical bodies on a collision course that are attracting one
another according to Newton’s Law of Universal Gravitation. (Place
the origin of coordinates at the point where the bodies will collide.)
m1 m2
Answer: m1 r¨1 = −G (1+m 2 (where r1 is the distance from the
1 /m2 )r1
center of mass to body m1 ).

4.4. Degrees of Freedom

The Lagrangian technique often involves changing from one set of
coordinates (usually x, y, z) to another (say, r, θ, φ). (It turns out that
coordinate transformations are central to the Lagrangian method; they
are considered in excruciating detail in graduate courses in mechanics.
If you want to know more about this subject, I recommend the excel-
lent, but advanced, books by Goldstein3 and by Fetter and Walecka.4)
3Goldstein, H. Classical Mechanics, Addison-Wesley Publishing Co., Reading,
MA: 1950. Note that there are three editions. The first edition is the shortest
and probably the best. The second edition included a significant amount of new
material but unfortunately it had many typographical errors which made it rather
difficult to read. The third edition (by H. Goldstein, C. Poole, and J. Safko) has
corrected the errors from the second edition and contains some interesting new
material.
4Fetter, A. L., and Walecka, J. D., Theoretical Mechanics of Particles and
Continua, McGraw-Hill, NY, 1980.
122 4. THE LAGRANGIAN METHOD

In discussing coordinates it is important to realize that every phys-

ical system has a particular number of degrees of freedom. The number
of degrees of freedom is the number of independent coordinates needed
to completely specify the position of every part of the system. To de-
scribe the position of a free particle you must specify the values of three
coordinates (say, x, y, and z). Thus a free particle has three degrees
of freedom. For a system of two free particles you need to specify the
positions of both particles. Each particle has three degrees of freedom,
so the system as a whole has six degrees of freedom. In general a me-
chanical system consisting of N free particles will have 3N degrees of
freedom.
For many systems, however, the number of independent coordinates
is much less than 3N. For example, the position of a particle on a flat
surface such as a tabletop can be described with two coordinates, such
as x and y. A condition that specifies the value of a coordinate or gives
a relation between coordinates is called a constraint. For the particle
on the tabletop, the constraint is
z = constant.
Similarly, a pendulum bob is constrained by the string to remain a
constant distance from the point of support. That is, the bob is con-
strained to the surface of a sphere whose radius equals the length of
the string. Similarly, a bar of soap slipping in a sink is constrained to
the surface of the sink. If the sink is a hemispherical bowl of radius a,
the equation of constraint is
x 2 + y 2 + z 2 = a2 .
This relationship indicates that there are only two independent coordi-
nates because the third coordinate (say z) is related to x and y through
p
z = a2 − x 2 − y 2 .
Constraints are important to us because each constraint reduces
the number of degrees of freedom by one.
Going back to the Lagrangian method, recall that there is one La-
grange equation for each coordinate. (See Equation 4.9.) It is obviously
beneficial to reduce the number of equations. You can do this by us-
ing constraints to get rid of as many nonindependent coordinates as
possible. When you describe the system in terms of independent coor-
dinates (say q1 , q2 , ..., qn ), you have minimized the number of Lagrange
equations. These independent coordinates are referred to as generalized
coordinates and are usually denoted by qi . A system with n degrees
of freedom can be described in terms of the n generalized coordinates
 4.5. GENERALIZED MOMENTUM 123

q1 , · · ·, qn . For such a system, the Lagrangian will be a function of

the generalized coordinates, the generalized velocities, and possibly the
time. That is,
L = L(q1 , q2 , · · ·qn ; q̇1 , q̇2 , · · ·q̇n ; t).
In this case, there will be n Lagrange equations of motion.5

4.5. Generalized Momentum

Consider again the problem of a mass m connected to a spring of
constant k as illustrated in Figure 4.3. If the mass is moving in the x
direction with speed ẋ it has momentum px = mẋ. We have seen that
the Lagrangian for this system is
L = 21 mẋ2 − 12 kx2 .
Taking the derivative of the Lagrangian with respect to ẋ yields
∂L ∂ 1 2 1 2


= mẋ − kx = mẋ.
∂ ẋ ∂ ẋ 2 2

But mẋ is just the linear momentum! Therefore, for this system, the
linear momentum is related to the Lagrangian by
∂L
px = .
∂ ẋ
Next, consider the problem of a simple pendulum consisting of a
mass m hanging from a string of length l. According to Equation (4.4),
the Lagrangian for this system is
L = 12 ml2 θ̇2 + mgl cos θ.
Taking the derivative of the Lagrangian with respect to θ̇ yields
∂L
= ml2 θ̇.
∂ θ̇
2
But ml θ̇ is the angular momentum of the pendulum! In this case the
derivative of the Lagrangian with respect to the angular velocity θ̇ is
the angular momentum.
In the first case, the Lagrangian was expressed in terms of x and
ẋ. That is, L = L(x, ẋ). The generalized coordinate was x and the
generalized velocity was ẋ. In the second case, the Lagrangian was a
function of θ and θ̇, that is, L = L(θ, θ̇). Here the generalized coordinate
was θ and the generalized velocity was θ̇. These two cases illustrate the
5I must warn you that physicists are somewhat careless in their usage of the
term “generalized coordinate.” You will often hear it applied in situations in which
the qi are not all independent.
124 4. THE LAGRANGIAN METHOD

general functional form of the Lagrangian for a one variable system,

L = L(q, q̇, t). That is, the Lagrangian depends on the generalized
position (q), the generalized velocity (q̇), and possibly the time (t).
The linear momentum of a mass on a spring was given by ∂L/∂ ẋ
and the angular momentum for the pendulum was given by ∂L/∂ θ̇. It
is reasonable to define a quantity called the “generalized momentum”
by
∂L
pi = . (4.11)
∂ q̇i
The generalized momentum pi is associated with the generalized co-
ordinate qi . These two quantities are called conjugates. Thus, for the
mass on a spring, the generalized momentum conjugate to x is mẋ and
for the pendulum, the generalized momentum conjugate to θ is ml2 θ̇. In
the first case the generalized momentum is the linear momentum and in
the second case the generalized momentum is the angular momentum.
Although the quantity ∂L/∂ q̇ is called the generalized momentum,
and is denoted by the symbol pi , be aware that it is frequently not
anything you would normally think of as a momentum. For example,
when we discussed a disk rolling down an inclined plane (see Worked
.2
Example 4.2) we found the Lagrangian to be L = 34 ms + mgs sin α.
Consequently the generalized momentum is
∂L 6
ps = = mṡ.
∂ ṡ 4
which is not what one might expect for linear momentum.
4.5.1. Ignorable Coordinates. We now come to a very interest-
ing point. For a mechanical system the Lagrangian might not depend
explicitly on a particular coordinate, say qi . If qi does not appear ex-
plicitly in the Lagrangian, then qi is called an ignorable coordinate.6
In such a case, it is obvious that ∂L/∂qi = 0. Recall that Lagrange’s
equation states that
 
d ∂L ∂L
− = 0.
dt ∂ q˙i ∂qi
But if ∂L/∂qi = 0, then
 
d ∂L
= 0.
dt ∂ q˙i
6When we say that q
i does not appear explicitly in the Lagrangian we mean that
if the Lagrangian is expressed in terms of the minimal set of generalized coordinates,
the quantity qi is not present. (Nevertheless, the generalized velocity q̇i may appear
in the Lagrangian.)
 4.5. GENERALIZED MOMENTUM 125

The term in parenthesis is the generalized momentum pi , so this equa-

tion states that dpi /dt = 0. If the time derivative of a quantity is
zero, the quantity is a constant. Therefore, if the coordinate qi does
not appear explicitly in the Lagrangian, then the conjugate generalized
momentum pi is constant.
I will summarize all of this in a single phrase which is easy to
remember, but perhaps not so easy to understand:
If the coordinate qi is ignorable, the
conjugate momentum pi is constant.
In studying a physical system, whenever you encounter a quantity
that remains constant during the motion, you have a very useful tool
for solving problems. Such constant quantities are called “constants of
the motion” or “first integrals” for reasons we shall discuss later.

Worked Example 4.5. The Lagrangian for a certain system

is given by L = (1/2)mṙ2 + (1/2)mr2 φ̇2 , where r and φ are gen-
eralized coordinates and m is constant. Determine the generalized
momenta pr and pφ . Is either of these a constant of the motion?
Solution: The generalized coordinates are r and φ, so the
conjugate generalized momenta are
∂L ∂L
pr = and pφ =
∂ ṙ ∂ φ̇
Therefore,
 
∂L ∂ 1 2 1 2 2
pr = = mṙ + mr φ̇ = mṙ
∂ ṙ ∂ ṙ 2 2
and  
∂L ∂ 1 2 1 2 2
pφ = = pφ = mṙ + mr φ̇ = mr2 φ̇
∂ φ̇ ∂ φ̇ 2 2
Since the Lagrangian does not depend on φ (it is ignorable), we
conclude that pφ is constant.

Exercise 4.8. Determine the generalized momenta pθ1 and pθ2 for
the double planar pendulum. Is either one of them a constant? Answer:
pθ2 = m2 l22 θ̇2 + m2 l1 l2 θ̇1 cos(θ1 − θ2 ). No.
Exercise 4.9. (a) Determine the generalized momentum for a sphere
of mass m and radius R rolling down an inclined plane of angle α. Let
126 4. THE LAGRANGIAN METHOD

the distance down the plane be given by s. (b) Is the generalized

momentum a constant? Answer: (a) ps = (5/2)mṡ. (b) No.

4.6. Generalized Force (Optional)

Lagrangian mechanics introduced the concepts of generalized coor-
dinates and generalized momenta. So it should not surprise you that
there is another quantity called the generalized force. The generalized
force is denoted by Qj and is defined in terms of a quantity called the
“virtual work.” However, before we can consider these concepts fur-
ther, we need to introduce yet another new concept, the “virtual dis-
placement” δx. The virtual displacement is an imaginary infinitesimal
displacement that is similar to the real infinitesimal displacement de-
noted dx, except that time is frozen for the virtual displacement. That
is, δx is imaginary, infinitesimal and instantaneous. Furthermore, it
satisfies any constraints on the system.
Now real work is defined in terms of forces and displacements and
we can write
X
dW = F·dr = Fx dx + Fy dy + Fz dz = Fi dxi .
i

Similarly, as you might suspect, the virtual work is given by

X
δW = Fi δxi .
i

Now since xi = xi (q1 , q2 , · · · qn , t) and by the rules of calculus

∂xi ∂xi ∂xi dxi
dxi = dq1 + dq2 + · · · dqn + dt,
∂q1 ∂q2 ∂qn dt
we conclude that the virtual displacement is given by
∂xi ∂xi ∂xi X ∂xi
δxi = δq1 + δq2 + · · · δqn = δqj .
∂q1 ∂q2 ∂qn j
∂q j

Note that the dt has disappeared because time is frozen during a virtual
displacement. Consequently, the virtual work can be expressed as
!
X X ∂xi X X ∂xi
δW = Fi δqj = Fi δqj .
i j
∂qj j i
∂qj
 4.6. GENERALIZED FORCE (OPTIONAL) 127

If we define the generalized force as

X ∂xi
Qj = Fi
i
∂qj
then we can write the virtual work in the form
X
δW = Qj δqj
j

and it has the same form as our usual expression for work, as a force
times a displacement.
Under certain circumstances, the generalized force can be obtained
from a scalar potential function V as
∂V
Qj = − .
∂qj
This relation is analogous to the well known relation between real
forces and the potential energy, F = −∇V.

Worked Example 4.6. Assume that for a certain system, the

generalized force Q can be obtained from the potential V according
to
∂V
Qi = − ,
∂qi
where the potential is assumed to be independent of velocity. Show
that for this situation, Lagrange’s equation can be written in the
so-called “Nielsen form”
d ∂T ∂T
− = Qi . (4.12)
dt ∂ q˙i ∂qi
Solution: According to the statement of the problem, the poten-
tial does not depend on velocity, so
∂L ∂ ∂T
= (T − V ) =
∂ q˙i ∂ q˙i ∂ q˙i
Similarly,
∂L ∂ ∂T ∂V ∂T
= (T − V ) = − = + Qi
∂qi ∂qi ∂qi ∂qi ∂qi
128 4. THE LAGRANGIAN METHOD

Therefore the Lagrange equation can be expressed as follows:

d ∂L ∂L
− = 0
dt ∂ q˙i ∂qi
 
d ∂T ∂T
− + Qi = 0
dt ∂ q˙i ∂qi
d ∂T ∂T
− = Qi .
dt ∂ q˙i ∂qi

4.7. The Calculus of Variations (Optional)

I am now going to discuss a branch of mathematics known as the
Calculus of Variations. I am primarily doing this so that you will un-
derstand how Lagrange’s equations are derived. However, the calculus
of variations is an interesting subject in its own right.
The calculus of variations considers problems involving maximums
and minimums for certain definite integrals. In doing so, it arrives at
answers to various interesting questions, such as: (1) What is the shape
of the curve of given length that encloses the largest area? (A circle.)
(2) What curve gives the shortest distance between two points in a
plane? (A straight line.) (3) What curve gives the shortest distance
between two points on a sphere? (A segment of a great circle.) (4)
What is the shape of a curve between two points, one higher than the
other, such that a bead slides from one point to the other in minimum
time? (A cycloid.)
In such problems you are asked to maximize or minimize some
quantity. The quantity itself will be expressed as an integral. You
may have to exert considerable cleverness to figure out this integral, as
it is usually not obvious.

y 2
ds
dy
dx
1
x
Figure 4.5. The quantity ds is an element of the path
from 1 to 2. Note that ds2 = dx2 + dy 2 .
 4.7. THE CALCULUS OF VARIATIONS (OPTIONAL) 129

Consider a specific example. Suppose you want to determine the

equation of the curve between two points in a plane such that the
distance along the curve is a minimum. (You already know the answer:
It is a straight line, and its equation is y R= mx + b.) Now the distance
2
between two points along a curve s is 1 ds. But ds2 = dx2 + dy 2
(Figure 4.5) so
s 2
Z 2 Z 2 Z x2 
p dy
I= ds = dx2 + dy 2 = 1+ dx
1 1 x1 dx (4.13)
Z x2 q
= 1 + (y 0 )2 dx.
x1
0
Here y ≡ dy/dx. (In the rest of this chapter, a prime will represent
differentiation with respect to x.) The integral I is the distance be-
tween points 1 and 2. The problem is to find the curve y = y(x) that
minimizes I.
I will come back to this problem in a moment, but first I want to
consider the calculus of variations in a general way. Problems in the
calculus of variations always lead to definite integrals of the form
Z x2 Z x2
dy
I= Φ(x, y, )dx = Φ(x, y, y 0 )dx.
x1 dx x1

The solution of the problem is the function y = y(x) that makes the
integral an extremum.7 Note that Φ is a function of a function be-
cause Φ is a function of y, and y itself is a function of x. Φ is called a
“functional” to distinguish it from an ordinary function.
Keep in mind that the integral I is a line integral between limits
x1 and x2 . It is the distance from point 1 to point 2 along a particular
path y = y(x). Obviously, if you use a different function y1 = y1 (x),
you will integrate along a different path and get a different value for
the integral.
Figure 4.6 illustrates the problem of the shortest distance between
two points in a plane. The three paths y1 (x), y2 (x), and y3 (x) have
different path lengths. The problem asks us to find the shortest of all
the possible paths between the fixed end points. Let us assume that
y0 (x) is the shortest path, and let us further assume that y1 (x) is a
path that only differs infinitesimally from y0 (x). That is, at each point
7By “extremum” we mean a maximum or a minimum, that is, values of a
function where the derivative is zero. The derivative of a function is also zero at an
inflection point, but in the calculus of variations we are always interested in either
the maximum or the minimum of the integral of a functional.
130 4. THE LAGRANGIAN METHOD

y y1(x)
y (x)
2

y3(x)

x x x
1 2

Figure 4.6. Three possible paths between two fixed end points.

on the path (except at the end points) y1 (x) is slightly different from
y0 (x). We can write the relationship between y1 (x) and y0 (x) in the
form
y1 (x) = y0 (x) + 1 η(x)
where 1 is a small quantity and η(x) is a function of x. Note that η(x)
is completely arbitrary except that it must be zero at the endpoints.
That is,
η(x1 ) = η(x2 ) = 0,
because all the paths between x1 and x2 must meet at the endpoints.
Remember that the problem is to find the minimal path y0 (x). I
will denote this minimal path simply by y(x) from now on. In general,
a nearby path is a member of a family of paths that are described by
the relationship
Y = Y (x, ) = y(x) + η (x) , (4.14)
where y(x) is the minimal path. Here  is taken as a continuous vari-
able. Different paths correspond to different values of . The minimal
path is obtained when  = 0.
The path length along any particular member of this family of
curves is
Z x2   Z x2
dY
I = I() = Φ x, Y, dx = Φ (x, Y, Y 0 ) dx.
x1 dx x1
Here the integral I is explicitly expressed as a function of . The desired
curve has  = 0; it is the curve that minimizes I. Therefore, the
condition on I is  
dI
= 0.
d =0
Observe that
d x2
Z x2 
∂Φ ∂Y 0
Z 
dI 0 ∂Φ ∂Y
= Φ(x, Y, Y )dx = + dx.
d d x1 x1 ∂Y ∂ ∂Y 0 ∂
 4.7. THE CALCULUS OF VARIATIONS (OPTIONAL) 131

From Equation (4.14)

∂Y
= η(x).
∂
Also,
∂Y 0
 
∂ dY ∂ 0
= = (y + η 0 ) = η 0 .
∂ ∂ dx ∂
If  = 0 then Y = y and Y 0 = y 0 , so
  Z x2  
dI ∂Φ ∂Φ 0
= η + 0 η dx = 0. (4.15)
d =0 x1 ∂y ∂y
The formula for integration by parts is
Z b Z b
b
udv = uv|a − vdu.
a a
Applying this to the second term in Equation (4.15) leads to
Z x2  x2 Z x2
∂Φ dη ∂Φ d ∂Φ
0
dx = 0
η − η dx.
x1 ∂y dx ∂y x1 x1 dx ∂y 0
But η = 0 at the limits x1 and x2 so the first term drops out, and
equation (4.15) becomes
Z x2  
∂Φ d ∂Φ
0= − ηdx.
x1 ∂y dx ∂y 0
Since η 6= 0 (except at the end points) this integral can be zero if and
only if the term in parenthesis is zero.8 That is,
∂Φ d ∂Φ
− = 0. (4.16)
∂y dx ∂y 0
This is called the Euler-Lagrange equation. It gives the condition on Φ
that minimizes the integral I.
To appreciate how to use the Euler-Lagrange equation, let us return
to the problem of the shortest distance between two points. We found
that for this problem the functional Φ is (Equation 4.13)
s  2
dy
1
Φ= 1+ = 1 + y 02 2 .
dx
8You might argue that
Z b
g(x)η(x)dx = 0
a
can be satisfied even if g(x) is not zero everywhere for some function η(x). But
that argument does not hold for an arbitrary function η(x).
132 4. THE LAGRANGIAN METHOD

Plugging this functional into the Euler-Lagrange equation (4.16) gives

∂
1 d ∂
1
1 + y 02 2 − 1 + y 02 2
= 0,
∂y dx ∂y 0
 
d 1
1
02 − 2 0
0− 1+y (2y ) = 0,
dx 2
" #
d y0
p = 0.
dx 1 + y 02
If the derivative of a function is zero, the function is constant, so

y0
p = const = C1 .
1 + y 02
Squaring both sides and rearranging,

y 02 = (C1 )2 (1 + y 02 ). (4.17)
02
Solving for y ,
y 02 (1 − C12 ) = C12
s
C12
y0 = = constant = m.
1 − C12
But y 0 = dy/dx, so
dy
= m.
dx
A final integration yields

y = mx + b,
the equation for a straight line! Thus we have proved that a straight
line is the shortest distance between two points. (Whew!)
4.7.1. Hamilton’s Principle. By this time you are surely won-
dering what any of this has to do with physics. The answer lies in
Hamilton’s principle. Hamilton’s principle states that the behavior
of any physical system will minimize the time integral of the
Lagrangian. In other words, a physical system that evolves over time
from t1 to t2 will follow a “path” q = q(t) such that the integral
Z t2
I= L(q, q̇, t)dt
t1

is minimized. The condition for this integral to be minimized is given by

the Euler-Lagrange equation of the calculus of variations. By replacing
 4.7. THE CALCULUS OF VARIATIONS (OPTIONAL) 133

q’s for y’s and t for x and changing the order of the terms, the Euler-
Lagrange equation (Equation 4.16) becomes
d ∂L ∂L
− = 0,
dt ∂ q̇ ∂q
which you recognize R as Lagrange’s equation!
The integral Ldt is called the “action.” Using that terminology,
Hamilton’s principle is:
The time development of a dynamical system will minimize the action.
Hamilton’s principle is sometimes referred to as, “The fundamental
principle of mechanics.”9
To make this a bit more explicit, consider a simple physical process,
such as the motion of a falling rock. This physical system can be
considered to evolve from some initial situation at time t1 to a different
final situation at time t2 . Initially the rock is at height h and has zero
velocity. Therefore the initial conditions are (q, q̇) = (h, 0). The√final
conditions (just before it hits the ground) are q = 0 and q̇ = 2gh.
Hamilton’s principle states that the behavior of the rock (its position
and
R t2 velocity at any instant of time) is such as to minimize the quantity
t1
L(q, q̇, t)dt. This leads to the variant of the Euler-Lagrange equation
called the Lagrange equation. Finally, the Lagrange equation generates
the equation of motion, which for the falling rock, is simply d2 q/dt2 =
−g.
4.7.2. Relation to Newton’s Second Law. As mentioned above,
Newton’s second law can be derived from the Lagrange equation. As
an illustration, consider the one-dimensional motion of a particle of
mass m. Assume the particle is acted upon by a conservative force F .
Recall that in one dimension, if F is conservative it can be obtained
from the potential energy V = V (x) by F = −dV /dx. The Lagrangian
is
1
L = T − V = mẋ2 − V (x),
2
and Lagrange’s equation is
    
d ∂L ∂L d ∂ 1 2 ∂ 1 2
0 = − = mẋ − V (x) − mẋ − V (x)
dt ∂ ẋ ∂x dt ∂ ẋ 2 ∂x 2
d ∂V (x)
= (mẋ) + = mẍ − F.
dt ∂x
9You may have studied Fermat’s principle in optics which states that the path
of a ray of light from one point to another is such as to minimize the time of flight.
This is a special case of Hamilton’s principle.
134 4. THE LAGRANGIAN METHOD

So
F = mẍ
as expected.
Thus, Newton’s second law is a consequence of Lagrange’s equations
and Lagrange’s equations are a consequence of Hamilton’s principle.10
I suppose you might conclude that Hamilton’s principle is more basic
and important than Newton’s laws. Nevertheless, it is true that New-
ton’s laws are the basic relations from which all of physics was derived.
Hamilton’s principle is a beautiful, sophisticated, and elegant way of
describing the way nature behaves, but it is not very easily understood.
Newton’s laws, on the other hand, can be grasped by anyone with an
elementary knowledge of mathematics.

4.8. Hamilton’s Equations (Optional)

In this section I will briefly discuss another formulation of the laws
of mechanics. First, however, I have to define a quantity called the
Hamiltonian. The Hamiltonian is a function of position and momen-
tum. For a single particle the one-dimensional Hamiltonian is defined
as:
H(p, q, t) = pq̇ − L(q, q̇, t), (4.18)
where p stands for the generalized momentum and q for the generalized
coordinate. Note that the Lagrangian is a function of velocity, position,
and time, L = L(q, q̇, t). The Hamiltonian is a function of momentum,
position and time, H = H(p, q, t). If you are asked to determine a
Hamiltonian, make sure your final expression does not explicitly con-
tain any velocities!
For more than one dimension and/or more than one particle the
Hamiltonian is: X
H= pi q̇i − L. (4.19)
i
More explicitly, for a system of N particles in three dimensions,

3N
X
H(p1 , p2 , · · · , p3N ; q1 , q2 , · · · q3N ; t) = pi q̇i −L(q1 , · · · q3N ; q̇1 , · · · q̇3N ; t).
i=1

10You might ask: “Why does nature always minimize the time integral of the
Lagrangian?” I do not know the answer to this question. I know that Hamilton’s
principle is an expression of the way the physical universe is put together. I know
that is how nature behaves, but I cannot tell you why it behaves that way. (Physics
tells you how, not why.)
 4.8. HAMILTON’S EQUATIONS (OPTIONAL) 135

Having obtained the Hamiltonian you can use it to obtain the equa-
tions of motion of the system. Recall that the equation of motion as
expressed by Newton’s second law as well as by Lagrange’s equation
is a second-order differential equation. On the other hand the Hamil-
tonian formulation yields two first-order equations for the momentum
and position. They are:
∂H ∂H
ṗi = − and q̇i = + . (4.20)
∂qi ∂pi
These are Hamilton’s equations of motion and they are completely
equivalent to Newton’s second law and to the Lagrange equations of
motion.
There are a number of different ways of deriving Hamilton’s equa-
tions. One way is to start with Equation (4.19) and write the differen-
tial of H as
X ∂L ∂L

∂L
dH = pi dq̇i + q̇i dpi − dqi − dq̇i − dt.
i
∂qi ∂ q̇i ∂t
Using the definition of generalized momentum to cancel the first and
last term in parenthesis, and using Lagrange’s equation to write
d ∂L d ∂L
= pi = ṗi =
dt ∂ q̇i dt ∂qi
we obtain
X ∂L
dH = (q̇i dpi − ṗi dqi ) − dt.
i
∂t
But the differential of H as obtained from H = H(pi , qi , t) is
X  ∂H ∂H

∂H
dH = dpi + dqi + dt.
i
∂pi ∂qi ∂t
A term-by-term comparison of the two expressions for dH yields Equa-
tions (4.20) as well as
∂H ∂L
=− .
∂t ∂t
For the sake of a specific example, consider the Hamiltonian of a
particle of mass m moving vertically in a uniform gravitational field
g. (For example, you might want to determine the Hamiltonian and
Hamilton’s equations for a ball of mass m thrown vertically upward.)
Begin with the Lagrangian. It is
1
L = T − V = mż 2 − mgz.
2
136 4. THE LAGRANGIAN METHOD

By the definition of generalized momentum,

∂L
p= = mż.
∂ ż
Therefore,
p
ż = .
m
Equation (4.18) then gives
p2
H = pż − L = pż − + mgz
2m
p p2
= p − + mgz
m 2m
p2
= + + mgz.
2m
Note that I was careful to replace ż by p/m. Inserting this Hamiltonian
into Hamilton’s equations yields the following equations of motion:

∂ p2
 
∂H
ṗ = − =− + mgz = −mg
∂z ∂z 2m
∂ p2
 
∂H p
q̇ = ż = + = + mgz = .
∂p ∂p 2m m
The second equation reads ż = p/m or p = mż which is just the
definition of momentum. The first equation states that ṗ = −mg or
d d
p = mż = mz̈ = −mg.
dt dt
That is, z̈ = −g, as expected.
Now
p2 m2 v 2 1
= = mv 2 = T,
2m 2m 2
and
V = mgz,
so
H = T + V = E = total energy.
Therefore, in this case, H = E. In fact, the Hamiltonian is nearly
always the total energy expressed in terms of generalized momentum
and position. Students frequently assume that H is always equal to
E. This is not true. The following conditions determine when H is
constant and when it is equal to E.
 4.8. HAMILTON’S EQUATIONS (OPTIONAL) 137

(1) If the Lagrangian does not depend explicitly on time, the

Hamiltonian is constant but not necessarily equal to the en-
ergy.
(2) If the transformation equations do not depend on time and
the potential energy does not depend on velocity, then the
Hamiltonian is equal to the total energy but it may not be
constant.
(3) If the constraints and the transformation equations and the
potential energy are all time independent, then the Hamilton-
ian is equal to the total energy and is constant.
The third condition is satisfied by many systems, and frequently
when requested to write the Hamiltonian (especially in quantum me-
chanical problems) a physics student will simply write H = p2 /2m + V.
However, this can be dangerous because there are problems in which
the Hamiltonian is not equal to the total energy. Fortunately, you prob-
ably will not encounter any of these situations in your undergraduate
courses in physics.
Students are often asked to determine the Hamiltonian for a partic-
ular system. A common error is to write an expression for H that in-
volves generalized velocities. This is not the way a Hamiltonian should
be written. The final expression for H must only contain momenta and
positions (and possibly the time) because H = H(p, q, t).
Finally, I would like to mention that if you transform to a new co-
ordinate system, you will have a new set of generalized momenta and
generalized coordinates, say Pi and Qi . If these new coordinates main-
tain the form of Hamilton’s equations unchanged, then they are called
“canonical conjugates.” Although we will not consider this subject any
further, I can assure you that you will be hearing much more about
canonical conjugates in your advanced courses.

Worked Example 4.7. (a) Write the Hamiltonian in polar

coordinates for a planet (mass m) in the gravitational field of a
star (mass M ). (Assume the planet is a particle moving in a planar
orbit. The star may be assumed to remain at rest.)
Solution: The Lagrangian for this system in polar coordinates
(r, θ) can be obtained from the fact that the velocity is
v = ṙr̂ + rθ̇θ̂
so
1 1  
T = mv 2 = m ṙ2 + r2 θ̇2 .
2 2
138 4. THE LAGRANGIAN METHOD

The potential energy is obtained from Newton’s Law of Universal

Gravitation
GM m
V =− .
r
Therefore,
1 1 GM m
L = T − V = mṙ2 + mr2 θ̇2 + .
2 2 r
The generalized momenta are
∂L
pr = = mṙ,
∂ ṙ
∂L
pθ = = mr2 θ̇.
∂ θ̇
The Hamiltonian is, therefore,
H = Σpi q̇i − L
= pr ṙ + pθ θ̇ − L
 
1 2 1 2 2 GM m
= pr ṙ + pθ θ̇ − mṙ + mr θ̇ + .
2 2 r
Replacing ṙ by pr /m and θ̇ by pθ /mr2 we obtain
pr pθ 1  pr 2 1 2  pθ 2 GM m
H = pr + pθ 2 − m − mr −
m mr 2 m 2 mr2 r
p2r p2θ 1 p2r 1 p2θ GM m
= + − − −
m mr2 2 m 2 mr2 r
2 2
1 pr 1 p θ GM m
= + 2
− .
2 m 2 mr r
Exercise 4.10. Write the Hamiltonian for a free particle. Answer:
2
p /2m.
Exercise 4.11. A student writes the Hamiltonian for a particle
falling in a uniform gravitational field as H = 21 mż 2 + mgz. What is
wrong with this?

4.9. Summary
To determine the equation(s) of motion do one of the following:
(1) Draw the free body diagram and apply F = ma. (That is,
determine the equation of motion by inspection.)
(2) Write the Lagrangian and apply Lagrange’s equations.
 4.9. SUMMARY 139

(3) Write the Hamiltonian and apply Hamilton’s equations.

The Lagrangian is given by
L = L(qi , q̇i , t) = T − V
and the Lagrange equations are
 
d ∂L ∂L
− =0; i = 1, 2, ..., n.
dt ∂ q̇i ∂qi
The qi and q̇i are called generalized coordinates and generalized veloc-
ities. The generalized momentum is defined by
∂L
pi = .
∂ q̇i
If the generalized coordinate qi does not appear in the Lagrangian, it is
called an ignorable coordinate. The generalized momentum conjugate
to an ignorable coordinate is a constant.
The number of generalized coordinates is equal to the number of
degrees of freedom. Each constraint reduces by one the number of
degrees of freedom (and hence, the number of generalized coordinates).
Lagrange’s equations are obtained from the calculus of variations.
The calculus of variations allows one to determine the extremum con-
ditions for a given parameter (such as time, distance, area, etc.). To
determine the function y = y(x) that will minimize or maximize the
quantity Z
Φ(x, y, y 0 )dx,
one applies the Euler-Lagrange equation
∂Φ d ∂Φ
− = 0.
∂y dx ∂y 0
The most difficult aspect of such problems is determining the functional
Φ(x, y, y 0 ).
Similarly, Lagrange’s equations are obtained from Hamilton’s prin-
ciple. This principle states that the time evolution of a mechanical
system will minimize the action, defined by
Z
Action ≡ Ldt.

Applying the technique of the calculus of variations to minimize the

action leads to Lagrange’s equations.
The Hamiltonian is defined by
X
H≡ pi q̇i − L.
140 4. THE LAGRANGIAN METHOD

The equations of motion (Hamilton’s equations) are

∂H ∂H
ṗi = − , and q̇i = +
∂qi ∂pi

4.10. Problems
Note: Problems 4.15 through 4.20 are based on optional section 4.7.
Problem 4.1. Determine the Lagrangian for a particle sliding on
the inner surface of a hemisphere of radius a (the “bar of soap in a
sink” problem). See Figure 4.7.

θ
ϕ y

x m

Figure 4.7. A particle of mass m slides on the inside

of a perfectly smooth hemispherical surface.

Problem 4.2. A bead of mass m, acted upon by a uniform gravita-

tional force, can slide freely on a frictionless circular hoop of radius R.
The hoop is oriented vertically and is rotating about a vertical axis with
constant angular velocity ω. Assume the hoop is massless. Determine
the Lagrangian for the bead.
Problem 4.3. A certain molecule is made up of four atoms. Three
of them have mass M and are located at the vertices of an equilateral
triangle of side a. The fourth atom has mass m and is free to move along
a line perpendicular to the triangle and passing through its center. As-
sume this atom is attracted to the other atoms by a force proportional
to the distance between them. (You can imagine the atom of mass m
is attached by three springs of force constant k to the atoms of mass
M.) Determine the Lagrangian for the system.
 4.10. PROBLEMS 141

Problem 4.4. A heavy ring of mass m can slide on a frictionless

rod. The rod is attached to a wall at one end with a bracket and the
other end hangs downward, so the constant angle between the rod and
the wall is α. One end of a massless spring of constant k is attached to
the bracket and the other end is connected to the ring. Determine the
Lagrangian for the system and the equation of motion. See Figure 4.8.
y x
s

m
Figure 4.8. The ring connected to the spring is free to
slide on the inclined rod.

Problem 4.5. A cart of mass M slides freely (no friction) on el-

evated horizontal tracks. Hanging from the cart is a pendulum with
string of length l and bob of mass m. Determine the Lagrangian for
this system. Obtain the equations of motion. By considering whether
or not any coordinates are ignorable, determine any constants of the
motion. See Figure 4.9.

θ l
m

Figure 4.9. A cart sliding freely on raised rails has a

pendulum hanging from it.

Problem 4.6. Obtain the equations of motion for a spherical pen-

dulum, that is, a pendulum that is not constrained to oscillate in a
plane.
Problem 4.7. A wedge of mass M and angle α is resting on a
frictionless horizontal plane. A rectangular box of mass m is on the
142 4. THE LAGRANGIAN METHOD

wedge. Since all surfaces are frictionless, the box will slide down the
wedge and the wedge will slide in the opposite direction on the plane.
Determine the accelerations of the two bodies.
Problem 4.8. A system of two masses and three springs is illus-
trated in Figure 4.8. Write the Lagrangian. Determine the generalized
momenta for this system. (Hint: Measure the positions of the masses
from their equilibrium points.)

k k k
1 2 3
M M
1 2

Problem 4.9. A toy cart is made up of a block of mass M and

four cylindrical wheels of radius R and mass m. It is rolling down a
fixed plane of angle α. Let the cart have a length d and let the wheels
be at positions ±d/2 relative to the center of mass of the cart. Assume
the center of mass lies on a line joining the centers of the wheels. See
Figure 4.10. Determine the Lagrangian for the system.

Figure 4.10. A block with four wheels rolls down a

plane. Problem 4.9.

Problem 4.10. A rope passes over a frictionless pulley. A monkey

of mass M is hanging from one end of the rope and a bunch of bananas
of mass m is attached to the other end. Since M > m, the monkey
is descending and the bananas are ascending and getting further and
further away from the monkey. Naturally, the monkey begins to climb
the rope. See Figure 4.11. You are asked to obtain the motion (x =
x(t)) for the monkey under two different assumptions. (a) Assume the
monkey climbs at a constant velocity relative to the rope. (b) Assume
the monkey exerts a constant force on the rope.
Problem 4.11. A wedge of mass M and angle α is free to slide on
a frictionless horizontal surface. On the wedge is mounted a mass and
spring arrangement that allows the mass m to slide up and down on
 4.10. PROBLEMS 143

Figure 4.11. The monkey and the bananas ( Problem 4.10).

the wedge. All surfaces are frictionless and the spring is massless. (a)
Write the Lagrangian for the system. (b) Obtain expressions for the
generalized momenta. (c) Obtain the equations of motion. See Figure
4.12.

X y
s m
M α
Figure 4.12. A block and spring on a sliding wedge.
The dot in the middle of the spring represents the equi-
librium position of the block, and is a distance X from
some fixed point. The stretch of the spring is measured
from the equilibrium position. The position of the block
is given by x and y where x = X + s sin α.

Problem 4.12. A pendulum is made of a bob of mass M but

instead of a string, the bob is suspended from a massless spring of
constant k. The system can swing back and forth and also move up
144 4. THE LAGRANGIAN METHOD

and down, but it is restricted to motion in a plane. Determine the

equations of motion.
Problem 4.13. Figure 4.13 shows two equal masses m. One is
hanging and the other is on a frictionless surface. They are connected
by an ideal horizontal string that passes over an ideal massless pulley.
The mass on the surface is connected to a fixed point by a massless
spring of constant k. Place the origin of coordinates at the pulley.
(a) Write the Lagrangian for the system in terms of a single variable.
(b) Obtain the equation of motion. (c) Determine the frequency of
oscillation of the system.

k
m

Figure 4.13. Two masses, a frictionless surface, a

spring, and an ideal pulley.

Problem 4.14. A marble of mass M and radius R rolls back and

forth on the bottom of a semi-circular bowl of radius a. (a) Write the
Lagrangian for the system. (b) Obtain the equation of motion. (c)
Determine the period with which the marble rolls back and forth. You
may assume the amplitude of the oscillations is small.
Problem 4.15. It is desired to find the equation for the shortest
distance between two points on a sphere. Determine the functional for
this problem. (Use spherical coordinates.)
Problem 4.16. The speed of light in a medium of index of refrac-
tion n is v = c/n = ds/dt. The time for light to travel from point A to
point B is
Z B
ds
.
A v
Obtain the law of reflection and the law of refraction (Snell’s law) by
using Fermat’s principle of least time.
 4.10. PROBLEMS 145

Problem 4.17. Determine the relation y=y(x) such that the fol-
lowing integral is an extremum. What is the shape of the curve?
Z x2
1/2
(x)(1 + y 02 )

dx.
x1

Problem 4.18. A bead slides down a wire. You want to determine

the shape the wire must have for the bead to slide from its initial
position to its final position in the least amount of time. (This is
called the “brachistochrone” problem.) Note that the quantity to be
minimized is the time and that dt = ds/v where the velocity v is given
by conservation of energy. I do not expect you to actually solve this
problem; it will be sufficient if you write the functional that describes
the system.11
Problem 4.19. Two pegs are at the same height above the ground.
A rope is draped over them, with both ends of the rope resting on the
ground. We wish to know the shape of the rope between the pegs.
The quantity that is minimized in this case is the potential energy.
Determine the appropriate functional.
Problem 4.20. A vase is designed by drawing a curve and then
rotating it about the z-axis, forming a “surface of revolution.” It is
desired that the vase have the minimum possible surface area. Obtain
the functional for this problem. See Figure 4.14 to get an idea of the
geometry.

axis

ds Area of shaded
Curve of
region is
length s
2πxds
Figure 4.14. A curve of length s is revolved about an
axis to generate a surface of revolution.

11This problem was apparently first devised by one the Bernoulli brothers who
sent it out as a “challenge problem” to the scientists of that day. It is said that
Newton solved the problem in a few hours. (The calculus of variations had not
been invented yet!)
146 4. THE LAGRANGIAN METHOD

Problem 4.21. In quantum mechanics, one replaces the momen-

h ∂
tum p by the “momentum operator” i 2π ∂x
. Using this replacement,
write the quantum mechanical Hamiltonian operator for a free particle.
(The quantity h/2π is usually written as ~.)
Problem 4.22. Obtain the Schrödinger equation as follows: Write
the quantum mechanical version of the Hamiltonian for a particle whose
potential energy is V (x). Let H = E, and allow the operators to act
on the function Ψ(x).
Problem 4.23. Write the Hamiltonian for a planar pendulum and
obtain the Hamilton’s equations of motion for the system.
Problem 4.24. Write the Hamiltonian for a mass on a spring (as
in Figure 4.3) and obtain the Hamilton’s equations of motion for the
system.

Computational Projects
Computational Techniques: To solve the first computational prob-
lem below you will have to use a Computer Algebra System (CAS) such
as Mathematica or Maple. We do not have the space nor the time to
get into the details of these excellent programs, but you should learn
some of the basics of one such system. They are easily available. For
example, Matlab incorporates symbolic computation using the Maple
kernel. (Maple is a CAS that was developed primarily at the University
of Waterloo in Canada and more recently at ETH in Zurich - Einstein’s
Alma Mater.) In Matlab’s version of Maple, if you want to obtain, say,
an expression for y = cos2 x + sin2 x you simply declare x, y and z as
symbolic variables, using the command
syms x y z
then write
y = cos(x)ˆ2+sin(x)ˆ2
Then write
simplify(y)
You will obtain (of course) y=1.
Another example, is to write
z = sin(x)/cos(x)
then write
simple(z)
 4.10. PROBLEMS 147

to get
z = tan(x)
Of course, you will have to consult the “help” files to get further infor-
mation, but let me assure you that using Maple is not difficult.
Computational Project 4.1. Use a symbolic manipulator such
as Maple to obtain the Hamiltonian for the double planar pendulum.
Computational Project 4.2. Solve for the motion of the double
pendulum. Assume the two bobs have the same mass and that the two
strings are of length 0.1 m. Compute examples of the motion for several
initial conditions.
Computational Project 4.3. A 2 kg object is launched hori-
zontally from a 5 m cliff with an initial velocity of 20 m/s. As you
know, it will follow a parabolic path. It hits the ground a distance 20.2
m horizontally from the launch point and will have been in the air for
1.01 s. The purpose of this project is to evaluate
R the action for this
process. The action is defined by the integral Ldt where L = T − V.
After you have written a computer program to evaluate the action for
the actual path of the projectile, evaluate the action for the following
hypothetical path between the same two end points and taking the
same amount of time. The hypothetical path is a straight line from
(0,5) to (20.2,0) at a constant speed of 20.6 m/s. You will find that the
action for the hypothetical path is greater than the action for the actual
path since the actual path for any dynamical system always minimizes
the action. You may wish to construct other paths between the end
points and show that their action is greater than that for the actual
path. (Suggestion: Use small time steps [about 1/1000 of a second],
evaluate L at each time step and estimate the integral by summing the
quantities L(t)∆t.)
Chapter 5
The Conservation of Energy

This chapter analyzes the conservation of energy in a much more

rigorous manner than the review in Section 1.3. Here you will also be
exposed to a number of other useful concepts such as energy diagrams,
the concept of metrics, and the del operator. The techniques presented
here are used quite frequently, not only in this course, but also in other
advanced physics courses.

5.1. The Work-Energy Theorem

When you push an object from one place to another, you do work
on it. To be precise, the work done on a particle by a force F during a
displacement from point 1 to point 2 along some given path, is defined
as Z r2
W = F·ds, (5.1)
r1
where F is the force acting on the particle during the displacement
ds. If there is no displacement, no work is done. By definition, the
displacement ds is the displacement of the point of application of the
force. For a particle, this is equal to the displacement of the particle.
For an extended rigid body ds refers to the displacement of the center
of mass.1
1The definition of work is often written as
Z
W = F·dr.

This definition may cause some confusion because dr represents the displacement
of the body, but it is not necessarily in the r̂ direction. It is wrong to write dr =drr̂
unless the displacement happens to be in the r̂ direction. For example, in spherical
coordinates,
dr = drr̂+rdθθ̂ + r sin θdφφ̂.
149
150 5. THE CONSERVATION OF ENERGY

It is reasonable to assume that doing work on a body will increase

its kinetic energy. For an extended body, this would include the kinetic
energy of rotation as well as the kinetic energy of translation. Since a
particle cannot rotate, it is easiest to consider, for the time being, the
work done on a particle.
The relation between work and kinetic energy is expressed by the
work-energy theorem:
Work-Energy Theorem. The increase in kinetic energy
of an object is equal to the total work done on it.

This theorem refers to the total or net work done by all the forces
acting on the particle.
The proof of this theorem is very simple. Let F be the net force
acting on a particle of mass m. Then using the definition v =ds/dt
and Newton’s second law in the form F = m dv dt
, we can write
Z r2 Z t2 Z t2
dv
W = F·ds = F · v dt = m · v dt.
r1 t1 t1 dt
Now
d 2 d dv dv dv
(v ) = (v · v) = · v + v· =2 · v.
dt dt dt dt dt
Therefore  
dv 1 d 2 d 1 2
m · v = m (v ) = mv .
dt 2 dt dt 2
So
Z r2 Z t2   Z T2
d 1 2
W = F·ds = mv dt = d(T ) = T2 − T1 = ∆T,
r1 t1 dt 2 T1

and the theorem is proved.

There are two things in this proof that I would like you to note.
One is the fact that
d 2 dv
v =2 · v.
dt dt
Keeping this relationship in mind will help you in solving a number
of problems. The other thing I would like you to be aware of is the
way the limits of integration change when the variable of integration
changes. Be sure to change the limits whenever you make a change of
variables!
To avoid confusion, in this chapter I will use ds for the infinitesimal vector
displacement.
 5.2. WORK ALONG A PATH. THE LINE INTEGRAL 151

5.2. Work Along a Path. The Line Integral

Equation (5.1) states that the work done by a force on a particle as
it is displaced from one point to another depends on the path between
the two points. In some cases, the work does not depend on the path;
in those cases the force is called “conservative” and has certain special
properties that we will study shortly. In general, however, the work
done along one path is different from the work done along some other
path. Therefore, to evaluate work, you will often have to evaluate a
path or line integral.
An example of a line integral is the work done by a force F as a
particle moves along a path described by the curve C:
Z
W = F·ds.
C
I will describe two ways to evaluate line integrals. The first is to
express the force and the displacement in Cartesian coordinates. Thus:
F = Fx ı̂+Fy ̂+Fz k̂
and
ds =dxı̂ + dy̂ + dz k̂.
So Z Z Z Z
W = F·ds = Fx dx + Fy dy + Fz dz. (5.2)
C C C C
Here the three integrals are evaluated along the curve C. Note that in
the first integral the force component Fx must be expressed in terms
of x, and likewise Fy is expressed in terms of y and Fz in terms of z,
where the relations between x, y, and z are determined by the path.

Worked Example 5.1. A particle is acted upon by the force

F = a(x + y)ı̂ + b(y − x)̂
Determine the work done by this force as the particle moves along
a straight line from the origin to the point (2, 1).
Solution: Using the definition of work,
Z Z Z
W = F·ds = Fx dx + Fy dy
ZC C
Z C

= a(x + y)dx + b(y − x)dy

C C
152 5. THE CONSERVATION OF ENERGY

Along the straight line, the relation between x and y is y = mx

where m = 1/2. In the first integral, replace y by x/2 (thus express-
ing y as a function of x) and in the second integral write x = 2y
(thus expressing x as a function of y). Then,
Z 2 Z 1
1
W = a(x + x)dx + b(y − 2y)dy
x=0 2 y=0
Z 2 Z 1
3
= axdx − bydy
x=0 2 y=0
1
= 3a − b.
2
The technique used in this example requires having an equation for
the curve in Cartesian coordinates. Obviously this procedure can
immediately be generalized to other coordinate systems.
It is interesting to note that if the path is first along the x axis
to (2,0) then up along the y axis to (2,1) we have the same initial
and end points, but the work done is quite different.
Z 2 Z 1
3
W = a(x + 0)dx + b(y − 2)dy = 2a − b.
0 0 2

The second approach is, in principle, the same as the first approach,
but it may be easier to apply in certain situations. This approach is
based on the fact that for motion along a smooth curve, the position
can be specified by a single variable, say λ. This single independent
variable can be the distance along the curve from some starting point,
or the time, or some other parameter (as the angle θ in the exampleR
below). Then writing both F and ds in terms of λ, the integral F·ds
reduces to a single integral over one variable. Thus, if F = F(λ) and
s = s(λ), then
Z Z
ds
W = F·ds = F(λ)· dλ.
C C dλ
The last integral can be expressed in a variety of ways. For example,
if F is given in Cartesian coordinates, we write
Z λ2  
dx dy dz
W = Fx + Fy + Fz dλ. (5.3)
λ1 dλ dλ dλ
This expression shows the close relationship to the first method. Nev-
ertheless, the second method is often more convenient, as illustrated
by the following example.
 5.2. WORK ALONG A PATH. THE LINE INTEGRAL 153

Worked Example 5.2. Consider the force

−y x
F= 2 2
ı̂ + 2 ̂.
x +y x + y2
R
Evaluate C F·ds along a semicircular path from (-1,0) to (+1,0),
as shown in the figure below.
Solution: The shape of the path suggests using polar coordi-
nates and the angle θ as a parameter. Since the radius of the semi-
circle is unity, we have x = cos θ, and y = sin θ and x2 + y 2 = 1.
Note that in terms of θ, Fx = − sin θ/(sin2 θ + cos2 θ) and similarly
for Fy .
Z θ=0   Z 0
dx dy
sin2 θ + cos2 θ dθ


W = Fx + Fy dθ =
θ=π dθ dθ π
Z 0
= dθ = −π
π

Y
+1

X
-1 +1

Figure 5.1. A semicircular path from (-1,0) to (1,0).

Worked Example 5.3. A particle is in a region of space where

the force on it is given by
F = xyı̂ − x2 ̂.
The particle is dragged along a path that is described by
s = s(t) = 3tı̂ + 2t2 ̂,
where t is the time. The force is in Newtons and the displacement
is in meters. Determine the work done by the force during the
time interval 0 ≤ t ≤ 1. Do this two ways, first by treating t as
a parameter (as in the second method described above), and then
by evaluating the work with equation (5.2). The force field and
the trajectory are shown in Figure 5.2. Note that as the particle
moves along the trajectory the force on it due to the force field is
154 5. THE CONSERVATION OF ENERGY

changing. We are not considering the work done by the force that
is dragging the particle through this field.
Solution: Treating t as a parameter, we write F and ds in
terms of t. Since s=xı̂+y̂ we appreciate that x = 3t and y = 2t2 .
Furthermore,
ds
ds = dt = (3ı̂ + 4t̂)dt
dt
Then the force can be expressed in terms of t as
F = (3t)(2t2 )ı̂ − (3t)2 ̂ = 6t3 ı̂ − 9t2 ̂.
Consequently,
Z Z t=1
3 2
W = F·ds = (6t ı̂ − 9t ̂) · (3ı̂ + 4t̂) = (18t3 − 36t3 )dt
t=0
18  4 1 9
=−t 0 = − J.
4 2
Now let us solve the problem using equation (5.2). Note that
we need to have expressions for dx and for dy. Since x = 3t and
y = 2t2 we can solve for t and obtain y as a function of x; that is,
t = x/3, so y = 2(x/3)2 = (2/9)x2 . Then
Z 3 Z 2 Z 3 Z 2 Z 3   Z 2 
2 2 2 9
W = Fx dx + Fy dy = xydx − x dy = x x dx − y dy
x=0 y=0 0 0 0 9 0 2
 3  2
2 x4 9 y2 81 36 9
= − = − =− J
9 4 0 2 2 0 18 4 2
Obviously, the two methods yield the same answer.

Force Field

Figure 5.2. The force field as a function of position for

the problem considered in Worked Example 5.3.
 5.3. POTENTIAL ENERGY 155

Exercise 5.1. A charged particle in an electric field E is acted

upon by a force qE. Assume the field is constant and evaluate the
work done by this force for the closed path illustrated in Figure 5.3(a).
Does the result depend on whether you traverse the path clockwise or
counterclockwise? Answers: zero, no.

y y
(a) (b)
18
2

x x
0 1 4 0 3

Figure 5.3. (a) A closed path. The corners are at

(1,0), (1,2), (4,2) and (4,0). (b) A parabolic path going
through the origin and the point (3,18).

Exercise 5.2. Evaluate the work done by the force F = yı̂+x̂

N along the parabolic path illustrated in Figure 5.3(b) as the particle
moves from (0,0) to (3,18) meters. Answer: 54 J.

5.3. Potential Energy

Forces are vectors. Although vectors were only invented about 100
years ago, it is hard to imagine trying to do physics without them.
Nevertheless, it is often much easier to deal with scalar quantities. As
you probably know, there are certain forces that can be represented
in terms of a scalar quantity called the potential energy. Examples of
forces that can be represented in terms of potential energies are the
electrostatic force, the gravitational force and the force exerted by a
spring. Scalars will usually make your life much simpler, so it is a good
idea to use them whenever possible.
For example, the force on a body of mass m at a height z near
the surface of Earth is F = −mg. If we define the scalar function
V = −mgz we can write the magnitude of the force as
dV
F =− .
dz
156 5. THE CONSERVATION OF ENERGY

As a second example, consider a stretched spring acting on an object

of mass m as illustrated in Figure 4.3. The force exerted by the spring
is F = −kxı̂. If V = 12 kx2 , then
dV
= kx = −F,
dx
and
dV
F =−
.
dx
As a third example, consider the electrostatic force:
Q1 Q2 r̂
F= .
4π0 r2
Q1 Q2 r̂
We can define V = 4π0 r
such that the electrostatic force is given by
dV
F =−
.
dr
You appreciate that in each case the derivative of V is the negative
of the magnitude of the force. (As you probably realized, the quantity
V is the potential energy. But don’t worry about the physical meaning
of V just yet. I’ll get to that pretty soon. Just consider V to be a
function whose derivative is related to the force.)
Replacing F by −dV /dr or −dV /dx in the examples above only
gives the magnitude of the force but not its direction. Is there some
way to obtain both the magnitude and the direction of the force from
the potential energy? The answer is yes. However, before continuing
with the discussion, I need to make a small digression to consider the
properties of the mathematical object called “del.”

5.3.1. The Del Operator. The del operator2 is defined as

∂ ∂ ∂
∇ ≡ ı̂ + ̂ + k̂ . (5.4)
∂x ∂y ∂z
Del is an operator, that is, it is a mathematical object that acts on a
function to produce a different function.
We shall be interested in three different kinds of operations involv-
ing del, and these are:
∇f The gradient of a scalar function f
∇ · F The divergence of a vector function F
∇ × F The curl of a vector function F.

2The symbol ∇ is also called “nabla”

 5.3. POTENTIAL ENERGY 157

Note that the gradient will generate a vector, the divergence (be-
ing a dot product) will generate a scalar, and the curl (being a cross
product) will generate a vector.

5.3.2. The Gradient. First consider the gradient, ∇f, where f =

f (x, y, z) is a scalar function. Using the definition of del, we have
∂f ∂f ∂f
gradient of f = grad f = ∇f = ı̂ + ̂ + k̂ . (5.5)
∂x ∂y ∂z
For example, if f = f (x, y, z) = 3x + 2y 2 then
 
∂ ∂ ∂
3x + 2y 2 = 3ı̂+4y̂


grad f = ı̂ + ̂ + k̂
∂x ∂y ∂z
The gradient of the scalar function f = 3x + 2y 2 produced the vector
3ı̂+4y̂. The fact that taking the gradient of a scalar function generates
a vector is one of the most important properties of the gradient.
The geometrical interpretation of the gradient is quite interesting.
To best appreciate it consider a specific example. Figure 5.4 is a contour
map, as on a regular geographic map, and it represents the shape of a
region containing two mountains. The curves on the figure are contour
lines or altitude “isopleths,” that is, lines drawn through points having
the same value of height or altitude.3 The map in our figure shows
two peaks, one higher than the other. You could (in principle at least)
determine an equation of the form h = h(x, y) that would give the
height above sea level at any point (x, y) on the map.
If you were at some spot on the mountain, say at the point marked
A, you could climb towards the peak directly (a very steep climb) or
you could walk along the isopleth passing through A. That would be a
level walk. If you walk along a contour line your altitude above sea level
does not change (dh = 0). By looking at the sketch, you can figure out
which of the three arrows points along the path of steepest ascent. It
is, of course, the arrow pointing in the direction in which the altitude
isopleths are closest together so that in moving a given distance in
that direction, you cross the largest number of isopleths. This is the
direction of the gradient, ∇h. The gradient does not necessarily point
toward the peak of the mountain. (What is the direction of ∇h at
point B?)

3In general an isopleth is a line drawn on a map through all points hav-
ing the same numerical value. You are familiar with the temperature isopleths
(“isotherms”) and pressure isopleths (“isobars”) drawn on weather maps.
158 5. THE CONSERVATION OF ENERGY

A
x

Figure 5.4. Contour plot of two mountains(altitude isopleths).

It is not difficult to prove that the gradient ∇h is a vector point-

ing in the direction of steepest ascent and whose magnitude gives the
“steepness” along this path of steepest ascent.4
Let us begin by considering the differential of an arbitrary scalar
function. You should recall from your calculus class that the differential
of a function of several variables, such as u = u(x, y, z), is given by
∂u ∂u ∂u
du = dx + dy + dz.
∂x ∂y ∂z
Keep this expression for du in mind for a moment. Now consider the
dot product of ∇u with the infinitesimal displacement vector ds. Since
ds = dxı̂ + dy̂ + dz k̂,
the dot product is
  
∂u ∂u ∂u 
∇u · ds = ı̂ + ̂ + k̂ · dxı̂ + dy̂ + dz k̂ ,
∂x ∂y ∂z
∂u ∂u ∂u
= dx + dy + dz.
∂x ∂y ∂z
But this is precisely the expression for du. Therefore,
du = ∇u · ds. (5.6)
The relationship given by equation (5.6) is so important that some
people consider it to be the definition of du. (I guess it just depends
4By “steepness” I mean the change of h for a given displacement.
 5.3. POTENTIAL ENERGY 159

on whether you think of du or ∇u as the more fundamental quan-

tity.) I will refer to relationship (5.6) as “the vector definition of the
differential.”
By Equation (5.6)
Z 2 Z 2
∇u · ds = du = u2 − u1 ,
1 1

which is the change in u due to the displacement from point 1 to point

2.
∇u and ds are both vectors. Their dot product, du, is a function
of the angle between the vectors ∇u and ds. From the definition of
dot product, if ds is perpendicular to ∇u, then ∇u · ds = 0, and so,
du = 0. Since ds is a displacement, this tells you that for a displacement
perpendicular to ∇u there is no change in u. Therefore, the lines of
constant u (contour lines) must be perpendicular to ∇u because a
displacement along a contour line results in no change in the value
of the function. (Actually, I should say surfaces of constant u are
perpendicular to ∇u because we have defined u = u(x, y, z) to be a
three dimensional function.)
Similarly, the largest change in du = ∇u · ds comes when the dis-
placement ds is parallel to ∇u, that is, in the direction of ∇u. The
largest change in du for a given displacement is along the path of steep-
est ascent, so ∇u points in the direction of steepest ascent. Further-
more, du = |∇u||ds| cosθ, and the maximum value of du occurs when
cos θ = 1, so  
 du 
 
 ds  = |∇u| .
max
That is, the magnitude of ∇u gives the value of the greatest increase
in u per unit displacement.
To summarize the preceding discussion you could state that the
direction of the gradient indicates the direction of greatest change of a
function and the magnitude of the gradient gives the maximum change
in the function per unit displacement.

Exercise 5.3. The surface of a sphere is described by x2 +y 2 +z 2 =

constant. Show that the gradient of this function points in the radial
direction.
Exercise 5.4. Evaluate the gradient of f = 5x3 +6y 2 +2z at (1,2,3).
Answer: 15ı̂+24̂+2k̂.
160 5. THE CONSERVATION OF ENERGY

Exercise 5.5. A family of coaxial cylinders is described by

f (x, y, z) = x2 + y 2 = constant.
Show that ∇f points away from the axis.
Exercise 5.6. The pressure in a body of still water is given by
P = ρgz + P0 where P0 is atmospheric pressure and z is the depth
below the surface. Show that the direction of maximum increase in
pressure is straight down. (If you got the wrong sign it is because you
did not consider that z is measured positive downward!)

5.3.3. The Relationship between Force and Potential En-

ergy. As described above, the potential energy is a function whose
derivative with respect to position gives the magnitude of the force.
But the expressions F = −dV /dr and F = −dV /dx only gave the
magnitude of the forces, but not their direction. The description is
not complete. It turns out (as you probably suspect by now) that the
gradient of the potential energy is related to the force. Specifically,
the force is equal to the negative gradient of the potential energy. In
symbols,
F = −∇V. (5.7)
This equivalence means you can work with the scalar function V in-
stead of the vector function F. This is a great advantage in solving
problems because it simplifies the calculations significantly. (You prob-
ably recollect from your introductory electricity class that determining
the electric field E by direct application of Coulomb’s law required
resolving E into components and then integrating over the charge dis-
tribution. You soon discovered it was much easier to first evaluate
the scalar function V, the electric potential, and then determine E by
differentiation.)
Not all forces can be represented by a potential energy function.
Forces that do not have a potential energy associated with them are
called nonconservative forces. Fortunately, most of the important forces,
such as the gravitational force and the electrical force, are conserva-
tive. All forces acting on the molecular and atomic level are believed
to be conservative. On the other hand, forces such as friction and air
resistance are nonconservative and do not allow one to take advantage
of the simplifications afforded by the potential energy.
How can you determine whether or not a force is conservative? The
answer is quite simple. A force is conservative if and only if its curl is
zero.
 5.3. POTENTIAL ENERGY 161

F is conservative iff ∇ × F = 0.
This statement is easy to prove. If F is conservative, then it can
be represented by the negative gradient of the potential energy, i.e.,
F = −∇V. Consequently,
∇ × F = ∇ × (−∇V ) = −∇ × (∇V ) = −curl ( grad V ) .
But the curl of the gradient of any function is zero. Therefore, if we
can write F as the gradient of V, then ∇ × F = 0.
For the sake of completeness and because the steps in the proof are
important to understand, I will now prove that curl grad V = 0 for
any function V . Using Cartesian coordinates,
   
∂ ∂ ∂ ∂V ∂V ∂V
∇ × ∇V = ı̂ + ̂ + k̂ × ı̂ + ̂ + k̂
∂x ∂y ∂z ∂x ∂y ∂z
 

 ı̂ ̂ k̂ 

=  ∂/∂x ∂/∂y ∂/∂z 
 ∂V /∂x ∂V /∂y ∂V /∂z 
 2
∂ 2V
 2
∂ 2V
 2
∂ 2V
  
∂ V ∂ V ∂ V
= ı̂ − − ̂ − + k̂ − .
∂y∂z ∂z∂y ∂x∂z ∂z∂x ∂x∂y ∂y∂x
But
∂ 2V ∂ 2V
= ,
∂y∂z ∂z∂y
because the order of taking partial derivatives does not affect the an-
swer. Therefore, each term in parenthesis is zero and
∇ × ∇V = 0.
Thus, we have proved that
curl grad (any scalar function) = 0.
5.3.4. Del in Other Representations. So far we have been us-
ing Cartesian coordinates, but it is often useful to work in some other
set of coordinates, such as cylindrical coordinates or spherical coordi-
nates. I am going to show you how to obtain representations for ∇ in
these coordinate systems. However, before doing so we must discuss
coordinate transformations.5
Coordinate Transformations: Suppose you want to transform
from the Cartesian coordinates x, y, z to a different coordinate system
5If you would like to have more information on this subject, an excellent refer-
ence is Chapter 2 of Mathematical Methods for Physicists, 5th Ed.” Arfken G, and
Weber H, Academic Press, NY, 2001.
162 5. THE CONSERVATION OF ENERGY

in which the coordinates are q1 , q2 , q3 . In the Cartesian coordinate sys-

tem the location of a point is specified by the intersection of three
planes, x = constant, y = constant and z = constant, as shown in Fig-
ure 5.5(a). For example, the point (1,2,0) is located at the intersection
of the planes x = 1, y = 2, z = 0.

z
y=const q 3=const
x=const q2=const
y t
z=const q 1=cons

x
(a) (b)
Figure 5.5. (a) A Cartesian coordinate system. (b) A
general coordinate system.

In Figure 5.5(b) I generalized the situation and specified the loca-

tion of the point by the intersection of three surfaces (q1 = constant,
q2 = constant, q3 = constant). Although it is not necessary, it is usu-
ally convenient to consider only orthogonal coordinate systems. For an
orthogonal coordinate system the surfaces are always perpendicular at
the lines of intersection. (All of the coordinate systems used in this
book are orthogonal.)
You realize, of course, that there must be a set of transformation
equations relating the Cartesian coordinates to the “new” coordinates.
The transformation equations can be written as:
x = x(q1 , q2 , q3 ),
y = y(q1 , q2 , q3 ),
z = z(q1 , q2 , q3 ).
The inverse transformations are:
q1 = q1 (x, y, z),
q2 = q2 (x, y, z),
q3 = q3 (x, y, z).
 5.3. POTENTIAL ENERGY 163

Just as ı̂,̂, k̂ are defined as the unit vectors of the Cartesian system, so
too you can define the quantities ê1 , ê2 , ê3 as the unit vectors of the new
coordinate system. These are defined so that ê1 points in the direction
of increasing q1 and is perpendicular to the surface q1 = constant, and
similarly for ê2 and ê3 .
The distance between two nearby points is ds. In the Cartesian sys-
tem ds is related to dx, dy, dz by the generalized Pythagorean relation
ds2 = dx2 + dy 2 + dz 2 .
If you express this distance in terms of the new coordinates you will
get a fairly complicated expression that reduces to the form
X
ds2 = h2ij dqi dqj i, j = 1, 2, 3. (5.8)
ij

To see how this comes about, note that since x is a function of q1 , q2 ,

and q3 , you can write
∂x ∂x ∂x
dx = dq1 + dq2 + dq3 .
∂q1 ∂q2 ∂q3
Squaring this expression leads to
 2      
2 ∂x 2 ∂x ∂x ∂x ∂x
dx = dq1 + dq1 dq2 + dq1 dq3 +···.
∂q1 ∂q1 ∂q2 ∂q1 ∂q3
Since Problem (5.4) asks you to go through this process, I did not write
out the whole expression, which has nine terms (some terms are the
same, so there are only six different terms). You can obtain similar
expressions for dy 2 and dz 2 . Each term involves a product of the form
dqi dqj . If you add and combine the three terms dx2 + dy 2 + dz 2 to
get ds2 you will obtain a sum of nine different terms involving dqi dqj
and it will have the form of Equation (5.8) above. The coefficients of
dqi dqj (the h2ij in Equation 5.8) are called scale factors. The set of nine
quantities hij (i, j = 1, 2, 3) is called the metric or metric tensor for the
coordinate system. It is easy to appreciate that
∂x ∂x ∂y ∂y ∂z ∂z
h2ij = + + .
∂qi ∂qj ∂qi ∂qj ∂qi ∂qj
A tensor can be expressed in matrix form, so the metric tensor can
be written as
 2 2 2 
h11 h12 h13
 h221 h222 h223 
h231 h232 h233
164 5. THE CONSERVATION OF ENERGY

For an orthogonal coordinate system, the scale factors have the

property
hij = 0 for i 6= j,
2
and the expression for ds (Equation 5.8) reduces to
ds2 = h21 dq12 + h22 dq22 + h23 dq32 , (5.9)
where I made a slight change in notation and replaced hii by hi .
With this expression for ds2 it is clear that for an orthogonal co-
ordinate system, the element of displacement along a given coordinate
is
dsi = hi dqi .
In vector form an infinitesimal displacement is
ds =h1 dq1 ê1 + h2 dq2 ê2 + h3 dq3 ê3 .
The element of area in the new system of coordinates is
dσij = dsi dsj = hi hj dqi dqj ,
and the element of volume is
dτ = ds1 ds2 ds3 = h1 h2 h3 dq1 dq2 dq3 . (5.10)
Now we can obtain an expression for del in generalized coordinates.
In Cartesian coordinates, ∂u/∂x describes how the function u = u(x)
varies for a given change in x. A small displacement along the x-axis
leads to a small change in u given by
 
∂u
du = dx.
∂x
Similarly, a small displacement along the q1 -axis (denoted by ds1 ) leads
to a change in u given by
 
∂u
du = ds1 .
∂s1
But
∂u ∂u
= .
∂s1 h1 ∂q1
Thus the definition of del in Cartesian coordinates
∂ ∂ ∂
∇ ≡ı̂ + ̂ + k̂ ,
∂x ∂y ∂z
leads to the following general definition of del for any coordinate system
1 ∂ 1 ∂ 1 ∂
∇ = ê1 + ê2 + ê3 . (5.11)
h1 ∂q1 h2 ∂q2 h3 ∂q3
 5.3. POTENTIAL ENERGY 165

Exercise 5.7. For cylindrical and spherical coordinates, sketch the

three planes that define the location of point (1,2,0). Label the axes
and the planes clearly.
Exercise 5.8. Sketch the element of area dσ12 for Cartesian coor-
dinates.

5.3.5. Cylindrical Coordinates . Let us apply these general

concepts to obtain an expression for del in cylindrical coordinates. Be-
gin with the transformation equations for cylindrical coordinates:
x = ρ cos φ,
y = ρ sin φ,
z = z.
Then,
dx = dρ cos φ − ρ sin φdφ,
dy = dρ sin φ + ρ cos φdφ,
dz = dz,
and
ds2 = dx2 + dy 2 + dz 2 , (5.12)
= dρ2 + ρ2 dφ2 + dz 2 .
By inspection of this last equation and comparing it to Equation (5.9),
the scale factors for cylindrical coordinates are

h21 = 1, h22 = ρ2 , h23 = 1.

The unit vectors are ρ̂, φ̂, k̂, so using Equation (5.11) yields the follow-
ing expression for ∇ :

∂ 1 ∂ ∂
∇ = ρ̂ + φ̂ + k̂ .
∂ρ ρ ∂φ ∂z
The element of volume in cylindrical coordinates is obtained immedi-
ately from Equation (5.10) as
dτ = ρdρdφdz.
166 5. THE CONSERVATION OF ENERGY

Geometrically this is the infinitesimal parallelepiped obtained from a

displacement of dρ along ρ̂, a displacement of ρdφ along φ̂, and a dis-
placement of dz along k̂. For the purposes of remembering this expres-
sion, it is probably best to visualize the volume element as illustrated
in Figure 5.6.

Exercise 5.9. Fill in the missing steps in the derivation of Equation

(5.12).

z
dz

dρ ρdϕ

y
r
x ρdϕ
dρ

Figure 5.6. Volume element in cylindrical coordinates.

The projected shaded area in the xy plane is the base
of the volume element. Note that the edge marked ρdφ
is an arc of a circle of radius ρ. The sides of the volume
element become flatter as they shrink to zero.

5.3.6. Spherical Coordinates . The relationships between Carte-

sian coordinates and spherical coordinates are
x = r sin θ cos φ,
y = r sin θ sin φ,
z = r cos θ.
Consequently,

dx = dr sin θ cos φ + dθr cos θ cos φ − dφr sin θ sin φ,

dy = dr sin θ sin φ + dθr cos θ sin φ + dφr sin θ cos φ,
dz = dr cos θ − dθr sin θ.
 5.3. POTENTIAL ENERGY 167

In Exercise 5.10 you will show that these relations lead to:

ds2 = dx2 + dy 2 + dz 2
= dr2 + r2 dθ2 + r2 sin2 θdφ2 . (5.13)
By inspection, the elements of the metric (the scale factors) for spher-
ical coordinates are
h21 = 1, h22 = r2 , h23 = r2 sin2 θ.
The unit vectors are r̂, θ̂, φ̂ so using Equation (5.11) the gradient is
expressed as:
∂ 1 ∂ 1 ∂
∇ = r̂ + θ̂ + φ̂ .
∂r r ∂θ r sin θ ∂φ
From Equation (5.10), the volume element (illustrated in Figure 5.7)
is
dτ = h1 h2 h3 dq1 dq2 dq3 = r2 sin θdrdθdφ.

dr
z dϕ
rs in θ rs in θ d ϕ
r rd θ
θ dθ

x ϕ
dϕ rs in θd ϕ

Figure 5.7. Volume element in spherical coordinates.

The volume element has curved edges but in the limit of
infinitesimal quantities the volume is given by (base ×
width × height).

Exercise 5.10. Fill in the missing steps between the two equations
in (5.13).
168 5. THE CONSERVATION OF ENERGY

5.4. Force, Work, and Potential Energy

To begin our consideration of the relation between force, work, and
potential energy let us recall the difference between conservative forces
and nonconservative forces. Work done by a nonconservative force
(such as friction) is at least partially converted into a non-recoverable
form of energy (such as thermal energy which is quickly dissipated
to the environment as heat). You cannot define a potential energy
function for a nonconservative force.
Think about lifting a rock. Imagine we raise it very slowly so the
kinetic energy (T = 12 mv 2 ) is essentially zero at all times. Call the
force that our muscles exert on the rock Fus . Then the work done by
our muscles is Z x2
Fus dx
x1
But Fus involves the tension in our muscles, the friction in our joints,
and so on. I’m pretty sure nobody has come up with a mathematical
expression for this force. Then how in the world can we evaluate the
integral? (Now comes the clever part.) To evaluate the integral we use
the fact that to lift the rock we exerted a force equal and opposite to
the force of gravity.6

Fus = −Fgrav .
Therefore, Z x2 Z x2
Fus dx = − Fgrav dx,
x1 x1
where the gravitational force is mg, directed downward.
When we raised the rock, the change in its potential energy was
Z x2
V2 − V1 = − Fgrav dx
x1
= negative of work done by the gravitational force
= (positive) work done by us.
Thus, the increase in potential energy is equal to the work done by
us, just as you would expect.
The work done by the nonconservative force (us) plus the work
done by the conservative force (gravity) is zero and since there is no
6Tobe precise, we need to exert a force slightly greater than the force of gravity
for an instant to get the rock moving, and when it reaches the top we exert a force
slightly less than the force of gravity so the rock will slow down and stop. Overall,
the average force we exert on the rock is equal to the gravitational force.
 5.4. FORCE, WORK, AND POTENTIAL ENERGY 169

net work done, the work-energy theorem tells us there is no increase in

kinetic energy.
Now suppose we drop the rock. There are no nonconservative forces
acting. The work done by the conservative gravitational force equals
the decrease in potential energy, i.e.,
Z Z
Wg = Fgrav dx = mgdx = mgh = −∆V.

According to the work-energy theorem, the work done by the gravita-

tional force is equal to the increase in the kinetic energy. Therefore, the
decrease in potential energy is equal to the increase in kinetic energy.
Generalizing to any conservative force (F = −∇V ), the work done
by the force during a displacement from r1 to r2 is
Z r2 Z r2 Z r2
W = F·ds = −∇V ·ds = − dV = V (r1 ) − V (r2 ).
r1 r1 r1

Note the use of the vector definition of the differential (see Equation
5.6). Furthermore, ∆V ≡ V (r2 ) − V (r1 ), so
W = −∆V.
That is, the work done by the conservative force will cause the potential
energy to decrease by the same amount. This might be the opposite
of what you expected. You have known for a long time that doing
work on an object will increase its potential energy. But please note
carefully that I said the work done by the conservative force causes an
equal decrease in the potential energy. When you raise a body from
the ground to a height h the conservative force (gravity) does negative
work and the potential energy increases. When the body falls back to
the ground, the conservative force does positive work and the potential
energy decreases.
You might wonder where the zero point of potential energy should
be situated. Recall that the change in potential energy is equal to the
negative of the work done by the conservative forces. Thus, you are
dealing only with the changes in V and the actual numerical values of
V at the end points of the process make no difference. The choice of
the zero level of V is completely arbitrary (but remember that once
you have made the choice you must stick with it).
For example Figure 5.8 shows a mass m at three different locations:
at the bottom of a well of depth d, on level ground, and on top of a
hill of height h.
If you chose the zero level of potential energy to be the level ground,
then V2 = 0; the particle in the well has negative potential energy
V1 = −mgd, and the particle on the hill has positive potential energy,
170 5. THE CONSERVATION OF ENERGY

3
h
2

dd
11

Figure 5.8. Illustration that potential energy is always

relative to an arbitrary value.

V3 = +mgh. The work done by the nonconservative force to take the

particle from the bottom of the well to the top of the hill is
W = V3 − V1 = mg(h + d).
If you select the zero of potential energy at the bottom of the well,
the values of V1 , V2 , V3 all change, but the work required to take the
particle to the top of the hill is still W = mg(h + d).
The quantity W = mg(h + d) is the work you have to do to take
the particle from the bottom of the well to the top of the hill. This
is not the work done by the conservative gravitational force. That
force was directed opposite to the displacement. The work done by the
gravitational field as the particle went from z1 = −d to z3 = +h is
W = −mg(h + d).
When dealing with potential energy, it is often convenient to define
some arbitrary reference position r0 as the position of zero potential
energy. In general,
Z r
V (r) − V (r0 ) = − F·ds.
r0

If V (r0 ) = 0, then Z r
V (r) = − F·ds.
r0
(The reference position is often taken to be at infinity.)

Worked Example 5.4. A particle has a potential energy

given by V = V (x) = kx2 ex/a (in joules), where k = 3 J/m2
and a = 1 m. What is the force on the particle at x = 0? What is
the force on the particle at x = 2.0 m? What is the work done by
 5.4. FORCE, WORK, AND POTENTIAL ENERGY 171

the conservative force during a displacement from x = 0 to x = 2

m?
Solution: Since this is a one dimensional problem, the relation
F = −∇V reduces to F = − dV dx
. Therefore,
dV d
= − (3x2 ex ) = − 6xex + 3x2 ex = −ex (6x + 3x2 )
 
F =−
dx dx
Therefore, for x = 0, F = 0 and for x = 2,
F = −e2 (12 + 12) = −24e2 = −177 newtons.
The work is
W = −∆V = − [V (x = 2) − V (x = 0)] = − 3(22 )e2 − 0 = −88.7 J.
 

Figure 5.9. A rope hanging off of a frictionless platform.

Worked Example 5.5. A rope of mass 2 kg and length 3 m is

partially hanging off the edge of a platform, as shown in Figure ??.
Initially, one meter of the rope is on the platform and two meters
are hanging down. Determine the work required to slowly pull the
rope onto the platform. (Assume that the platform is frictionless.)
Solution: The force required to pull the rope onto the platform
is equal to the weight of the portion of the rope hanging off the
edge. Initially this is 2 meters and finally it is zero meters. The
mass per unit length of the rope is λ = (2/3) kg/m and the weight
of the hanging portion is λgx where x is the length hanging off the
edge. That is F = λgx. The work done to pull it up is
Z Z 2  2 2
x 2 4
W = F dx = λgxdx = λg = (9.8) = 13.1 J.
0 2 0 3 2
This problem can also be solved by determining the change in po-
tential energy of the rope. Taking the platform as the zero of
potential energy we have Vf inal = 0 because the entire rope is on
172 5. THE CONSERVATION OF ENERGY

the platform. The initial value of the potential energy is

Z 0 Z 0
Vinitial = dmgx = gλ xdx = −2λg.
x=−2 −2
So the work done by the conservative force (gravity) is
Wg = −∆V = −(Vf inal − Vinitial ) = −2g(2/3) = −13.1J, (5.14)
and the work done by the applied force is W = −Wg = +13.1J.
A third approach is to simply determine the change in potential
energy of the center of mass of the hanging portion of the rope.
Once again, Vf inal = 0 whereas
Vcm(initial) = mghcm = (2/3)(2)(9.8)(−1) = −13.1J
and once again Wg = −∆V = −13.1 J, and the work done by the
applied force is 13.1 J.

Exercise 5.11. A particle of mass m is in a uniform gravitational

field where the acceleration is g. Is this field conservative?
Exercise 5.12. Show that the gravitational force, as given by New-
ton’s law of universal gravitation, F = − (GM1 M2 /R2 ) r̂, is conserva-
tive.
Exercise 5.13. A particular force can be expressed as F = 3x2 ı̂+2y 2 ̂.
Is this force conservative? (yes)
Exercise 5.14. Two positive charges (Q1 and Q2 ) are separated
by a distance r1 . They are pushed toward one another until their
separation is r2 . What is the change in potential energy? How much
work was done by external forces in pushing the charges together?
Recall that the electrostatic force is given by F = (Q1 Q2 )/(4πε0 r2 )r̂.
Answer: Q1 Q2 (r1 − r2 )/4π0 r1 r2 .
Exercise 5.15. The force exerted by a spring is F = −kx where
k is a constant of proportionality and x is the displacement from equi-
librium. Obtain an expression for the potential energy of a spring.
Answer: If x0 = 0, V = (1/2)kx2 .
Exercise 5.16. The gravitational potential energy of an object of
mass m at a distance r from the center of the Earth is V = −GmME /r
where ME is the mass of the Earth. From this obtain Newton’s Law
of Universal Gravitation.
Exercise 5.17. What is the change in potential energy of a 1 kg
mass when it is taken from the surface of the Earth to infinity? (Ignore
 5.5. THE CONSERVATION OF ENERGY 173

the effect of any other bodies in the universe.) Answer: 6.24 × 107
joules.

5.5. The Conservation of Energy

The work-energy theorem states that the net work done on a parti-
cle is equal to the increase in its kinetic energy. If Wnet is the algebraic
sum of all the work, then
∆T = Tf − Ti = Wnet .
This principle is never violated. (However, it is easy to think of situa-
tions in which it does not appear to be true.)
Suppose the only force acting on the particle is conservative. In
that case, the work done can be written as W = −(Vf − Vi ) = Vi − Vf .
Combining this with the work-energy theorem leads to
Tf − Ti = Vi − Vf
or
Tf + Vf = Ti + Vi .
This equation is the simplest expression of the law of conservation of
energy. It states that if the only work performed on a system is done
by conservative forces, then the sum of the initial kinetic energy plus
the initial potential energy (before the work was performed) is equal
to the sum of the final kinetic energy plus the final potential energy
(after the work was performed).
The sum of kinetic energy and potential energy is called the total
energy. It is denoted by E :
E = T + V.
The conservation law simply states that the total energy is constant.
This statement holds as long as the only forces acting on the
system are conservative forces.
In high school physics courses one often hears the energy conser-
vation principle expressed as, “Energy can neither be created nor de-
stroyed but it can be converted from one form to another.” A more
sophisticated version of this statement might be, “In mechanical sys-
tems the effect of a conservative force is to transform potential energy
into kinetic energy, or kinetic energy into potential energy.”
The situation gets a little more complicated when nonconservative
forces are involved. Suppose you pick a book from the floor and place
it on a table, at height h above the floor. The initial energy of the
174 5. THE CONSERVATION OF ENERGY

book was zero because it had zero kinetic energy to begin with (and
the floor is the zero point of potential energy). When the book is on
the table, once again it has zero kinetic energy and its total energy is
E = V = mgh. Where did this energy come from? Clearly you are
the ultimate source of the energy of the book. The net work done on
the book (by you and gravity) was zero, so the book’s kinetic energy
did not increase. The increase in potential energy was equal to the
negative of the work done by the gravitational force, but this energy
can be recovered (if the book falls off the table, for example). The
work done by you is not recoverable because the force you exerted was
nonconservative. If you could evaluate the chemical energy supplied
to your muscles, do you think it would be equal to mgh? (Actually it
would not, because energy is also internally converted into heat in a
working person’s body.)
Mechanics primarily deals with problems involving conservative
forces, so the conservation of energy principle is nearly always written in
the simple form E = T +V = constant, or equivalently, Ti +Vi = Tf +Vf
where subscript i represents some initial state of the system and sub-
script f represents some final state. If you wish to use the energy
principle when nonconservative forces are acting you can include them
as a work term. That is, if Wnc is the work done on the system by
nonconservative forces as the system evolves from initial state i to final
state f , then
Tf + Vf = Ti + Vi + Wnc
Z f
= Ti + Vi + Fnc · ds.
i

In solving problems, make sure you have the right sign on Wnc , de-
pending on whether the nonconservative work increases or decreases
the total energy of the system. This is usually obvious from the state-
ment of the problem or from a consideration of the behavior of the
system.

Exercise 5.18. A 40 kg child climbs 5 meters up a flagpole then

slides to the bottom. The child is moving at 3 m/s upon reaching the
bottom. What was the work done by friction? Answer: 1780 J.
Exercise 5.19. Evaluate the escape velocity of an object from the
Earth. (This means that the object is initially on the surface of the
Earth and ends up an infinite distance away.) If the object just barely
 5.5. THE CONSERVATION OF ENERGY 175

escapes, it reaches infinity with zero velocity. Ignore any other bodies
in the universe. Answer: 11.2 km/sec.
Exercise 5.20. Evaluate the velocity required for a body at 1 AU
from the Sun to escape from the solar system. Assume the solar system
consists only of the Sun (a good approximation).

5.5.1. A Reflection on the Conservation of Energy. Conser-

vation laws are extraordinarily useful for solving physics problems. To
determine the speed of a falling object it is easier to apply the law
of conservation of energy than to integrate the equation of motion.
However, conservation laws are much more than mere computational
aids; they are fundamental laws of nature that tell us how the phys-
ical universe behaves.7 Consider, for example, the universe defined
as the total collection of all material objects. There are no external
forces doing work on the universe (because there is nothing outside of
it!). Therefore, the total energy of the universe must be a constant
quantity. Similarly, the total linear momentum and the total angular
momentum of the universe are also constant quantities.
It has been suggested that conservation laws are not laws of nature,
but just mental constructs designed to help us comprehend and deal
with the physical world. The law of conservation of energy is often
cited as an example of this point of view. In your introductory physics
course you learned the law of conservation of energy in the form:
kinetic energy + potential energy = constant.
When you studied thermodynamics you learned that mechanical energy
could be converted into heat and you modified the law to:
kinetic energy + potential energy + thermal energy = constant.
Then you studied electricity and magnetism and learned of other forms
of energy associated with the electromagnetic fields. These also had to
be included in the conservation law. And so on...
This procedure gives the impression that every time we find a new
form of energy we simply add it to the left hand side of the equation and
7A physicist who finds that a conservation law has been violated knows that it
is either a mistake or a very important discovery. For example, it was believed for
many years that parity was always conserved. The discovery of a reaction in which
parity did not remain constant caused a furor among physicists and eventually led
to a much deeper insight into the nature of the physical world. (We will consider
the conservation of parity briefly in Chapter 8.)
176 5. THE CONSERVATION OF ENERGY

define energy as that quantity whose total value is constant. However,

such an approach tends to trivialize the conservation laws and we shall
not adopt it. We shall consider the conservation laws to be fundamental
to the structure of the physical universe. As shown in Chapter 8, the
conservation laws are related, on a very profound level, to symmetries
found in nature. Thus, energy conservation is related to a symmetry
in time and momentum conservation is related to a symmetry in space.
(I really don’t expect you to understand what I am talking about here
- but don’t worry, it will all become clear as we go along!)

5.6. Energy Diagrams

Imagine a particle that is initially at the top of a perfectly smooth
hill, as illustrated in Figure 5.10 which shows the particle perched on
a peak at point A. There are peaks at A, C, and E and valleys at B
and D. The only force acting on the particle is gravity.
The particle is given an infinitesimal push to the right. It slides
down the hill towards B speeding up as it goes. It climbs the hill at
C, slowing down somewhat, then speeding up as it descends into the
valley at D. Finally it climbs another big hill and stops on the tip of
the hill at point E.
In other words, the motion is what you would expect from an ordi-
nary amusement park roller coaster ride, if you could find a completely
frictionless roller coaster.

A B C D E x

Figure 5.10. A particle on the top of a frictionless hill.

If given an infinitesimal shove, the particle will slide past
B, C, D, and end up on the tip of E.

You could consider the sketch in Figure 5.10 to be a diagram of

potential energy versus position because gravitational potential energy
 5.6. ENERGY DIAGRAMS 177

(mgh) is proportional to h, the height of the particle above zero level.

Figure 5.11 shows the same situation, but now I labeled the vertical
axis V (x) and I drew a line intersecting the vertical axis at a value
labelled E. This line gives the total energy. We call such a plot an
energy diagram.

V(x)
E

x
A B C D E

Figure 5.11. An energy diagram. The vertical axis is

energy (traditionally labeled as potential energy) and the
horizontal axis is position. Note that the horizontal line
at energy E indicating the total energy is constant and
equal to the same value for all x.

This plot is a very simple energy diagram, but it illustrates a num-

ber of important points. First of all, note that the total energy is
constant (the horizontal line indicates the particle has the same total
energy whatever the value of x, the position of the particle). Secondly,
note that I knew where to draw the total energy line because I knew
that at point A the particle had no kinetic energy (TA = 0) and the
total energy at that location was equal to the potential energy.
Since E = T + V at any point, the kinetic energy is given by the
difference between the total energy and the potential energy at that
point. The energy diagram allows you to determine the kinetic energy
for any value of x by simply noting the “distance” between the line
for the total energy and the potential energy curve. The velocity is
proportional to the square root of T so this also gives an estimate of
how fast the particle is moving.
It is very important for you to know how to interpret energy di-
agrams. Consider, for example, the energy diagram of Figure 5.12
in which I plotted the potential energy of a mass m connected to a
spring of constant k. (Recall that for this system the potential energy
is V = 21 kx2 .) The total energy is E. The mass will therefore oscillate
178 5. THE CONSERVATION OF ENERGY

V(x)

forbidden forbidden
region region

-x 1 x=0 +x
1

Equilibrium point

Figure 5.12. Energy diagram for a mass on a spring.

between the points labelled ±x1 . These two points are called “turning
points.” Even though the actual physical system is a block of mass m
sliding back and forth horizontally, you can represent it by the round
dot sliding up and down in a “potential well,” as shown in the up-
per part of the figure. If the dot starts from rest at a turning point
(say −x1 ) it will slide down to x = 0, picking up speed and reaching
maximum velocity at the bottom, then slowing down as it climbs the
potential hill and coming to a stop at the other turning point (at +x1 ).
The physical motion of the block is given by the projection of the dot
onto the horizontal (x) axis. At the turning points the velocity of the
block is zero. The regions beyond the turning points are labeled “for-
bidden region.” A particle in the forbidden region would have V (x) > E
(because the turning point is the location where V (x) = E). The al-
lowed region is where V (x) < E. It is obvious that V (x) > E is
impossible because T = E − V, and T would be negative if V > E.
Since T = 12 mv 2 , a negative value of the kinetic energy requires either a
negative mass or an imaginary value for the velocity. There is no such
thing as a negative mass, and an imaginary value for velocity is not
allowed because velocity is an observable quantity and all observable
physical quantities are real.8
8Physicists
frequently use complex quantities to make the mathematics of a
problem easier, but only the real part has physical significance. Keeping this simple
fact in mind will help you to avoid confusion, especially in your study of electro-
magnetic waves where the electric and magnetic fields are expressed as complex
 5.7. SOLVING FOR THE MOTION: THE ENERGY INTEGRAL 179

5.7. Solving for the Motion: The Energy Integral

Let us consider the motion of a particle of mass m and total energy
E in a region where the potential energy has a form such as illustrated
in Figures 5.11 or 5.12 and all the forces are conservative. (We limit
ourselves to one dimensional motion for simplicity.) To obtain an ex-
pression for the position as a function of time, we can start with the
fact that the total energy is constant:
T + V = constant = E.
Using the definition of kinetic energy write
1 2
mv + V (x) = E.
2
Then,
2
v2 = (E − V (x)) ,
m
or
r
dx 2p
= E − V (x).
dt m
This fairly simple differential equation can be solved by separation of
variables. Since E is a constant and V is a function only of x,
r
dx 2
p = dt.
E − V (x) m
This equation can be integrated, at least in principle. The expression
Z x r
dx 2
p = t (5.15)
x0 E − V (x) m
is called the energy integral.
You can go no further unless you have an explicit expression for
V (x). If such an expression is given, carry out the integration over x
and obtain a mathematical expression involving x and x0 and E. That
is, obtain
t = t(x, x0 , E).
Finally, invert this equation and solve for x in terms of x0 , E, and t to
get an expression of the form
x = x(t) = x(x0 , E, t).
functions whose real parts correspond to the actual measurable quantities. This
concept is also important in quantum mechanics where you must keep in mind that
measurable physical quantities are expressed as real numbers even though the wave
functions are complex.
180 5. THE CONSERVATION OF ENERGY

Solving for the motion means obtaining an expression for x as a function

of time and various constants. The constants are usually the initial
position and the initial velocity. Here x(t) is given in terms of initial
position and total energy, which is not quite the same thing, but is
obviously equivalent.
Note that solving for the motion using the energy only requires one
integration, whereas starting with the equation of motion requires two
integrations. The reason is that the conservation of energy equation
can be expressed as
1 2
mv = E − V (x)
2
which is essentially an expression for the velocity. If we start with
the equation of motion, the first integration gives the velocity. For
this reason, the conservation of energy equation is often called a “first
integral.”
Since the total energy E remains constant (as long as the forces
are conservative), E is called a “constant of the motion.” As you have
noted, knowing E makes the problem easier to solve. In general, know-
ing the constants of the motion makes problem solving much easier.

Worked Example 5.6. A particle of mass m = 0.1 kg moves

in a one dimensional potential field given by V = −2x + 3x2 J.
Determine the motion if the total energy is 1 J. Assume that at
time t = 0 the particle is at x = 1/3 m.
Solution: Begin with equation (??):
r Z x Z x(t)
2 dx dx
t= p = √
m x0 E − V (x) 1/3 1 + 2x − 3x2
x(t)
1 ±(1 − 3x)
= − √ sin−1

 .
3 2 
1/3

The last expression came from evaluating the standard form inte-
gral
2ax − b
Z
dx 1
√ = −√ sin−1 √ .
ax2 + bx + c −a b2 − 4ac
 5.8. THE KINETIC ENERGY OF A SYSTEM OF PARTICLES 181

Therefore,
r  
2 1 −1 1 − 3x
t = − √ sin ±
m 3 2
1 − 3x √
± = − sin 60t
2
1 2 √
x = ± sin 60t.
3 3
The particle oscillates sinusoidally in a potential well between the
turning points at x = − 31 and x = +1.

Exercise 5.21. A particle moves in a constant potential energy

field, V = V0 = constant. Assume E is known. Determine the position,
velocity and acceleration as a function
p of time, that is, x = x(t), v =
v(t), and a = a(t). Answer: x(t) = 2(E − V0 )/mt + x0 .
Exercise 5.22. Show that a particle in a region of space where the
potential energy is constant will move with zero acceleration.

5.8. The Kinetic Energy of a System of Particles

Nearly everything that has been said in this chapter has referred
to a single particle or to a rigid body that can be treated as a particle
located at the center of mass of the body. But a rigid body can rotate
as well as translate. We now show that the energy of a system of
particles can be expressed as the energy of translation of the center of
mass, plus the energy of the particles relative to the center of mass.
We shall show that
TT otal = TCM + TwrtCM
where TCM is the translational kinetic energy of a particle whos mass
is the total mass of the body and which is moving with the center of
mass of the body, and TwrtCM is the kinetic energy of all the particles
with respect to the center of mass. (This is esentially what we assumed
when we considered a disk rolling down an inclined plane and wrote
1 1
T = Ttrans + Trot = M v 2 + Iω 2 .
2 2
Keep in mind that a rigid body is a collection of particles that
maintain fixed distances from one another.
182 5. THE CONSERVATION OF ENERGY

Before going any further, you should refresh your memory on the
definition of center of mass. In particular, recall that for a collection
of N particles located at ri , the position of the center of mass, denoted
rc , is PN
mi ri
rc = Pi=1N
. (5.16)
i=1 m i
So
XN
M rc = mi ri , (5.17)
i=1
P
where M = mi is the total mass.

mi
z i

r'
i
r cm
i
rc
y

Figure 5.13. A system of particles. Particle i with

mass mi is located at position ri . The center of mass
(cm) is at rc . The arrow marked r0i is the position of
particle i with respect to the center of mass.

We now show that the total kinetic energy of a system of particles

can be written as the sum of the kinetic energy of translation of a
particle of mass M moving with the center of mass and the kinetic
energy due to the motion of all the particles relative to the center of
mass. To prove this, begin with the fact that the kinetic energy of
particle i is
Ti = 21 mi vi2 = 21 mi (ṙi ·ṙi ),
where the vector ri is measured from the origin. As shown in Figure
5.13 the position vector of particle i can be described in terms of the
position of the center of mass by
ri = rc + r0i ,
where r0i is the position of particle i relative to the center of mass.
Therefore, the kinetic energy of particle i can be expressed as
Ti = 12 mi (ṙc + ṙ0i ) · (ṙc + ṙ0i ) = 21 mi ṙc2 + 2ṙc · ṙ0i + ṙi02


 5.9. WORK ON AN EXTENDED BODY. PSEUDOWORK 183

The total kinetic energy is obtained by adding the kinetic energies of

all the particles in the system.
N
X X X X
T = Ti = 12 ṙc2 mi + ṙc · mi ṙ0i + 1
m ṙ02 .
2 i i
i
The middle term is zero because
X d X d X
mi ṙ0i = mi r0i = mi (ri − rc ),
dt dt
and
X X X
mi (ri − rc ) = mi ri − mi rc = M rc − rc M = 0.
Consequently, the total kinetic energy of a system of particles is given
by X X
T = 21 ṙc2 mi + 1
m ṙ02 ,
2 i i
or X
T = 12 M vc2 + 1
m v 02 .
2 i i
(5.18)
If you imagine a particle of mass M at the center of mass, then the
total kinetic energy of a system of particles is equal to the sum of the
translational kinetic energy of M plus the kinetic energy of all of the
particles relative to the center of mass, and the proposition is proved.

Exercise 5.23. Two particles are on a collision course along the

x-axis. In the laboratory coordinate system a particle of mass 2 kg
moves towards the right with a speed of 2 m/s, and a particle of mass
1 kg moves to the left with a speed of 1 m/s. (a) Determine the total
kinetic energy of the system. (b) Determine the velocity of the center
of mass. (c) Determine the kinetic energy of each particle relative to
the center of mass. (d) Verify Equation (5.18). Answers: (a) 4.5 J (b)
1 m/s.

5.9. Work on an Extended Body. Pseudowork

The work done on a particle was defined by Equation (5.1),
Z r2
W = F·ds,
r1
where F was the force acting on the particle as it moved from r1 to r2 .
The displacement ds can be interpreted either as the displacement of
184 5. THE CONSERVATION OF ENERGY

the particle or as the displacement of the point of application of the

force, since the two are equivalent. Furthermore, it is clear that the
source of the increase in the particle’s kinetic energy is the agent that
is producing the force. However, the situation gets more complicated
when you consider the work done when a force acts on an extended
body.
A simple example of an external force acting on an extended body
is an ice skater who puts her hand on a wall and pushes herself away
from it. The external force is the reaction force of the wall on her hand.
(She pushes on the wall and the wall pushes back on her with an equal
and opposite force.) The force acting on the skater stops as soon as her
hand loses contact with the wall. The point of application of the force
did not move at all. Did the wall do work on the skater? The answer
is obvious from an energy point of view: The energy of the wall did
not change! The source of the kinetic energy imparted to the skater
was chemical energy from her own body. This chemical energy was
converted into mechanical energy through the action of her muscles.
Nevertheless, the only unbalanced external force acting on the skater
was the reaction force the wall exerted on her hand.
To determine the appropriate form of the equation for work in the
case of an extended body, we begin with Newton’s second law in the
form
Fe = mac ,
where ac is the acceleration of the center of mass of the extended body
and Fe is the external force. Multiplying both sides of this equation by
an infinitesimal displacement of the center of mass, dsc leads to
dvc dsc
Fe dsc = mac dsc = m dsc = mdvc = mvc dvc .
dt dt
Integrating from initial point i to final point f, gives
Z f Z f    
1 2 1 2
Fe dsc = mvc dvc = mv − mv = ∆T.
i i 2 c f 2 c i
Therefore, the change
R f in kinetic energy of the extended body is equal
to the expression i Fe dsc . This looks suspiciously like the work. But
it isn’t exactly like the work because the point of application of the
force did not move. Furthermore, the body exerting the force was not
the source of the energy increase. For this reason, some authors prefer
to call this expression the “pseudowork” and they call the equation
Z f
Fe · dsc = ∆T
i
 5.10. SUMMARY 185

the “energy equation” to distinguish it from the work-energy theorem

in which the external force is the source of the energy.9

5.10. Summary
In this chapter you have been exposed to a number of mathematical
concepts and a number of physical concepts. In this summary the two
sets of concepts are listed separately, first the math, then the physics.

5.10.1. Mathematical Concepts. R

Line Integral. By definition, Work = F·ds. The integral is a line
integral, and it must be evaluated along the path taken by the particle.
You have seen two ways to evaluate a line integral: (1) Express the
force and the differential of displacement in component form, thus
Z Z Z Z
W = F·ds = Fx dx + Fy dy + Fz dz,
C C C C

or (2) Express the force and displacement in terms of some parameter

λ, and evaluate
Z Z
ds
W = F(λ)·ds = F(λ)· dλ.
C C dλ
Del Operator. The del operator in Cartesian coordinates is de-
fined as
∂ ∂ ∂
∇ = ı̂ + ̂ + k̂ .
∂x ∂y ∂z
Operations involving del include
∇f The gradient of the scalar function f
∇ · F The divergence of the vector function F
∇ × F The curl of the vector function F.
This chapter concentrated on del operating on a scalar function.
This generates a vector called the gradient that is denoted by ∇f .
Geometrically, think of the gradient as a vector pointing in the direction
of the greatest increase in f, whose magnitude gives the rate of increase
of f in that direction.

9A very interesting article on this subject is the paper by Bruce Sherwood,

Pseudowork and Real Work Am. J. Phys, 51, 597-602, 1983.
186 5. THE CONSERVATION OF ENERGY

Metric. The differential displacement vector ds has different forms

in different coordinate systems. In terms of the generalized coordinates
q1 , q2 , q3 ,
X
ds2 = h2ij dqi dqj , i, j = 1, 2, 3.
ij

The scale factors hij depend on the geometrical properties of the coor-
dinate space and collectively are called the metric.
Del and Volume Element in Other Representations. Using
the transformation properties of coordinates we derived expressions for
the volume element and del in terms of generalized coordinates as

dτ = ds1 ds2 ds3 = h1 h2 h3 dq1 dq2 dq3 ,

and
1 ∂ 1 ∂ 1 ∂
∇ = ê1 + ê2 + ê3 .
h1 ∂q1 h2 ∂q2 h3 ∂q3
In cylindrical coordinates these yield
dτ = ρdρdφdz,
∂ 1 ∂ ∂
∇ = ρ̂ + φ̂ + k̂ ,
∂ρ ρ ∂φ ∂z
and in spherical coordinates they yield
dτ = r2 sin θdrdθdφ,
∂ 1 ∂ 1 ∂
∇ = r̂ + θ̂ + φ̂ .
∂r r ∂θ r sin θ ∂φ

5.10.2. Physical Concepts. I will now summarize the many im-

portant physical concepts introduced in this chapter.
Work. The work done on a particle by a force F during a displace-
ment from r1 to r2 is
Z r2
W = F·ds
r1

As noted above, the integral is a line integral and must be evaluated

along the path followed by the particle. However, if the force is conser-
vative, then the value of the integral depends only on the end points.
Work-Energy Theorem. The total work done by all of the ex-
ternal forces equals the increase in kinetic energy of the particle.
W = ∆T = Tf − Ti .
 5.10. SUMMARY 187

Potential Energy. If a force is conservative (∇ × F = 0) we can

associate with it a potential energy V. The relation between force and
potential energy is
F = −∇V.
The work done by a conservative force can be expressed in terms of the
change of potential energy, thus

W = −∆V = −Vf + Vi .

Conservation of Energy. For conservative forces, equating the

two expressions for work leads to the conservation law

Tf + Vf = Ti + Vi .

Energy Diagrams. An energy diagram is a plot of the potential

energy (V ) as a function of position (x). The total energy is represented
by a constant horizontal line on such a diagram. The distance between
the total energy line and the potential energy curve (E − V ) is equal
to the kinetic energy and is proportional to the speed of the particle
squared.
Solving for the Motion using Energy. When the force is a
function of position, energy methods are the easiest way to determine
the motion. Basically, this involves evaluating an integral of the form
Z x r
dx 2
p = t.
x0 E − V (x) m

Energy of a System of Particles. The kinetic energy of a system

of particles can be expressed as the sum of the kinetic energy of the
center of mass and the kinetic energy with respect to the center of
mass:
X
T = 12 M vc2 + 1
m v 02
2 i i
= Tc + Twrt .
cm

Work on an Extended Body. Although the equations for the

work done on an extended body are the same as the equations for the
work done on a particle, the two concepts are not the same. In particu-
lar, the source of energy may be different in the two cases. Nevertheless,
the correct answer is obtained by simply replacing the displacement of
the particle by the displacement of the center of mass of the extended
body.
188 5. THE CONSERVATION OF ENERGY

5.11. Problems
Problem 5.1. The force R on a particle is given by F =3xı̂+2y̂ N.
Determine the line integral F·ds for the straight line path that starts
at the origin and ends at the point (3,6) m. Answer: 49.5 J.
Problem 5.2. In this problem we assume that somehow an elec-
tron is dragged past a proton in a straight line path. The electron is
subject to the electrostatic force and to whatever external force keeps
it moving in a straight line at constant speed. (a) Evaluate the work
done by the electrostatic force as the electron is taken from x1 to x2
along the trajectory of Figure 5.14. (b) What is the work done by the
external force required to keep the electron on the straight line path?
(c) Evaluate the work done by the electrostatic force if the electron is
taken around a semicircular path with the proton at the center.

x ^r
x1 x2
0
q
- q

r ds = dx
d
F
+

Figure 5.14. An electron is dragged past a proton from

x1 to x2 along the straight line path.

Problem 5.3. The magnitude of the force on a particle is F = − kr

where r is the distance from the origin to the particle. The force is
directed towards the origin at all times. (This is an example of a
central force.) Determine the work done when the particle is moved
along a semicircular path of radius R from the origin to point (2R, 0).
Determine the work done when the particle is moved from the origin
to the point (2R, 0) along a straight line path. (Note: The origin is not
at the center of the semicircular path.)
Problem 5.4. Derive Relations (5.8) and (5.10) from the transfor-
mation equations and the expression for ds2 in Cartesian coordinates.
Problem 5.5. (a) Determine the metric for plane polar coordi-
nates. (b) Obtain an expression for ds in polar coordinates, as well as
the element of area.
Problem 5.6. Determine ds2 , the scale factors, the vector ds, the
volume element and the ê vectors for the paraboloidal coordinates
 5.11. PROBLEMS 189

u, v, φ :
x = uv cos φ,
y = uv sin φ,
1 2
u − v2 .


z =
2
Problem 5.7. The “elliptic cylindrical” coordinates, u, v, z, are
defined by
x = a cosh u cos v,
y = a sinh u sin v,
z = z.
Determine the metric. Write an expression for ds.
Problem 5.8. Consider the u, v, z coordinates defined by
1 2
u − v2 ,


x =
2
y = uv,
z = z.
(a) Determine the metric for these coordinates. (b) Evaluate ∇ in these
coordinates.
Problem 5.9. The prolate spheroidal coordinates η, θ, φ are related
to the Cartesian coordinates by the following transformation equations
x = a sinh η sin θ cos φ,
y = a sinh η sin θ sin φ,
z = a cosh η cos θ.
Determine the expression for ∇ using these coordinates.
Problem 5.10. (a) Using the expression for ∇ given by Equation
(5.11), obtain ∇ · V in cylindrical coordinates. Note that the deriva-
tives also act on the unit vectors, as in going from Equation (2.8) to
∂V
Equation (2.9). Answer: ∇ · V = ρ1 ∂ρ ∂
(rVρ ) + ρ1 ∂φφ + ∂V∂z
z
. (b) Show that
the same result is obtained
h from i
∇ · V = h1 h12 h3 ∂q∂1 (h2 h3 V1 ) + ∂q∂2 (h1 h3 V2 ) + ∂q∂3 (h1 h2 V3 ) .

Problem 5.11. Using spherical coordinates, determine the volume

of a sphere. Pay particular attention to the limits on the integrals over
θ and φ.
190 5. THE CONSERVATION OF ENERGY

Problem 5.12. Near the surface of the Earth the potential energy
is mgh. Use this fact to obtain an equation for the position of a falling
particle as a function of time. Show that the acceleration of the particle
is g.
Problem 5.13. Upon finding that the top of a certain mountain
is a perfect cone, a surveyor describes it mathematically in cylindrical
coordinates by the formula z = h0 − ρ. (a) Sketch the contour lines.
(b) Show that the direction of steepest ascent everywhere points toward
the summit.
Problem 5.14. A mountain rises above a flat plain. The height of
the mountain above the plane is described by the relation
z(x, y) = 2000 exp − x2 + 2y 2 /8000


meters.
(a)How high is the mountain? (b) If you are standing at x = 20 m, y =
10 m, how high above the plain are you? (c) At that location, what is
the direction of quickest descent? (d) What is the maximum rate of
descent at this point? (That is, how many meters do you descend for
every meter of horizontal displacement?) Answer: (d) -13.12 m/m.
Problem 5.15. The temperature in a certain location is given by
T = T0 − A(2x2 + y 2 + z 2 ) (in kelvin). Distances are in meters and
the constant A has the value 0.5 K/m2 . The value of T0 is 300 K. (a)
At point (1,2,2) what is the direction and rate (in K/m) of maximum
temperature increase? (b) If you move one meter in that direction
what is the temperature at the new position? (c) The temperature
at the new position as calculated by the formula is not equal to the
value obtained by multiplying the rate of temperature decrease by the
displacement. Explain this discrepancy.
Problem 5.16. (a) Determine the center of mass of the system
composed of three particles of masses 1, 2, 3 kg located at (0,1), (1,0)
and (1,1) respectively. (b) Assume the 1 kg particle has a velocity
of 1̂ m/s, the 2 kg particle has a velocity of 2ı̂ m/s and the 3 kg
particle has a velocity of 3(ı̂+̂) m/s. Determine the total kinetic energy
of the system. (c) What is the velocity of the center of mass? (d)
Determine the kinetic energy of the particles relative to the center of
mass. Answers: (a) (5/6)ı̂+(4/6)̂, (b) 31.5J, (c) 136
ı̂+ 10
6
̂ m/s, (d) 9.08
J.
Problem 5.17. Two positive charges (Q1 and Q2 ) are located on
the x-axis at positions x = ±a. A negative charge (−q) of mass m is
constrained to move along the y-axis. The electrostatic potential of the
 5.11. PROBLEMS 191

system is
qQ1 qQ2
V =− −
4π0 r1 4π0 r2
where r1 and r2 are the distances from the negative charge to the
positive charges. The negative charge is released from rest at y = b.
Obtain an expression for the maximum speed of the negative charge.
Where does this occur?
Problem 5.18. A football player of mass M decides to go bungee
jumping. He suits up in a halter with a long elastic cord of unstretched
length b. After tying the free end of the rope to a high bridge he leaps
out into space. A few moments later he finds himself suspended far
above the ground. He climbs the bungee cord back up to the bridge.
Determine the ratio of the work done climbing the elastic rope to the
work done in climbing an inelastic rope of length b. You may assume
the elastic rope obeys Hooke’s law and has a force constant k.
Problem 5.19. A girl on a swing is pushed harder and harder by
her older brother. Eventually, she is swinging in an arc whose highest
point is 1.5 m above its lowest point. Assume the swing has massless
ropes 2.8 m long and that the mass of the girl (plus swing seat) is 30
kg. What is the tension in the ropes when the girl passes through the
lowest point?
Problem 5.20. A block of mass 5 kg is on a horizontal surface.
The coefficient of sliding friction is 0.5. The mass has a speed of 10 m/s
and is moving towards a stationary spring that is 3 meters away. The
spring constant is 4000 N/m. The block strikes the spring, compresses
it, and bounces back. How far from the spring does the block come to
rest?
Problem 5.21. A radioactive nucleus decays by emitting an alpha
particle with energy 6.0 MeV. The total energy released in the disinte-
gration is 6.2 MeV. Determine the mass of the recoiling nucleus (ignore
relativistic effects).
Problem 5.22. A particle of mass m and total energy E is moving
in a one-dimensional potential given by V (x) = −bx. Determine the
motion.
Problem 5.23. A rail gun is a device that accelerates a projectile
to extremely high speeds using magnetic forces. Suppose a particle of
mass m is fired vertically from the surface of the Earth. Ignoring air
resistance, show that the maximum height reached by the particle is
192 5. THE CONSERVATION OF ENERGY

given by
GM m
−R
|E|
where R is the radius of the Earth and M is its mass. (Assume that
its total energy E is negative.)
Problem 5.24. Prove that for a particle the rate of change of
kinetic energy dT
dt
is equal to the dot product of the force acting on it
and the instantaneous velocity of the particle, that is, F · v.
Problem 5.25. Let h = r − R be the position of a particle above
the surface of Earth where R is the radius of the Earth. Prove that in
the limit
hR
the gravitational potential reduces to mgh.
Problem 5.26. A particle of mass m is in a one-dimensional po-
tential energy field given by
2
V (x) = −Ae−ax ,
where A and α are constants. (a) Plot the energy diagram. (b) Deter-
mine the turning points if the particle has a total energy E = −0.5A.
(c) Determine the turning points if the particle has a total energy
E = −Ae−1 . (d) Assume the particle has zero energy and is located at
x = −∞. It is given an infinitesimal shove towards the origin. What is
its velocity when it passes through the origin?
Problem 5.27. In a certain region of space the potential energy
can be expressed as
A
V = −p ,
x2 + y 2 + z 2
where A is a constant, and the origin is excluded. (a) Obtain an expres-
sion for the force. (b) Determine the work required to take a particle
from (x1 , y1 , z1 ) to ∞. (c) Express the force in spherical coordinates.
Problem 5.28. The potential energy of a vibrating diatomic mol-
ecule as a function of the separation (s) between the two atoms is
approximately given by the “Morse Function,”
V (s) = V0 (1 − e−(s−s0 )/δ )2 − V0 ,
where s0 , δ, V0 are constant parameters. (a) Obtain an expression for
the force on the atoms. (b) Determine the separation between the
atoms when the potential energy is a minimum. (c) What is the
 5.11. PROBLEMS 193

minimum value of the potential energy? Answer (a) − (2V0 /δ) (1 −

e−(s−s0 )/δ )e−(s−s0 )/δ .
Problem 5.29. Coulomb’s law tells us that the electrostatic force
between two charged particles is
Q1 Q2
F= r̂
4πε0 r2
where Q1 and Q2 are the charges on the particles and r is the sepa-
ration between them. The quantity 4πε0 is constant. Show that the
electrostatic force is conservative. Obtain an expression for the poten-
tial energy.
Problem 5.30. A particle is subjected to a force given by
F =K[(2x + y)ı̂ + (x + 2y)̂].
(a) Prove that this force is conservative. (b) Obtain an expression for
the potential energy.
Problem 5.31. Newton’s law of universal gravitation states that
the force between two particles of masses m1 and m2 is
Gm1 m2
F=− r̂,
r2
where G is a constant and r is the distance between the particles. Use
this force expression to obtain the gravitational potential energy of the
system. Select an appropriate point for the zero of potential energy.
Problem 5.32. By evaluating the work required to bring three
masses m1 , m2 , and m3 from infinity to their final positions at r1 , r2 ,
and r3 determine the potential energy of the system. (The only force
acting on the particles is their mutual gravitational attraction.)
Problem 5.33. A particle of mass 2 kg moves along the x axis. Its
potential energy as a function of position is V (x) = −3x + x2 joules.
(Here x is measured in meters.) The particle passes through the origin
with a speed of 4m/s. (a) Sketch the potential energy as a function
of x. (b) Determine the speed of the particle when it is at x = 1. (c)
Where are the turning points?
Problem 5.34. Consider the potential
a b
V = − + 2.
r r
(a) Plot the potential as a function of r. (b) Obtain an expression for
the force. (c) Obtain an expression for the turning points as a function
of a, b, and the total energy E.
194 5. THE CONSERVATION OF ENERGY

Problem 5.35. A particle moving in a potential given by V (x, y) =

ax2 +by 2 is a two-dimensional oscillator with different force constants in
the two directions. (This is called a nonisotropic oscillator.) Determine
the motion, assuming that at time t = 0 the particle is passing through
the origin with velocity v =v0x ı̂ +v0y ̂.
Computational Projects
Computational Project 5.1. Assume that the potential energy
is given by V (x) = −(1/x6 ) + (1/x12 ). Integrate Equation (??) numer-
ically to determine the position of a particle of unit mass and total
energy 10−4 energy units. Plot x = x(t) and v = v(t) for a particle
coming in from infinity. (Infinity is approximately at x = 5.)
Chapter 6
Conservation of Linear Momentum

We now turn our attention to problems that can be solved using the
law of conservation of linear momentum. Examples of such problems
include the motion of a rocket, collisions in one and two dimensions,
and the behavior of a system when an impulsive force acts on it.

6.1. The Law of Conservation of Momentum

The law of conservation of linear momentum is based on Newton’s
second law: The rate of change of momentum of a system is equal to
the net external force acting on the system:
dP
= F,
dt
where F is the net external force and P is the total linear momentum.
Therefore, if F = 0, then P =Σi mi vi = constant.1 In words, the law
of conservation of linear momentum is:
If no net external force acts on a system,
the total momentum of the system is constant.
For example, in the collision of two bodies, if external forces are negli-
gible, the conservation of momentum principle can be expressed as
Pfinal = Pinitial ,
where Pinitial is the total momentum of the two bodies before the colli-
sion and Pfinal is the total momentum after the collision. Many prob-
lems can be solved using this simple relation.
1In Chapter 8 you will be exposed to a more general derivation of the law of
conservation of linear momentum.
195
196 6. CONSERVATION OF LINEAR MOMENTUM

If a net external force does act on a particle, and if the mass of a

body changes with time, then Newton’s second law must be expressed
in the form
dP dv dm
F= =m +v . (6.1)
dt dt dt
Note that dv/dt is the acceleration of the body; in this case, one cannot
write Newton’s second law in the form F =ma.

Exercise 6.1. A bullet of mass m and velocity v is fired at a wooden

block of mass M at rest on a frictionless surface. The bullet embeds
itself in the block. (a) Determine the speed of bullet plus block after
the impact. (b) Is kinetic energy conserved? Explain. (c) Evaluate the
loss of kinetic energy. Answers: (a) mv/(M +m), (c) ∆T = − 12 m+M mM 2
v .

6.2. The Motion of a Rocket

As an application of the law of conservation of linear momentum to
a system of variable mass, consider the motion of a rocket. Imagine the
rocket to be in empty space, far from any stars or any other material
bodies so there are no external forces acting on it. The pilot decides
to fire the rocket engines for a short time, say dt. The rocket engines
burn an amount of fuel dM . The burned fuel is expelled from the
rocket exhaust tubes at a velocity u with respect to the rocket. How
much faster is the rocket traveling after this fuel burn?
Let the initial mass of the rocket be M and let V be its initial ve-
locity, relative to some inertial reference frame. The initial momentum
is, therefore,
Pi = M V.
After the burn, the mass of the rocket is M − dM and its velocity
is V + dV. The mass of the burned fuel is dM and its velocity in
the inertial reference frame is the vector sum V + u. The total final
momentum of the rocket plus burned fuel is
Pf = (M − dM )(V + dV) + dM (V + u).
See Figure 6.1
Since there is no net external force acting on the rocket/fuel system,
the total momentum must be constant; that is, the momentum after
the burn must be equal to the momentum before the burn. Therefore,
(M − dM )(V+dV) + dM (V + u) = M V.
 6.2. THE MOTION OF A ROCKET 197

Figure 6.1. A rocket before and after it fires an amount

of fuel dM.

Next we carry out the indicated multiplications and discard all second
order differentials. (The term dM · dV is a second order differential.
The product of one infinitesimal quantity with another infinitesimal
quantity generates a very small quantity indeed!) This procedure yields
(and you should verify this result yourself):
M dV = −udM.
Dividing both sides by dt gives
dV dM
M = −u . (6.2)
dt dt
Consider the left hand side of (6.2). Since dV/dt is the acceleration of
the rocket and M is its mass, the left hand side is mass times acceler-
ation. Therefore, the right hand side looks like the force on the rocket,
usually called the thrust. The thrust depends on the rate at which fuel
is burned (dM/dt) and on the velocity with which the burned fuel is
ejected, u. It is convenient to express Equation (6.2) in scalar form as
dV dM
M = −u .
dt dt
The right hand side appears to have the wrong sign, but recall that
dM/dt is negative. Since M is changing, this equation is not of the
form F = ma. In fact, when solving problems, it is usually safer to
write
dM
dV = −u .
M
(What would you write if the rocket is slowing down by firing its retro-
rockets?)
To generate a large thrust, a rocket motor is designed to burn fuel
very rapidly (large dM/dt) and to expel it at as high a speed as possible
(large u). The whole purpose of burning the fuel is to convert it to a
198 6. CONSERVATION OF LINEAR MOMENTUM

gas, thus greatly increasing its volume and causing it to be expelled

through the rocket tubes at the highest possible velocity.

Worked Example 6.1. A rocket of total mass m0 in inter-

planetary space is coasting at speed v0 . It approaches an interest-
ing asteroid and must slow down to the speed of the asteroid which
happens to be v0 /2. Determine the amount of fuel that must be
burned to achieve this reduction in speed.
Solution: Since the rocket is slowing down we write
dM
dV = u .
M
Integrating
1 v0 /2
Z Z mf
dM
dV = ,
u v0 m0 M
1 mf
(v0 /2 − v0 ) = ln ,
u m0
v0 mf
− = ln ,
2u m0
mf = m0 e−v0 /2u .
The amount of fuel that must be burned is
m0 − mf = m0 (1 − e−v0 /2u ).

Worked Example 6.2. The rocket equation was obtained

by considering the difference in momentum between an initial and
final state, assuming a small amount of fuel was burned. Obtain
the same result by showing that the total momentum is constant.
Assume the rate of fuel burn is constant and equal to ṁ.
Solution: At any given moment the total momentum of the
system is the momentum of the rocket (including unburned fuel)
and the momentum of the expelled (burned) fuel. We shall solve
the problem using scalar quantities. Let the mass of the rocket
at any instant be MR and the mass of burned fuel be mf . The
ejection speed of the fuel in inertial space is VR − u where u is
a positive quantity. The rate of change of mass of the rocket is
dMR /dt = −ṁ and the rate of change of the mass of burned fuel is
 6.2. THE MOTION OF A ROCKET 199

dmf /dt = +ṁ. To determine the momentum of the burned fuel at

time τ we need to keep in mind that while it was being burned, the
rocket was moving at a changing speed. Therefore, the momentum
of the burned fuel at time τ is
Z τ Z τ
Pf (τ ) = ṁ(VR (t) − u)dt = ṁ (VR − u)dt
0 0
The momentum of the rocket at time τ is
PR (τ ) = MR (τ )VR (τ )
The total momentum is Ptot = PR + Pf so the rate of change of
momentum is
 Z τ 
dPtot d
= MR (τ )VR (τ ) + ṁ (VR − u)dt
dt dt 0
d
= (MR VR ) + ṁ(VR − u)
dt
dMR dVR
= VR + MR + ṁVR − ṁu = 0
dt dt
But ṁ = − dM
dt
R
, so
dVR dMR
MR = −u .
dt dt
Note that in this expression the ejection speed u is positive and
rate of mass decrease (dMR /dt) is negative.

Exercise 6.2. A rocket of initial mass 500 kg burns fuel at a rate of

5 kg/sec. The exhaust speed of the gases is 300 m/s. What is the initial
acceleration of the rocket? (Ignore gravity.) What is its acceleration
after one minute? Answers: 3 m/s2 , 7.5 m/s2 .

Exercise 6.3. Joe and Bill are railroad men who are riding side
by side on flatcars rolling on parallel tracks. The tracks are straight
and perfectly horizontal. There are absolutely no frictional forces or
air resistance. It starts to snow. Joe sweeps the snow off his flatcar as
soon as it lands, sweeping it off the side, perpendicular to the direc-
tion of motion of the flatcar. Bill, the lazy one, simply lets the snow
accumulate on his flatcar. Who travels further in the same interval
of time? Answer this question conceptually and also mathematically.
(The snow falls perfectly vertically.)
200 6. CONSERVATION OF LINEAR MOMENTUM

Exercise 6.4. Beginning with Newton’s second law in the form

F = dP/dt and the definition of derivative, generalize equation (6.2)
to include the effect of an external force. Answer: M dV
dt
+ u dM
dt
= F.
Exercise 6.5. Consider a rocket that is rising into the air. The
burning exhaust gases stream out through the nozzles and push down
on the air below. But what pushes upward on the rocket? (An equiv-
alent question: When you inflate a balloon and then release it, it flies
wildly about the room. What is pushing the balloon?)

6.3. Collisions
Another application of the principle of conservation of linear mo-
mentum is in the analysis of collisions. Generating and studying col-
lisions between elementary particles are a principal way physicists ex-
plore the underlying properties of nature. A collision between two
bodies often involves a strong, short range interactive force between
the bodies. When two extended bodies (such as automobiles or bil-
liard balls) come in contact, we idealize the collision and assume no
forces act except during the instant of contact. Taking the two bod-
ies as the entire system, these forces are internal forces. During an
ideal collision, no external forces are acting. Consequently, the total
momentum of the system is constant.
In a contact collision, there is a very short range repulsive force that
acts while the surfaces of the two bodies are touching.2 A glancing
collision, such as illustrated in Figure 6.2, occurs when the velocity
vectors of the two bodies are not aligned along the line of centers.
The distance b between the initial velocity vectors is called the impact
parameter.
A collision may not involve the actual physical contact of the two
bodies. The force exerted by one body on the other may be a long range
force such as the force of electric repulsion between an alpha particle
and the nucleus as in Rutherford’s experiment, or the force of gravity
when a comet in a hyperbolic orbit approaches from infinity, swings
2Theorigin of these forces is the repulsion between the electrons in one body
and the electrons in the other. When the two “electron clouds” start to overlap,
there is a repulsive Coulomb force between them. Fortunately, we do not need
detailed information about the force between the bodies because we can solve the
problem using the law of conservation of momentum in which internal forces play
no part.
 6.3. COLLISIONS 201

Figure 6.2. A glancing collision.

about the Sun, and travels back out to infinity. The same physics
applies regardless of the range of the forces.3
As a basic collision problem, consider the situation illustrated in
Figure 6.3 where body M1 with velocity V1 makes a glancing collision
with body M2 that is initially at rest. (We can always find a coordinate
system in which one body is initially at rest.) The two bodies move
off with velocities V10 and V20 at angles θ1 and θ2 relative to V1 , as
shown in the figure. To keep things simple, assume the bodies are not
rotating.
There are no external forces acting on the system so the law of
momentum conservation states that

Pi = Pf . (6.3)
That is, the initial momentum and the final momentum are equal. This
is a vector equation so it is equivalent to the three scalar equations
Pxi = Pxf , Pyi = Pyf , Pzi = Pzf ,
(if two vectors are equal, their components must be equal). It is con-
venient to place the origin of coordinates at the original position of
body M2 and to let the x-axis be defined by the direction of V1 . Select
the z-axis perpendicular to the plane of the motion, i.e., perpendicular
to the plane containing V10 and V20 . Then Pzf = 0. By momentum
conservation, Pzi = 0. Therefore the problem is two-dimensional; the
motion takes place entirely in the xy-plane.
3We often refer to colliding bodies as particles, even though they may be as-
tronomical objects. It is appropriate to use the term particle when we are dealing
with long range forces. In a glancing collision, the surfaces of two extended bodies
come in contact and they should not be called particles. Nevertheless, physicists
are a bit careless in the usage of this term, and in dealing with collisions the word
particle is often used when, strictly speaking, it should not be.
202 6. CONSERVATION OF LINEAR MOMENTUM

Figure 6.3. Illustration of the parameters involved

when two bodies undergo a glancing collision.

The momentum conservation equations in the xy-plane can be writ-

ten
0 0
Pxi = Pxf , or M1 V1x = M1 V1x + M2 V2x ,
0 0
Pyi = Pyf , or 0 = M1 V1y − M2 V2y .

In terms of the angles θ1 and θ2 these two equations are

M1 V1 = M1 V10 cos θ1 + M2 V20 cos θ2 , (6.4)

0 = M1 V10 sin θ1 − M2 V20 sin θ2 . (6.5)

Note the minus sign on the last term.

Conservation of momentum led to two equations. These equations
involve the seven parameters M1 , M2 , V1 , V10 , V20 , θ1 , θ2 . Obviously five
of these parameters must be “known” quantities so that you can solve
for two unknowns. In most collision problems you will know the speed
of the incoming particle, V1 . You will also probably know the masses
of the particles, or at least their ratio (which is sufficient). This leaves
you with four unknowns, namely the final speeds, V10 and V20 , and the
final directions, θ1 and θ2 . Two of these must be determined (perhaps
experimentally) before you can solve the problem for the other two.
If, however, the collision is elastic, then there is one more equation
you can use, namely the equation expressing the conservation of kinetic
energy. By definition, an elastic collision is one in which the kinetic
 6.3. COLLISIONS 203

energy is conserved.4 That is, Ti = Tf , or in terms of the problem,

1 1 1
M1 V12 = M1 V102 + M2 V202 . (6.6)
2 2 2
This condition gives you a third equation, allowing you to solve for
three unknowns. Frequently, the unknown quantities will be the final
velocities and the direction of one particle, for example, V10 , V20 , and θ2 .
The quantities you will need to know are (usually) the velocity of the
incoming particle and its deflection angle, as well as the masses.

Figure 6.4. Rutherford’s experiment. An alpha parti-

cle is scattered through the angle θ by a gold nucleus.
This collision experiment was crucial in determining the
structure of atoms.

To help you visualize a typical problem, Figure 6.4 illustrates Ruther-

ford’s famous experiment in which he bombarded the nuclei of gold
atoms with alpha particles emitted by a radioactive substance. The
gold nuclei were essentially at rest and the alpha particles approached
with a known velocity. After interacting with a gold nucleus the alpha
particles hit a screen painted with fluorescent material. This caused a
tiny flash of light to appear at the point where an alpha particle hit the
screen. The pinpoints of light were observed by eye by Rutherford’s
graduate students who tabulated the final positions of all the alpha
particles. In this problem the known quantities were M1 , M2 , V1 and
θ1 . The unknown quantities were V10 , V20 , and θ2 . (You will solve this
problem after studying central force problems; see Problem 10.20.)
4A collision is usually elastic if the two bodies do not come in contact, or if
neither body is deformed by the collision. Thus the interactions between celestial
bodies and between elementary particles are often elastic. Collisions of billiard balls
are frequently assumed to be elastic since the deformation is negligible.
204 6. CONSERVATION OF LINEAR MOMENTUM

Going back to the general case, I will now show you how to manip-
ulate momentum conservation and kinetic energy conservation, Equa-
tions (6.4), (6.5), and (6.6), to solve for the unknown quantities in
terms of the known quantities. Let me warn you that the algebra is a
bit tedious so you might want to get up and get a cup of coffee before
we start. Also, I am not going to give you all the intermediate steps,
so while you are up you had better get your pencil because you will
not understand the final result unless you can work out all the steps.
To begin, rewrite Equations (6.4) and (6.5), placing all the terms
with subscript 1 on one side. Thus:

M1 V1 − M1 V10 cos θ1 = M2 V20 cos θ2 , (6.7)

M1 V10 sin θ1 = M2 V20 sin θ2 . (6.8)
Next square both equations and add them together to eliminate θ2 .
You obtain
M12 V12 − 2M12 V1 V10 cos θ1 + M12 V102 = M22 V202 . (6.9)
This equation, together with the kinetic energy equation (6.6) give you
two equations in the two unknowns V10 and V20 . (The initial velocity V1
and the angle θ1 are assumed known.)
Now eliminate V20 from Equations (6.6) and (6.9). If you take the
expression for M22 V202 given by Equation (6.9) and plug it into Equation
(6.6), you obtain the following equation for V10 :
[(M1 + M2 )] V102 − [2M1 V1 cos θ1 ] V10 + (M1 − M2 )V12 = 0. (6.10)
 

All the terms in square brackets are known, so this is a quadratic

equation for V10 whose solution is
q
0
2M1 V1 cos θ1 ± (2M1 V1 cos θ1 )2 − 4 (M1 + M2 ) (M1 − M2 ) V12
V1 = ,
2 (M1 + M2 )
or
s 2
0 M 1 M1 V1 M1 − M2 2
V1 = V1 cos θ1 ± cos2 θ1 − V .
M1 + M2 M1 + M2 M1 + M2 1
Dividing through by V1 you obtain a nicer looking expression:
 s 
0
 2
V1 M1 cos θ1 ± cos2 θ1 − 1 + M2  .
= (6.11)
V1 M1 + M2 M1
 6.3. COLLISIONS 205

Let us interpret this result by discussing several special cases. Use

your experience with colliding bodies to visualize each situation.5 Case
1: Head-on Collisions: If two objects hit “head-on” then the motion
is one-dimensional and θ1 = θ2 = 0. The velocity vector of the incom-
ing particle, V1 , is pointed directly at the center of body M2 . Assume
M1 is moving from left to right. Recall from experience that if the
masses are equal, M1 stops and M2 moves off to the right. If M1 > M2
both masses move off to the right with M2 moving faster than M1 . If
M1 < M2 , then M1 bounces back (to the left) and M2 moves off to
the right. All of this information is incorporated in Equation (6.11).
I will now show you how to “read” Equation (6.11) to extract this
information.
Since we are imagining a head-on collision, the angle θ1 is zero.
Setting cos θ1 = 1 in equation(6.11), and doing a little bit of algebra,
leads to
M1 ± M2
V10 = V1 . (6.12)
M1 + M2
Subcase 1.1: M1 = M2
Consider first the situation in which M1 = M2 . Equation (6.12)
then reduces to
V10
 
1 1
= [1 ± 1] = .
V1 2 0
That is, V10 is either equal to V1 or it is equal to zero. From experi-
ence we expect V10 to be zero (the incoming object stops). What does
V10 = V1 correspond to? It corresponds to a miss, that is, no collision
at all. It tells us that the final velocity of M1 is equal to its initial
velocity. Furthermore, Equations (6.6) and (6.9) both tell us that for
a miss, V20 = 0. The condition θ1 = θ2 = 0 is equally satisfied by a
head-on collision or a miss. (Note how much physical information is
coded into the equations!)
We are not interested in the situation where one particle misses the
other, so let us set V10 = 0. What is V20 , the final velocity of the particle
that was initially at rest? Again, from experience we expect it to be
equal to V1 , the initial velocity of the incoming object. And, indeed,
Equation (6.7) for M1 = M2 and θ1 = 0 yields
V1 − V10 = V20
But V10 = 0, so,
V20 = V1 .
5You can make a rough check of the results we obtain by carrying out experi-
ments on a smooth surface with coins of the same or different masses.
206 6. CONSERVATION OF LINEAR MOMENTUM

This equality corresponds to our experience that the struck object flies
off with the velocity the incoming particle had before the collision.
Subcase 1.2 M1 6= M2
Now allow the two colliding bodies to have different masses, but
still assume a head-on collision so θ1 = θ2 = 0. Again, Equation (6.11)
reduces to Equation (6.12). Selecting the plus sign gives V10 = V1 ,
implying no collision at all. This case is of no interest, so select the
minus sign. That is,
M1 − M2
V10 = V1 . (6.13)
M1 + M2
Therefore, if M1 > M2 , the velocity V1 0 is positive. Plugging Equation
(6.13) into (6.7) and keeping in mind that θ1 = θ2 = 0, you get the
velocity acquired by M2 :
M1
V20 = 2V1 . (6.14)
M1 + M2
Equations (6.13) and (6.14) agree with your experience and intuition.
If a heavy (massive) object collides with a light (less massive) object,
the heavy object will keep on moving in its original direction. If a
billiard ball strikes a ping-pong ball, M1 >> M2 , and
. M1
V10 = V1 = V1 .
M1
That is, the billiard ball keeps on moving at essentially its original
velocity. What happens to the ping-pong ball? According to Equation
(6.14), if M1 >> M2 , then V20 = 2V1 . That is, the light object moves
off with a speed of twice the speed of the incoming heavy object.
It is easy to appreciate that if M1 < M2 , the situation is simply
reversed. If a light object hits a heavy object then M1 < M2 and V10 is
negative. That is, the light object bounces back. For example, if you
throw a ball against a wall, then M1 is the mass of the ball and M2
is the mass of the Earth (assuming the wall is attached to the Earth).
So M1 << M2 and according to Equation (6.13), V10 , the final velocity
of the ball, will be V10 = −V1 . The ball bounces back with its initial
speed. Similarly, V20 = 0. The Earth stands still.

Worked Example 6.3. A spacecraft flyby of a planet can

be used to give the spacecraft a “boost” and increase its velocity
by the process known as the “slingshot effect.” To underscore the
physics of this process, this example is unrealistic because we will
 6.3. COLLISIONS 207

assume a spacecraft approaches Mars, swings around the planet,

and heads off in the opposite direction. This would be (essentially)
a head-on collision. A realistic problem would have the spacecraft
approach the planet at some angle to its velocity vector and be
“scattered” at another angle.
In this problem we assume the spacecraft has a mass of 2000
kg and that Mars has a mass of 6.4×1023 kg. The initial speed
of the spacecraft is -12 km/sec and the initial speed of Mars is
+23.36 km/sec. (The speeds are relative to a coordinate system
at rest with respect to the Sun.) (a) Determine the final speed of
the spacecraft. (b) Evaluate the ratio of the final to initial kinetic
energy of the spacecraft.
Solution: As mentioned, this is (essentially) a two body head-
on collision. To analyze it we transform to a reference frame in
which one particle (Mars, in this case) is at rest. Then the initial
velocities are:
V2 = 0,
V1 = 23.36 + 12 = 35.36 km/s.
Equations (6.13) and (6.14) give the final velocities.
2000 − 6.4 × 1023 .
 
0 M1 − M2
V1 = V1 = 35.36 = −35.36 km/s.
M1 + M2 2000 + 6.4 × 1023
 
0 M1 2000 .
V2 = 2V1 = (2)(35.36) = 0.
M1 + M2 2000 + 6.4 × 1023
These are the speeds in a reference frame in which Mars is at rest.
Transforming back to the “inertial” reference frame we find the
speed of the spacecraft is
vs0 = V10 − vmars = −35.36 − 23.36 = −58.72 km/s.
The ratio of kinetic energies is
1
Tf ms vs2 (58.72)2
= 12 = = 23.9.
Ti m v 02
2 s s
(12)2
(Note: This problem is artificial because it assumes a one di-
mensional situation in which the spacecraft’s velocity vector and
the planet’s velocity vector are opposite to one another. Such a
situation could be set up but would require thrusts to adjust the
spacecraft’s speed.)
208 6. CONSERVATION OF LINEAR MOMENTUM

Exercise 6.6. Fill in the missing steps to obtain Equation (6.12)

from (6.11).
Exercise 6.7. (a) A ball of mass m traveling at speed v hits a ball
of mass 3m at rest. Determine the final velocities of the two balls. (b)
A ball of mass 3m traveling at speed v hits a ball of mass m at rest.
Determine the final velocities of the two balls. Assume head-on elastic
collisions. Answers: (a) -v/2 and v/2, (b) v/2 and 3v/2.

Case 2: Glancing Collisions: Let us now analyze a glancing

collision. The final directions of the two objects are given by nonzero
angles θ1 and θ2 , shown in Figure 6.3. As before, there are three possi-
bilities, M1 = M2 , M1 > M2 , and M1 < M2 . Once again the discussion
is based on Equation (6.11). Note that cos2 θ ranges from 1 to 0 so the
term under the radical varies from a minimum of (M2 /M1 )2 − 1 to a
maximum of (M2 /M1 )2 .
Subcase 2.1: M1 = M2
Let us begin by assuming M1 = M2 . Equation (6.11) then reduces
to
V10 1h p i 
0
= cos θ1 ± cos2 θ1 = . (6.15)
V1 2 cos θ1
Thus, the final velocity of M1 is either zero or V1 cos θ1 . The solution
V10 = 0 can be discarded because it implies a head-on collision, as
considered previously. Recall that V1 and θ1 are assumed known, so
the lower solution of Equation (6.15) yields the first of our unknown
quantities, the velocity of M1 after the collision:
V10 = V1 cos θ1 .
Plugging this into Equation (6.6) yields
V12 = (V1 cos θ1 )2 + V202 ,
or
2
(V20 ) = V12 (1 − cos2 θ1 ) = V12 sin2 θ1 , (6.16)
so
V20 = V1 sin θ1 .
This gives us the second unknown quantity, V20 . The third “unknown”
is θ2 and it can be obtained immediately from the conservation of mo-
mentum along the y−axis, Equation (6.5), as
V10
θ2 = sin−1 .
V1
 6.3. COLLISIONS 209

Worked Example 6.4. Prove that in any glancing elastic

collision between bodies of equal mass, the sum of the deflection
angles is θ1 + θ2 = π/2.
Solution: Given V10 = V1 cos θ1 and V20 = V1 sin θ1 . Plug into
the conservation of momentum equation V10 sin θ1 = V20 sin θ2 to get
V1 cos θ1 sin θ1 = V1 sin θ1 sin θ2
cos θ1 = sin θ2
But cos θ1 = sin π2 − θ1 , so θ2 = π2 − θ1, or

π
θ1 + θ2 = .
2

Exercise 6.8. Show that the null solution for Equation (6.15) im-
plies a head-on collision.
Exercise 6.9. Equation 6.16 leads to the two solutions V20 = ±V1 sin θ1 .
Using the facts that θ1 + θ2 = π/2 and V10 = V1 cos θ1 , show that only
the positive solution is obtained. (The negative solution corresponds
to θ1 = 0 and V10 = V1 , that is, a miss.)

Subcase 2.2: M1 > M2

If the mass of body 1 is greater than the mass of body 2, you expect
the heavier object (M1 ) to continue moving in the forward direction.
That is, you expect θ1 to be less than π/2. To appreciate that this is
true, note that the quantity under the radical in Equation (6.11) must
be positive. (Otherwise the velocity V10 would be a complex number
and this is not possible because physically measurable quantities must
be real.) The quantity under the radical is non-negative for M1 > M2
only if
2 M22
cos θ1 ≥ 1 − 2 .
M1
This means that the angle θ1 can range from zero to some maximum
value θmax given by
1
M22 2

−1
θmax = cos 1− 2 .
M1
210 6. CONSERVATION OF LINEAR MOMENTUM

In the limit M2 /M1 → 1 you obtain θmax = cos−1 (0) = π/2. In the
limit M2 /M1 → 0 you obtain θmax = cos−1 (1) = 0. Therefore,
π
0 < θ1 < .
2
This tells you that if the incoming object is the more massive body, it
will be scattered through an angle θ1 smaller than π/2.
Subcase 2.3: M1 < M2
The case of a glancing collision in which M1 < M2 can also be
analyzed with Equation (6.11). It is left as an exercise to show that if
the target M2 is much more massive than M1 , then the incoming body
will bounce off with its original speed (but with a change in direction).
Note that this case reduces to Subcase 2.2 if you interchange the two
bodies.
In conclusion, you have seen that an elastic collision between two
bodies can be analyzed using the conservation of linear momentum
and the conservation of kinetic energy. The equations obtained are
amazingly complex for such a simple problem. An important benefit
you should get from the analysis of collisions is an appreciation for how
to extract physical meaning from mathematical relations. Be aware
that in a collision the conservation of momentum always holds, but
conservation of kinetic energy may not.

Exercise 6.10. Show that if M1 << M2 the speed of M1 after the

collision is (essentially) the same as its speed before the collision.
Exercise 6.11. A billiard ball with velocity v strikes a second bil-
liard ball at rest with a glancing collision. The first ball is observed to
emerge from the collision at an angle of 20◦ . Determine the speed and
direction of the second ball.

6.4. Inelastic Collisions. The Coefficient of Restitution

In the previous section the collision between the two bodies was
elastic: There was no loss of kinetic energy. Of course, this is not
always the case. It is customary to represent the gain or loss of kinetic
energy by a quantity called the “Q value,” defined by
Q = Tf − Ti ,
where Tf is the total final kinetic energy and Ti is the total initial
kinetic energy.
 6.5. IMPULSE 211

For an elastic collision, Q = 0. This condition is generally not met;

most collisions are either endoergic in which kinetic energy is lost, or
exoergic in which kinetic energy is gained. For example, in the collision
of two putty balls having equal but opposite momenta, all of the kinetic
energy is lost. This collision is completely inelastic and Q is negative.
On the other hand, a collision between two molecules may involve an
exothermic chemical reaction in which chemical energy is transformed
into mechanical energy. For such a collision, Q is positive.
A closely related concept is the coefficient of restitution. This
was originally described by Isaac Newton who observed that for any
head-on collision of two non-rotating bodies the ratio of relative final
velocities to relative initial velocities is a constant. That is, if V1 and
V2 are the initial velocites and V10 and V20 are the final velocities, then
Newton’s Rule is
|V20 − V10 |
e= .
|V2 − V1 |
For an elastic collision, e = 1. This is easily demonstrated from
Equations (6.13) and (6.14). For a completely inelastic collision in
which all energy is lost, e = 0.
If the collision is a glancing collision, then the velocities to be used
in Newton’s formula are the velocity components along the line joining
the two bodies.6

Exercise 6.12. Prove that e = 0 for a completely inelastic collision

and e = 1 for an elastic collision. You may assume a head-on collision.
Exercise 6.13. Using Equations (6.13) and (6.14) show that Q = 0
for a head-on elastic collision. (This is obviously true from the defi-
nition of elastic collision; the purpose of the exercise is to give you
experience in manipulating the relations.)

6.5. Impulse
When a bat hits a baseball, a large force acts for a short period of
time. Such a blow gives rise to an impulse. By definition, an impulse
6Thecoefficient of restitution actually depends on various other factors, such
as the medium in which the collision occurs, but Newton’s formula is a good
approximation.
212 6. CONSERVATION OF LINEAR MOMENTUM

is the time integral of a force. Denoting the impulse by J, we can write

Z τ
J= Fdt,
0
where τ is the time during which the force acts. In general, it would be
difficult to evaluate the integral on the right because the force is usually
an unknown, and probably complicated, function of time. Nevertheless,
it is easy to evaluate J because from Newton’s second law, F = dp dt
, so
Z τ Z τ Z pf
dp
J= Fdt = dt = dp = pf − pi = ∆p.
0 0 dt pi
That is, the impulse is simply equal to the change in momentum.

Exercise 6.14. A force F = 3 sin 5t N acts on a particle of mass

2 kg that was initially at rest. The force acts during the time interval
from t = 0 to t = π/10 s. What is the final velocity of the particle?
Answer: 0.3 m/s.
Exercise 6.15. A stationary block sitting on a frictionless surface
is acted upon by a force (in newtons) given by
F = 2t for 0 ≤ t ≤ 2,
F = 4 for 2 ≤ t ≤ 5,
F = −t for 5 ≤ t ≤ 7.
(Times in seconds.) Determine the final momentum of the block.
Answer: 4 kg m/s.

6.6. Momentum of a System of Particles

In this book we have claimed more than once that internal forces
do not affect the total linear momentum of a mechanical system. For
example if a bomb explodes, pieces fly off in all directions, but the total
momentum is unchanged. We now prove the claim by considering a sys-
tem of N particles and determining the effect of internal and external
forces on the momentum. The particles have masses m1 , m2 , · · · , mN
and are located at positions r1 , r2 , · · · , rN .
All of the particles exert forces on each other. These forces are
internal forces. Denote by Fij the force exerted on particle i by particle
j.
Fij = force acting on particle i, due to particle j.
 6.6. MOMENTUM OF A SYSTEM OF PARTICLES 213

Newton’s second law, applied to particle i is, then,

dpi (e)
X
= Fi + Fij ,
dt j=1,N
j6=i

(e)
where Fi is the external force acting on i. Note that the summation
over the internal forces is subject to the condition j 6= i because a
particle cannot exert a force on itself.
There is one such equation for each particle. Adding all N equations
yields
N N N N
X dpi dPtot X (e) X X
= = Fi + Fij ,
i
dt dt i i j6=i
P
where Ptot = pi is the total momentum of the system. The last term
(the double sum) is zero because by Newton’s third law, Fij = −Fji .
By writing out a few terms you will see that the double sum consists
of pairs of terms that cancel each other out. Therefore,
dPtot X (e) (e)
= Fi = Ftot (6.17)
dt i
(e)
where Ftot is the total (or net, or resultant) external force acting on the
system. Thus, we have shown that the internal forces have no effect on
the total momentum.
This result has another important consequence. If the masses of
the particles are constant, we can write
dPtot X
= mi r̈i ,
dt i

and Equation (6.17) can be written

X (e)
mi r̈i = Ftot .
i

The definition of center of mass (rc ) for a system of particles having

total mass M is (see Equation 5.17)
X
M rc = mi ri .
i

Differentiating twice with respect to time gives

X
M r̈c = mi r̈i . (6.18)
i
214 6. CONSERVATION OF LINEAR MOMENTUM
P (e)
But i mi r̈i = Ftot so
(e)
M r̈c = Ftot . (6.19)
This important results states that the center of mass of a system of
particles moves like a particle of mass M acted upon by the resultant
of the sum of all the external forces, regardless of their point of ap-
plication. A corollary is that if a system is not acted upon by a net
external force, the center of mass will move with constant velocity.

m1
r2-r1
r1
m2
r2
O

Figure 6.5. Relative positions of two particles. Parti-

cles m1 and m2 are located at r1 and r2 with respect to
the origin. The relative coordinate r gives the position
of m2 with respect to m1 .

6.7. Relative Motion and the Reduced Mass

When we study the motion of two interacting bodies such as a star
and a planet, or an electron and a nucleus, we are often interested in
their relative motion and do not care about their motion with respect
to an inertial reference frame. For such problems it is convenient to
introduce the concepts of relative coordinate and reduced mass. The
introduction of these quantities allows us to replace the two-body prob-
lem (with two equations of motion) by a single one-body problem (and
only one equation of motion).7
Consider a system consisting of two particles (m1 and m2 ) that are
exerting equal and opposite forces on each other. Assume no external
forces are acting. Let F be the force on m2 due to m1 . Then the force
on m1 is −F. Consequently, the equations of motion of the two particles
are:
m1 r̈1 = −F,
7This
is a very important simplification allowing us to obtain (for example)
the motion of a planet relative to the Sun. There is no such simplification for a
system of three bodies, although physicists have been trying to solve the “three
body problem” for hundreds of years. Some mathematicians claim the problem is
actually unsolvable.
 6.7. RELATIVE MOTION AND THE REDUCED MASS 215

and
m2 r̈2 = +F.
Figure 6.5 shows two particles, m1 and m2 , at positions r1 and r2
with respect to the inertial origin O. The “relative” vector r gives the
position of m2 with respect to m1 and (by tip-to-tail addition) is given
by
r = r2 − r1 .
Differentiating the relative coordinate twice with respect to time yields
r̈ = r̈2 − r̈1 .
Substituting for r̈2 and r̈1 from the equations of motion gives
 
F F m1 + m2
r̈ = + = F
m2 m1 m1 m2
or  
m1 m2
r̈ = F.
m1 + m2
The quantity mm11+m
m2
2
is called the “reduced mass.” It is denoted by µ.
So the equation of motion for the relative coordinate is
µr̈ = F. (6.20)
As an example, consider a system composed of a star and a planet.
The star is much more massive than the planet and for all intents
and purposes the star remains at rest and the planet orbits around it.
If the mass of the star is m1 and the mass of the planet is m2 and
if m1 >> m2 , then µ ∼ = m2 . On the other hand, for a binary star
system both masses may be approximately equal and the two stars
orbit around their common center of mass. If both stars have the same
mass, say m, the reduced mass is µ = 21 m.
Consider the Sun-Earth system. The relative coordinate gives the
distance from the Sun to the Earth. F is the force the Sun exerts on the
Earth. In an analysis of this system you should write µr̈ = F and not
m2 r̈ = F. The reason is that the Sun is accelerating, so a coordinate
system with origin at the Sun is not an inertial coordinate system
and Newton’s second law does not hold. (Actually for the Sun and
Earth, m2 and µ are so nearly equal that the error in using m2 r̈ = F
is negligible.)

Exercise 6.16. Where is the center of mass of the Sun-Earth sys-

tem? (Look up the necessary values.) Answer: 4.5 × 105 meters from
center of the Sun.
216 6. CONSERVATION OF LINEAR MOMENTUM

Exercise 6.17. Determine the reduced mass of the Sun-Jupiter

system and the reduced mass of the Earth-Moon system. Answer: For
Earth-Moon, µ = 7.26 × 1022 kg.

6.8. Collisions in Center of Mass Coordinates (Optional)

When studying two body collisions, physicists frequently use a co-
ordinate system that is moving with the center of mass because it is
often an easier and quicker way to solve the problem. This approach
is particularly useful when studying collisions between elementary par-
ticles. If you specialize in high energy physics, you will become very
familiar with center-of-mass coordinates.
Let me remind you that if there are no external forces acting on
the system, the center of mass moves with constant velocity (Equation
6.19). In that case, a coordinate system moving with the center of mass
is an inertial (i.e., non-accelerating) coordinate system.

m1 v'1cm
v1cm ϕ v2cm
m1 ϕ m2
m2
v'2cm
Figure 6.6. A collision as seen in the center-of-mass
coordinate system.

From the center-of-mass point of view, in a two body collision both

bodies are moving, approaching one another and the center of mass.
They collide at the center of mass. After the collision, they recede from
the center of mass.
In the center-of-mass frame, the velocities of the two particles before
the collision are denoted v1cm and v2cm . After the collision the particles
0 0
recede in opposite directions with velocities v1cm and v2cm . The angle
between the incoming and outgoing directions is denoted φ, as shown
 6.8. COLLISIONS IN CENTER OF MASS COORDINATES (OPTIONAL) 217

in Figure 6.6. Note that there is now a single angle, so we have already
achieved some simplification of the problem.
Figure 6.7 shows the positions of the two particles and their center
of mass relative to the origin O of an arbitrary inertial frame. The
center of mass (indicated by the small open circle) lies on the line
joining the two particles and is at rc relative to O. The coordinates
r1cm and r2cm give the positions of the particles relative to the center
of mass. Note that
r1cm = r1 −rc
r2cm = r2 −rc

Figure 6.7. The center-of-mass coordinates. r1cm and

r2cm give the positions of m1 and m2 with respect to the
center of mass. The open circle specifies the position of
the center of mass.

The total initial momentum in the center of mass system is zero.

To prove this, take the time derivative of the definition of the center of
mass, rc = (m1 r1 + m2 r2 )/M, where M = m1 + m2 :
M ṙc = (m1 ṙ1 + m2 ṙ2 ) = (m1 ṙ1cm + m1 ṙc + m2 ṙ2cm + m2 ṙc ),
so,
M ṙc − (m1 + m2 )ṙc = m1 ṙ1cm + m2 ṙ2cm ,
0 = m1 v1cm + m2 v2cm .
Therefore, the total initial momentum in the center of mass system is
p1cm + p2cm = 0. (6.21)
The total final momentum must also be zero,
p01cm + p02cm = 0. (6.22)
218 6. CONSERVATION OF LINEAR MOMENTUM

The energy equation in the cm system is

p21cm p22cm p02

1cm p02
+ = + 2cm + Q, (6.23)
2m1 2m2 2m1 2m2

where the factor Q represents any energy gained or lost during the
collision. For the rest of this section I will assume elastic collisions so
Q = 0.
Let us go back to the basic collision of Section 6.3 in which a particle
of mass m1 and speed V1 strikes a particle of mass m2 at rest, but now
we analyze the problem in the center of mass coordinate system. I will
use capital letters (such as V1lab ) for speeds in the laboratory frame,
and small letters (such as v1cm ) for speeds in the center of mass frame.
The speed of the center of mass in the lab frame is denoted Vc and
is obtained by taking the time derivative of the definition of center of
mass,
m1 r1 + m2 r2
rc = ,
M
m1 ṙ1 + m2 ṙ2
∴ Vc = .
M
Since ṙ1 = V1lab and ṙ2 = V2lab = 0 the velocity of the center of mass
is
m1
Vc = V1lab . (6.24)
M
In the laboratory system the origin of coordinates is located at
the initial position of the particle at rest. After the collision the two
particles move off at angles θ1 and θ2 . In other words, the laboratory
frame is described by Figure 6.3.
The transformation between laboratory frame velocities and center
of mass velocities is illustrated in Figure 6.8, which shows the rela-
0 0
tionship between v1cm and V1lab , as seen in the laboratory frame of
reference. In constructing the figure I used the vector relationship

V0 1lab = v0 1cm + Vc . (6.25)

This equation tells you that to convert velocities from the cm frame to
the lab frame you simply add the velocity of the center of mass.
We now obtain some useful relationships between the velocities in
the two coordinate systems. The velocity of the center of mass in the
lab system is given by Equation (6.24). Then, according to Figure 6.8
 6.8. COLLISIONS IN CENTER OF MASS COORDINATES (OPTIONAL) 219

Vc
v'1cm
V'1lab
ϕ
θ1
direction of V1lab
Figure 6.8. Relationship between velocities in the cen-
ter of mass system and the laboratory system for the
final velocities of particle number 1.

and Equation (6.25),

v1cm = V1lab − Vc , (6.26)
 m1 
= V1lab 1 − ,
M
= V1lab (m2 /M ),
and similarly for the other velocities.
Looking at Figure 6.8 you can see that the velocity components are
given by
0 0
V1lab sin θ1 = v1cm sin φ,
and
0 0
V1lab cos θ1 = v1cm cos φ + Vc .
These two equations allow you to determine the relationship between
θ1 and φ. Dividing one equation by the other yields
0
v1cm sin φ sin φ
tan θ1 = 0
= 0
. (6.27)
v1cm cos φ + Vc cos φ + (Vc /v1cm )
Although this gives an expression for θ1 in terms of φ, it is not in a very
convenient form. As you will see in a moment, it is possible to express
θ1 in terms of φ and the masses of the particles. To do so, begin by
0
expressing both Vc and v1cm in terms of the relative velocity, vrel :
vrel = V2lab −V1lab = V02lab −V01lab .
(This equation implies that the relative velocity is constant. However,
that is only true for an elastic collision.)
Since V2lab = 0, you can write
vrel = −V1lab .
220 6. CONSERVATION OF LINEAR MOMENTUM

Use Equation (6.24) to form the ratio

m1
Vc V1lab m1
= M =− ,
vrel −V1lab M
so  m  m
1 1
Vc = |Vc | = − vrel  = vrel .
 
M M
0
Similarly, you can obtain an expression for v1cm as follows:
m m2 0 
1
0
v1cm 0
= V1lab −Vc = V01lab − 0
V1lab + V ,
M M 2lab
0
 m1  m2 0 m2 0 0
= V1lab 1− − V2lab = (V1lab − V2lab ),
M M M
m2
= (−vrel ) ,
M
and consequently,
0 m2
v1cm = vrel .
M
0
Having obtained expressions for Vc and v1cm you can write Equation
(6.27) as
sin φ sin φ
tan θ1 = (m1 /M )vrel
= . (6.28)
cos φ + ( ) cos φ + (m1 /m2 )
(m2 /M )vrel

It is left as a problem to show that

sin φ
tan θ2 = . (6.29)
1 − cos φ
If V2lab = 0 and V1lab and θ1 are known lab frame variables, you can
determine the center of mass coordinate φ from (6.28). Then Equation
(6.29) yields θ2 . The value of v1cm is obtained from Equation (6.26).
0 0
Since vrel = −V1lab you can determine v1cm from v1cm = (m2 /M )vrel .
0 0
Finally, v2cm = −(m1 /m2 )v1cm . If desired, you can then transform back
to laboratory coordinates.

Worked Example 6.5. Using the center-of-mass coordinate

system, show that in an elastic collision between two particles of
equal mass, the scattering angles in the laboratory system add up
to π/2. (That is, show that θ2 + θ1 = π/2.)
Solution: If m1 = m2 , Equation (6.28) can be written as
sin φ φ
tan θ1 = = tan .
cos φ + 1 2
 6.9. SUMMARY 221

(The last step requires the “half angle” trigonometric identity.)

This result indicates that φ/2 = θ1 . Similarly, Equation 6.29 leads
to
sin φ φ
tan θ2 = = cot .
1 − cos φ 2
Combining these relations gives
tan θ2 = cot θ1 .
Another trigonometric identity is
tan θ1 + tan θ2
tan(θ1 + θ2 ) =
1 − tan θ1 tan θ2
so
tan θ1 + tan θ2 tan θ1 + tan θ2
tan(θ1 + θ2 ) = = → ∞.
1 − tan θ1 cot θ1 0
Therefore,

θ2 + θ1 = π/2.
This result indicates that if the two masses are equal, the angle
between the two outgoing particles in the lab system is a right
angle.

Exercise 6.18. Use the center-of-mass system to show that if

m1 << m2 the scattering angle in the cm system is (almost) equal
to the scattering angle in the lab system. That is, show that θ1 ≈ φ.

6.9. Summary
The law of conservation of linear momentum is applicable to nu-
merous problems. In this chapter we have been particularly interested
in two problems: the motion of a rocket and the collision of two masses.
Conservation of momentum is based on Newton’s second law. Since
F =dp/dt, if the net external force is zero, the momentum is constant.
A system such as a rocket (or a conveyor belt) whose mass is chang-
ing with time but which is not acted upon by external forces, is ana-
lyzed by requiring that the initial and final momenta be equal. For the
rocket, this leads to
dV dM
M = −u ,
dt dt
222 6. CONSERVATION OF LINEAR MOMENTUM

or,
dM
dV = −u ,
M
where u is the speed of the ejected gases relative to the rocket.
The analysis of a glancing collision of two objects is also based on
the conservation of linear momentum. If the masses are given, the
problem can be expressed in terms of the parameters V1 , V10 , V20 , θ1 , θ2 .
Momentum conservation yields two equations. Thus you need to be
given three of these parameters. In the case of elastic collisions, con-
servation of kinetic energy gives you an additional equation and you
only need to be given two of the parameters.
Assuming an elastic collision and that V1 and θ1 are given, the
conservation laws lead to an equation for V10 , the final velocity of M1 .
 s 
0
 2
V1 M1 cos θ1 ± cos2 θ1 − 1 + M2  .
=
V1 (M1 + M2 ) M1

The final velocity of M2 is given by

V202 = [M12 V12 − 2M12 V1 V10 cos θ1 + M12 V102 ]/M22 .
Finally, θ2 can be determined from Equation (6.8),
M2 V20
sin θ1 = sin θ2 .
M1 V10
A sharp blow involves a force F that acts during a short time in-
terval τ. The impulse J of such a force is the time integral of the force
and is equal to the change in momentum. Thus,
Z τ
J= Fdt = ∆p = pf −pi .
0
A system of particles is acted upon by internal and
Pexternal forces.
The center of mass moves like a particle of mass M = mi acted upon
by the total external force:
(e)
M r̈c = Ftot .
When studying the relative motion of two particles, it is convenient
to introduce the reduced mass µ, given by
m1 m2
µ= .
m1 + m2
Then, the equation of motion for the relative coordinate r is F =µr̈.
Collisions are frequently studied in the center-of-mass coordinate
system in which the total initial and final momenta are zero. For
 6.10. PROBLEMS 223

elastic collisions, the scattering angles in the lab and cm systems are
related by
sin φ
tan θ1 = 0
,
cos φ + (Vc /v1cm )
sin φ
tan θ2 = ,
1 − cos φ
where θ1 and θ2 are the scattering angles in the laboratory coordinate
system and φ is the (single) scattering angle in the center-of-mass sys-
tem.

6.10. Problems
Problem 6.1. A 90 kg railroad worker is on a handcar of mass 200
kg. The handcar is moving at 5 m/s when it passes under a tree. (a)
The railroad worker leaps upwards, grabs a limb and hangs on. Does
the speed of the handcar change? If so, determine its final velocity. (b)
Now consider the converse problem. The empty handcar is moving at
5 m/s when it passes under a tree and a 90 kg railroad worker drops
out of the tree onto the handcar. In this case, does the speed of the
handcar change? If so, determine its final velocity.
Problem 6.2. Prove that in a one dimensional elastic collision
between two bodies, the relative velocity between the bodies has the
same magnitude before and after the collision, but the opposite sign.
Problem 6.3. (The ballistic pendulum.) A ballistic pendulum
can be used to determine the speed of the bullet fired from a rifle by
determining its effect when it hits and is embedded in a pendulum.
Consider a ballistic pendulum that consists of a suspended block of
wood of mass M . A bullet of mass m and initial velocity v is fired
into and becomes embedded in the block. To block swings upward a
height h. (See Figure 6.9) Derive an equation for v in terms of the given
quantities.

Problem 6.4. A raindrop is falling through fog and is picking up

tiny water droplets as it falls. (a) Justify that the rate of change of
mass of the raindrop is proportional to r2 v, where r is the radius of the
raindrop and v is its downward velocity. (b) Prove that the acceleration
of the drop is g/7. (Ignore air resistance.)
Problem 6.5. A box filled with sand is placed on a sled and slides
down an ice covered hill (so that friction is negligible). Sand is leaking
224 6. CONSERVATION OF LINEAR MOMENTUM

Figure 6.9. The ballistic pendulum.

out of a hole in the box at a constant rate. The slope of the hill is α.
(a) Show that the equation of motion of the box (plus sled) is just
dv
m = mg sin α.
dt
(b) Now assume that the sand is somehow thrown out of the box with
a velocity −v, that is a velocity in the direction opposite to the motion
of the box but with the same speed as the box. Show that the equation
of motion in this case is
dv dm
m =v + mg sin α.
dt dt
Problem 6.6. A jet boat (or jet ski) operates on the following
principle: Water is drawn through an inlet into a turbine and pumped
out at high speed through a smaller opening. The manufacturer of a
jet boat states that the turbine draws 50 gallons of water per second
and expels it at a pressure of 80 psi. The manufacturer states that the
thrust developed is over 2000 lbs. Determine whether or not this is a
realistic value. (Hint: Look up Bernoulli’s equation.)
Problem 6.7. A spherical asteroid of mass m0 is moving freely
in interstellar space with velocity v0 . It runs into a dust cloud whose
uniform density is ρd . Assume that every particle of dust that hits the
asteroid sticks to it. (a) Obtain an expression for the velocity of the
asteroid as a function of time. (b) Obtain an expression for the force
exerted on the asteroid by the dust as a function of time. Answer (a)
h i−3/4
4/3
v = v0 m0 m0 + 4kv 3K
0 m0
2/3 t where k = ρd π and K = m/r3 .

Problem 6.8. During a war being waged in Antarctica, an armored

car of mass 2000 kg with a machine gun mounted on its roof drives onto
a frozen lake at 30 km/hr. The ice is, of course, perfectly frictionless,
and the truck will continue to slide in a straight line at a constant
velocity directly into the enemy camp unless it can make a 90◦ turn
 6.10. PROBLEMS 225

and slide to safety. G. I. Joe (who studied physics in college) jumps

to the roof, swivels the machine gun, and begins firing in a direction
perpendicular to the motion. The bullets have a mass of 500 grams
each and leave the gun with a velocity of 800 m/s. The gun fires at
a rate of 200 bullets per minute. How long must G. I. Joe fire the
machine gun for the car’s motion to be deviated by 90◦ ? Since the
bullets are being fired so rapidly, you can assume the mass decrease is
continuous. Answer: 17.55 sec.
Problem 6.9. The man on the flying trapeze is hanging by his
knees from the cross bar. The woman trapeze artist stands on the
circus floor. The trapeze starts at an angle of 60◦ from the vertical.
At the bottom of the swing, the man grabs the woman and they both
swing upward. To what angle will the trapeze swing with both artists
on it? For simplicity, assume the man is a point mass mM and the
woman is a point mass mW . The rope has length l and negligible mass.
Problem 6.10. A rocket motor is undergoing a “bench test.” It is
attached to a fixed support by four large springs of constant 106 N/m.
The motor burns fuel at a rate of 50 kg/s. When the motor is running
the springs are observed to stretch 1.5 cm. Determine the exhaust
speed of the burned fuel.
Problem 6.11. Most airports have a conveyer belt behind the
check-in counter for luggage to be transported to the airplane. Bags
of mass m are dropped onto the belt a rate of k per second. (You can
assume the mass increase is constant.) What is the increased power
load during the time the bags are being placed on the belt? Show that
the extra power required is twice the rate of increase of kinetic energy.
Explain what is happening to the “missing” power.
Problem 6.12. The “people mover” at the San Francisco airport
is essentially a horizontal, very long conveyor belt running at speed
v. Suppose that initially there are no people on the belt. The motor
driving the belt is drawing W watts of electrical power. A flight arrives
and passengers (all having the same mass) step onto the belt, one after
another at one second intervals, so that the mass being carried increases
at a constant rate k kg/sec. Assume the people stepping onto the belt
had an initial speed of v/2. If the belt is to continue running at the
same speed, how much extra electrical power must be supplied to the
motor?
Problem 6.13. A physics student is holding a vertical hanging
chain by its top link. The bottom link is just touching the top surface
226 6. CONSERVATION OF LINEAR MOMENTUM

of a scale. The student lets go of the chain and observes the reading
on the scale while the chain is dropping. The student claims that
the reading on the scale is three times the weight of the length of
the chain on the scale. The Laboratory Instructor doubts the result
obtained by the student. Show that the student’s observation is correct.
(The reading of the scale is the force exerted by the scale to stop the
downward motion of the chain.)
Problem 6.14. Many years ago on a cold winter morning in Chicago,
Bonnie and Clyde stole an armored truck full of money. (The mass of
the truck was 2000 kg and its top speed was 240 km/hr or 66.6 m/s.)
Officer Dick Tracy and his driver spotted them and gave chase. (The
police car had a mass of 1500 kg and its top speed was also 240 km/hr.)
As luck would have it, the two vehicles ran off the bank onto Lake
Michigan which was covered with perfectly frictionless ice, so the two
vehicles continued to move at constant speed and maintained a con-
stant separation. Dick Tracy grew impatient, so he opened the moon
roof, stood up, and started to shoot at the armored truck with his
machine gun. The machine gun fired 120 bullets per minute with a
muzzle speed of 1000 m/s. Each bullet had a mass of 0.05 kg. All of
the bullets hit and were embedded into the armored truck. After one
minute of this, what was the speed of the police car and what was the
speed of the truck? Answer: 62.6 m/s, 69.4 m/s.
Problem 6.15. An Atwood’s machine uses two containers filled
with water on either side of the ideal, frictionless pulley. Initially,
both buckets contain the same amount of water and weigh the same.
However, one of the containers has a small hole in it, and water is
leaking out at a rate k (kg/s). The leaking container will, therefore,
move upward. Obtain an expression for the velocity of this container.
Answer: v = −gt+(2mg/k) ln [(2m)/(2m − kt)] , where m is the initial
mass of the buckets plus water.
Problem 6.16. A rocket of mass 40,000 kg is in empty space.
Determine its velocity increase after burning all its fuel if the mass of
the fuel is 90% of the mass of the rocket. The rate of fuel burn is
constant. The speed of the exhaust gas relative to the rocket is 3000
m/s.
Problem 6.17. A rocket of mass 1500 kg is designed to expel
exhaust gas at 1000 m/s. Determine the minimum required burn rate
if the rocket is to rise from the surface of the Earth. What burn rate
is required for it to have an initial acceleration of 1 m/s2 ?
 6.10. PROBLEMS 227

Problem 6.18. A small rocket is launched from the surface of the

Earth. Since it does not rise very high, we are justified in assuming g =
const. Obtain an equation for the height reached when all of the fuel
is burned. The mass of the fuel is m and the initial mass of rocket plus
fuel is M0 . The exhaust speed of the gases is u. Assume the exhaust
speed and the burn rate are constant. Ignore air resistance.
Problem 6.19. Johnny Whizz, the inventor, designs an automobile
that will just hover at the surface of the Earth. It has four rocket
motors, one in each wheel well. The rocket motors expel burned fuel
with an exhaust velocity of 1000 m/s. Assume the mass of the fuel is
80% of the total mass of the car. Evaluate the maximum time the car
can hover above the ground.
Problem 6.20. Rockets are not always launched straight up. Con-
sider a rocket launcher consisting of a ramp inclined at 30◦ above the
horizontal. The mass of the rocket is 4000 kg of which 3000 kg are fuel.
The exhaust speed is 1000 m/s and the fuel is burned at a constant
rate of 200 kg/s. You may assume a flat, airless, nonrotating Earth.
(a) Determine the velocity and direction of the rocket at the time all
the fuel is burned. (b) Determine its position at this time.
Problem 6.21. A billiard ball is placed in contact with the upper
surface of a bowling ball and they are dropped from a height h onto a
cement floor. (Since they fall at the same rate, they are essentially in
contact but you might like to think of the billiard ball as lagging the
bowling ball by an infinitesimal amount.) The bowling ball hits the
ground and bounces back up, immediately striking the billiard ball.
How high does the billiard ball rise? You may assume the bowling
ball has a mass 20 times greater than a billiard ball. All collisions are
elastic.
Problem 6.22. A particle of mass 5 kg and initial speed 5 m/s
undergoes a head-on elastic collision with a particle of mass 3 kg with
initial speed −3 m/s where the negative sign indicates that it is ap-
proaching the first particle. Determine the final speeds of the two
particles.
Problem 6.23. An air hockey puck of mass 2M collides in a glanc-
ing collision with a puck of mass M. The heavier puck had an initial
velocity of 3 m/s and comes off at an angle of 30o after the collision.
Determine the velocity and direction of the lighter puck. Assume an
elastic collision.
Problem 6.24. Consider a glancing collision between two objects
of nearly equal mass (M2 = M1 +δ where δ is a small quantity). Obtain
228 6. CONSERVATION OF LINEAR MOMENTUM

an expression for V10 /V1 and show that it reduces to 1 or cos θ as δ → 0,

as in Subcase 2.1. What is the value of V10 /V1 in the extreme of δ being
a large value (that is, M2 >> M1 )?
Problem 6.25. When a spacecraft carries out a flyby maneuver
near a planet, the speed of the spacecraft can be dramatically increased.
The Cassini spacecraft approached Jupiter at a speed of 9.36 km/sec as
measured relative to the Sun. Jupiter’s orbital speed is 13.02 km/sec.
As it approached the planet, the spacecraft velocity formed an angle of
about 37o to Jupiter’s velocity vector. Cassini was deflected through
an angle of 52.8o . Determine the increase in speed of the spacecraft due
to this slingshot maneuver.
Problem 6.26. At time t = 0 three particles of masses m1 = 1
g, m2 = 2 g and m3 = 3 g are at rest and lined up along the x-axis at
positions x1 = −1 cm, x2 = 0 cm, and x3 = +2 cm. The forces acting
on the particles are F1 =-3̂ dynes, F2 = 0, and F3 =+12̂ dynes. (a)
Determine the position of the center of mass at t = 0. (b) Determine the
positions of the three particles at t = 10 s. (c) Determine the position
of the center of mass at t = 10 s from Equation (5.17). (d) Determine
the position of the center of mass at t = 10 s by using Equation (6.19).
Problem 6.27. Two masses, m1 and m2 , undergo a completely
inelastic collision. Show that the loss of kinetic energy is equal to 12 µv 2
where µ is the reduced mass and v is the relative velocity of the two
masses.
Problem 6.28. Consider a planet-star system with masses MS and
MP . The planet is located at a distance r from the star. As seen from
an inertial coordinate system, the star is located at RS . (a) Determine
the acceleration (in inertial space)of the star and the planet due to
their mutual gravitational attraction. (b) Determine the acceleration
of the planet with respect to the star. (Note that it is not equal to
−(GMS /r2 )r̂ but it reduces to this value if MS >> MP .)
Problem 6.29. A ball is dropped from a height h. The coefficient
of restitution is e. Determine that the time required for the ball to come
to rest is s  
2h 1 + e
t= ,
g 1−e
and that the total distance the ball travels is
1 + e2
 
d=h .
1 − e2
 6.10. PROBLEMS 229

Hint: Express the time as an infinite series. Note that the sum of a
geometric series of the form a, ar, ar2 , ar3 , · · · is S = a/(1 − r).
Problems 6.30 to 6.32 are based on Optional Section 6.8
Problem 6.30. Show that in the center of mass system, the con-
servation of kinetic energy (equation 6.23) reduces to
2 0
P1cm P2
= 1cm
2µ 2µ
if Q = 0.
Problem 6.31. Show that in an elastic collision between two par-
ticles the relative velocity does not change.
Problem 6.32. Show that the scattering angle φ in center of mass
coordinates is related to the angle θ2 in the lab coordinates by
sin φ
tan θ2 = .
1 − cos φ
(In other words, derive Equation 6.29.)
Computational Projects
Computational Techniques: The Runge-Kutta Method
In Section 3.7 I described the Euler Cromer algorithm for solving
ordinary differential equations (ODE’s). Here I will describe the Runge-
Kutta method which is more accurate and perhaps more elegant, but
less transparent. It is based on the truncated Taylor Series expansion
for a function g(t):

dg 
g(t + τ ) = g(t) + τ
dt  ξ

The second term is usually evaluated at t, but to make the solu-

tion more accurate, the Runge-Kutta technique evaluates it half way
through the time step. That is ξ = t + τ /2.
An economical way of expressing the ODE is to define two vectors,
x and f in the following way. Assume a two dimensional system so the
position is given by x and y and the velocity by vx and vy . Then
x(t) = [x(t) y(t) vx (t) vy (t)]
f (x,t) = [vx (t) vy (t) ax (t) ay (t)]
and the ODE can be written
dx
= f (x(t), t).
dt
230 6. CONSERVATION OF LINEAR MOMENTUM

The second-order Runge-Kutta algorithm is obtained by first defining

a new vector x∗ by:
1
x∗ = x∗ (t + τ /2) = x(t) + τ f (x(t), t).
2
Then
x(t + τ ) = τ f (x∗ , t + τ /2).
It is not too difficult to appreciate that this is equivalent to the trun-
cated Taylor Series. However, the most common ODE solver is not the
second-order Runge-Kutta, but the fourth-order Runge-Kutta, which
is also based on the Taylor Series but in a much less transparent way.
In vector form, the value of x at time t + τ is given by
1
x(t + τ ) = x(t) + τ (F1 + 2F2 + 2F3 + F4 ) ,
6
where
F1 = f (x, t),
1 1
F2 = f (x + τ F1 , t + τ ),
2 2
1 1
F3 = f (x + τ F2 , t + τ ),
2 2
1
F4 = f (x + τ F3 , t + τ ).
2
Computational Project 6.1. A man is standing on a stationary
railroad flatcar loaded with large rocks, all having the same mass of 10
kg. He is able to throw the rocks at a speed of 2 m/s, relative to himself.
There are 50 rocks on the flatcar. Determine the speed of the flatcar
after the man has thrown all of them straight back off the end of the
car. The mass of the man plus empty flatcar is 1000 kg. Naturally, this
advanced propulsion system depends on the development of completely
frictionless railroad cars. Solve this problem numerically, and also solve
it using equation 6.2. Compare your answers. Why do they not agree?
Which answer do you think is correct?
Computational Project 6.2. Write a program to determine
the altitude reached by a rocket launched from the surface of Earth.
Assume the rate at which fuel is burned is proportional to the amount
of fuel left and the exhaust speed of the ejected gases is constant. Plot
the position of the rocket as a function of time for dM/dt = −0.01M if
the initial mass of the fuel is 50,000 kg and u = 2500 m/s. The mass of
the empty rocket is 10,000 kg. Ignore air resistance, but keep in mind
that the gravitational force decreases with the distance from the center
of Earth.
 6.10. PROBLEMS 231

Computational Project 6.3. A rocket is launched from the

surface of the Earth. The initial mass of the rocket is 1000 kg, and it
is 90% fuel. The exhaust gases have a speed (relative to the rocket) of
250 m/s, and the burn rate is 50 kg/s. The resistance of the air can
be expressed as a retarding force given by F = −0.5Cd ρAv 2 where the
drag coefficient Cd is 0.35 and the density of the air can be assumed to
be constant and equal to 1.20 kg/m3 . The cross sectional area of the
rocket, A, is 0.8m2 . Do not assume g is constant, but you may set it
equal to 9.8 m/ sec2 at the ground. Determine the altitude reached by
the rocket when all of its fuel is burned. Plot altitude as a function of
time. Explain the shape of the curve.
Computational Project 6.4. Solve Computational Project 6.3
using a realistic profile for air density. (You can obtain tables of air
density as a function of altitude using the United States Standard At-
mosphere or the Smithsonian Meteorological Tables. These can be
found in your library or on the internet.)
Computational Project 6.5. A 5 MeV alpha particle is ap-
proaching a gold nucleus. The impact parameter is 1 Å. You may
assume the gold nucleus is initially at rest. Plot the trajectory of the
two particles and determine the angles at which both are scattered.
Computational Project 6.6. You are required to design a two
stage rocket that will accelerate a 5000 kg payload to Earth’s escape
velocity. Assume that 95% of the mass of the rockets is fuel. Assume
the exhaust velocity of the rocket motors is 2000 m/s. Investigate the
possible ranges of masses for the two stages and determine the con-
figuration that will minimize the take-off weight. Determine why a
single-stage rocket, burning the same amount of fuel, cannot accom-
plish the same objective.
Chapter 7
Conservation of Angular
Momentum

In this chapter we consider the conservation of angular momen-

tum. Although some of the concepts discussed here were mentioned in
Chapter 1, you will now be exposed to them in much greater detail.
We begin by applying the conservation of angular momentum to a
particle, and then generalize to a rigid body rotating about a fixed axis.
To keep the analysis simple, it will be limited to symmetrical bodies
with fixed rotational axes. In Chapters 15 and 16 you will study the
general rotational motion of a rigid body.

7.1. Definition of Angular Momentum

Suppose a particle of mass m is located at position r and is moving
with velocity v. Its linear momentum is p =mv. Its angular momentum
is given by

l ≡ r × p. (7.1)
See Figure 7.1. According to the definition of cross product, the angular
momentum is perpendicular to the plane defined by r and p. By the
right hand rule, the angular momentum of the particle illustrated in
Figure 7.1 points into the page.
The magnitude of the angular momentum is given by
l = mvr sin θ.
From the figure it is easy to appreciate that r sin θ is a constant; call it
b. Then l = mvb.
233
234 7. CONSERVATION OF ANGULAR MOMENTUM

Keep in mind that the angular momentum depends on the choice of

the origin of the coordinate system (because the vector r is part of the
definition). When working with angular momentum, remember that
once you have chosen an origin, you must not change it.

Figure 7.1. Illustrating angular momentum for a par-

ticle. A particle of mass m moving with velocity v is
located at r. The linear momentum of the particle is
p =mv and its angular momentum is l = r × p, a vector
pointing into the page.

An important special case is the angular momentum of a particle

moving in a circular path. If the center of the circle is chosen as the
origin, the angle between r and v is π/2 and
l = mvr sin(π/2) = mvr.
But for circular motion, v = ωr, where ω is the angular velocity. (See
Equation 1.12.) Consequently,
l = mr2 ω

Exercise 7.1. Evaluate the magnitude and direction of the angular

momentum vector for a 1000 kg automobile driving down a straight
road at 100 km/hr, with respect to a point 20 meters to the side of
the road. What is the angular momentum with respect to a point 20
meters on the other side of the road? Answer: l = 5.56 × 105 kg m2 /s.
Directions = down and up.
 7.2. CONSERVATION OF ANGULAR MOMENTUM 235

7.2. Conservation of Angular Momentum

7.2.1. Torque. Before we express the law of conservation of an-
gular momentum, let us recall the definition of torque. Suppose that
an extended rigid body is mounted on a fixed axis and is free to rotate
about that axis. (See Figures 1.5 and 1.6.) Let a force F be applied at
some point. The torque exerted by the force is
N = r × F,
where r is the vector from the origin (usually the axis of rotation) to
the point of application of the force. It is obvious from this definition
that the torque depends on the location of the axis or rotation as well
as the point of application of the force, and, of course, the magnitude
and direction of the force. However, if the net torque is zero then the
torque does not depend on the location of the axis of rotation. This
useful fact is explored in the following worked example.

Worked Example 7.1. Prove that for a body in equilibrium

the net torque about any point is zero.
Solution: Consider the relation between the torque about
some point (call it O) and the torque about some other point (O0 )
displaced by a distance d from O. The total torque about O is
X X
NO = NiO = riO × Fi ,
where riO is the vector from O to the point of application of force
Fi . See Figure 7.2.
Similarly, the torque about O0 is
X X
NO 0 = NiO0 = riO0 × Fi ,
where riO0 is the vector from O0 to the point of application of force
Fi . But as shown in Figure 7.2
riO = d + riO0
so X X X
NiO0 = riO0 × Fi = (riO − d) × Fi (7.2)
or X X X
NiO0 = NiO − d× Fi .
P
For a body in equilibrium, the net force is zero ( Fi = 0). There-
fore, if the torque about a particular point is zero, then the torque
about any other point will also be zero. This is a useful fact. For
236 7. CONSERVATION OF ANGULAR MOMENTUM

a body in equilibrium, the torque on the body with respect to any

point is zero.

Figure 7.2. A laminar body acted upon by various

forces (only one of them is shown). If the sum of the
forces is zero and the torque about point O is zero, then
the torque about any other point O0 is also zero.

7.2.2. Torque and Angular Momentum. In Section 1.5.4, we

noted that the rate of change of the angular momentum is equal to the
applied torque. We now prove this assertion, first for a particle, and
then, in the next section, for a collection of particles.
For a particle, l = r × p and so,
dl d dr dp
= (r × p) = ×p+r× .
dt dt dt dt
The first term is zero because
dr
× p = v × p =m(v × v) =0.
dt
Since dp/dt = F, and since the torque is defined by N = r × F, the
equation for dl/dt reduces to
dl dp
=r× = r × F = N. (7.3)
dt dt
That is, the rate of change of the angular momentum of a particle is
equal to the net torque exerted on it by external forces. If there is no
net external torque acting on a particle, the time rate of change of its
angular momentum is zero. That is,

dl
if N = 0 then dt
= 0 and l = constant.
 7.3. ANGULAR MOMENTUM OF A SYSTEM OF PARTICLES 237

If the torque acting on the particle is zero, its angular momentum

is conserved.

Exercise 7.2. A particle of mass m has an initial velocity v = v0 ı̂.

d2 l
It is subjected to a constant force F = f0 ̂. Evaluate dt 2 at the initial
d2 l
time. Answer: dt2 = v0 f0 k̂.

7.3. Angular Momentum of a System of Particles

Having defined the angular momentum of a single particle, let us
consider the angular momentum of a collection of particles. Since the
angular momentum of the i-th particle is li = ri × pi , the total angular
momentum of N particles is just the vector sum of the angular momenta
of all the particles. That is,
N
X N
X
L= li = (ri × pi ).
i=1 i=1

You will not be surprised to find that the internal torques these
particles exert on each other do not affect the total angular momentum.
That is, just as internal forces do not affect the total linear momentum,
so too, internal torques do not affect the total angular momentum.
Suppose the N particles have masses m1 , m2 , · · · , mN and at some
instant of time, these particles are located at positions r1 , r2 , · · · , rN .
All of the particles exert (internal) forces on each other. The force on
particle i, due to particle j, will be denoted Fij . Particle i may also
(e)
be subjected to an external force (denoted Fi ). Consequently, the
equation of motion for particle i is,
N
X
(e)
mi r̈i = Fi + Fij .
j=1
j6=i

Crossing ri into this equation of motion yields

(e)
X
ri × (mi r̈i ) = ri × Fi + ri × Fij . (7.4)
j6=i
238 7. CONSERVATION OF ANGULAR MOMENTUM

Now note that the rate of change of the angular momentum vector of
particle i is
dli d
= (ri × mi ṙi )
dt dt
= mi (ṙi × ṙi ) + mi (ri × r̈i )
= ri × mi r̈i .
Therefore, the left hand side of Equation 7.4 is just the rate of change
of the angular momentum, so
dli (e)
X
= ri × Fi + ri × Fij .
dt j6=i

There is one such equation for each particle. Adding all these equations
gives !
N N
X dli X  (e)
 X X
= ri × Fi + ri × Fij . (7.5)
i
dt i=1 i j6=i
But
X dli d X d
= li = L,
i
dt dt i dt
where L is the total angular momentum of the system, defined as the
vector sum ofthe angular momenta  of all the particles. Furthermore,
P  (e) (e)
i ri × Fi is the total torque Ntot on the system due to external
forces. Consequently, Equation (7.5) can be written as
dL (e)
XX
= Ntot + ri × Fij .
dt i j6=i

The last term, containing the double summation, is equal to zero,

as we now prove. Consider two particles, i and j. The corresponding
terms in the double sum are:
ri × Fij + rj × Fji .
By Newton’s third law, Fij = −Fji , so we can write this pair of terms
as
ri × Fij − rj × Fij = (ri − rj ) × Fij = rij × Fij ,
where rij is the relative coordinate and points from particle j to par-
ticle i. If Newton’s law is obeyed in the strong form,1 then the force
1ALagrangian analysis of the constancy of angular momentum does not invoke
the strong form of Newton’s third law. See, for example, “A Student’s Guide to
Lagrangians and Hamiltonians” by Patrick Hamill, Cambridge University Press,
2014, pages 33-36.
 7.3. ANGULAR MOMENTUM OF A SYSTEM OF PARTICLES 239

between the particles is directed along rij and so rij and Fij are paral-
lel. Therefore, the cross product rij × Fij is zero, and all the terms in
the double sum cancel out pairwise, leaving
dL (e)
= Ntot . (7.6)
dt
That is, the rate of change of the total angular momentum is equal
to the net external torque on the system. In particular, the law of
conservation of angular momentum can be expressed as follows:
If the net external torque acting on a system of
particles (or an extended body) is zero, the total
angular momentum of the system will remain constant.
This is the law of conservation of angular momentum. It is one of
the basic physical principles governing material objects and, as far as
we know, it is never violated. When you study quantum mechanics
and are introduced to the concepts of the orbital and the spin angular
momentum of electrons, you will begin to appreciate the power of this
conservation law.
Furthermore, the law of conservation of angular momentum is a
very useful tool for solving physics problems. There is a wide variety
of problems in which no external torque acts on a particle or system
of particles. Then the angular momentum is constant and the angular
momentum before some event is equal to the angular momentum after
the event.
For example, suppose a comet is attracted to some star, approaches
it, swings about the star in a hyperbolic path, and then travels back out
to “infinity.” The force exerted by the star on the comet is along the
line joining them, so the torque is zero. Therefore, the initial angular
momentum is equal to the final angular momentum, where “initial”
and “final” can refer to any two points along the trajectory.
A common problem found in introductory physics textbooks has
a running child leap onto a playground merry-go-round. This type of
problem can be solved by setting the total initial angular momentum
(the value for child plus merry-go-round before the child jumps aboard)
equal to the total final angular momentum (the value after the child
has jumped on).

Exercise 7.3. A little girl (mass = 20 kg) runs at 3 m/s and jumps
onto the rim of a playground merry-go-round that consists of a 50 kg
240 7. CONSERVATION OF ANGULAR MOMENTUM

disk with radius 1 m. The frictionless merry-go-round was initially at

rest. What is the final angular velocity of the system (child plus disk)?
Answer: 4/3 rad/sec.

Another type of angular momentum problem involves bodies upon

which a net external torque is acting. Consider, for example, a spool
with thread wrapped around it. A tension in the thread exerts a torque
on the cylinder causing its angular momentum to increase.

Figure 7.3. The Earth-Moon system, looking down on

the North Pole. The Earth rotates counterclockwise and
the Moon orbits the Earth counterclockwise. The Moon
exerts a force on the tidal bulges of the Earth. The force
on the further bulge is somewhat smaller than the force
on the nearer bulge so there is a net torque on the Earth
acting clockwise in the figure, causing the Earth to slow
down. The figure is extremely out of scale. The Earth-
Moon distance is roughly 60 Earth radii, and the bulges
are only a few degrees away from the Moon-Earth line.

An interesting example in which a torque causes a change in angular

momentum is the effect of the Moon on the rotation rate of the Earth.
As you know, the Moon raises tidal bulges in the oceans of the Earth, a
large one on the face of Earth facing the Moon and a smaller one on the
opposite side.2 These liquid bulges are dragged along with the rotating
Earth so they are not directly “under” the Moon, but somewhat ahead
of the position of the Moon by a small amount. The Moon pulls on the
tidal bulges as illustrated in Figure 7.3. The force on the nearer bulge
is greater than the force on the further bulge so there is a net torque
on the Earth. The effect of this torque is to slow down the rotation
2The reason there are two tidal bulges is the subject of Problem 10.6.
 7.3. ANGULAR MOMENTUM OF A SYSTEM OF PARTICLES 241

rate of the Earth. As the Earth rotates more and more slowly, the day
grows longer. In several billion years the day will be so long that it will
be equal to the orbital period of the Moon. That is, the day will be as
long as the month.
If we consider Earth-Moon to be an isolated system, there are no ex-
ternal torques on it, and the total angular momentum must be constant.
We have seen that the angular momentum of the Earth is decreasing,
so how can the total angular momentum remain constant? The answer
is that the angular momentum of the Moon is increasing. This requires
an increase in the distance between the Earth and the Moon. That is,
the Earth is slowing down, losing rotational angular momentum, and
the Moon is receding from the Earth, gaining orbital angular momen-
tum in such a way as to conserve the total angular momentum of the
system.3

Worked Example 7.2. You have invented a certain device

that is to be used in outer space (so gravity can be neglected). It
consists of a small weight of mass m connected to a massless string
of length d and it is spun around in a circle whose center is at a fixed
point. The initial angular velocity is ω0 . The string is pulled in until
it is half of its initial length. (a) Evaluate the change in angular
momentum. (b) Compare the final angular velocity to the initial
angular velocity. (c) Evaluate the change in kinetic energy.(d) How
much work was done in pulling the mass closer to the center?
Solution: (a) The initial angular momentum is l = md2 ω0 .
The force is along the string so the torque = N = r × F =0 and
angular momentum is conserved. Therefore ∆l = 0.
(b) By conservation of angular momentum
lf = li =⇒ md2f ωf = md2 ω0 .
But df = 0.5d, so
d2
ωf = ω0 = 4ω0.
(d/2)2

3The tidal interaction between Earth and Moon has many fascinating aspects.
It has been used to explain the eccentricity of the Moon’s orbit, but recent studies
seem to indicate that the gravitational force of Jupiter may be more important.
If you are interested, you might look up the paper by Matija Cuk, “Excitation of
lunar eccentricity by planetary resonances,” Science, 318, 244 (2007),
242 7. CONSERVATION OF ANGULAR MOMENTUM

(c)The change in kinetic energy is

∆T = Tf − T0 .
Since v = ωr and T = (1/2)mv 2 , the kinetic energy can be ex-
pressed as T = (1/2)m (ωr)2 . Therefore, T0 = 21 md2 ω02 and Tf =
1
2
m(d/2)2 ωf2 . Consequently,
1 1
∆T = m(d/2)2 (4ω0 )2 − md2 ω02
2  2

1 2 2 1 3
= md ω0 (16) − 1 = md2 ω02 = 3T0 .
2 4 2
(d) By the work energy theorem, the work done is 3T0 =
3
2
md2 ω02 .

Exercise 7.4. An Atwood’s machine has two weights tied to the

ends of a string that passes over a pulley of radius 2 cm. The two
masses are 500 gr and 700 gr. Assuming the string does not slip,
determine the rate of change of angular momentum of the pulley.The
moment of inertia of the pulley is 100 gr cm2 . (Note: Since the masses
are accelerating, you cannot assume tension equals mg.) Answer: 8000
dyne cm.
Exercise 7.5. A child’s top of mass M is spinning at an angular
velocity ω. When it slows down, it leans over a bit so that the axis
is at an angle θ with the vertical. The center of mass is a distance R
from the tip. Determine the gravitational torque with respect to the
point of contact with the floor. What is the direction of this torque?
Draw a diagram indicating the direction of the rate of change of angular
momentum. (This is why the top precesses.) Answer: N = RM g sin θ
perpendicular to ω and g.
Exercise 7.6. A dumbbell consists of two point masses M con-
nected by a massless rod of length d. It is rotating about an axis through
its center of mass at a rate ω0 . One of the masses comes loose and flies
off. Determine the angular velocity of the remaining mass.

7.3.1. Angular Momentum Relative to the Center of Mass.

Recall that the kinetic energy of a system of particles can be expressed
as the sum of the kinetic energy of the center of mass plus the kinetic
energy of all the particles with respect to the center of mass. (See
 7.3. ANGULAR MOMENTUM OF A SYSTEM OF PARTICLES 243

Section 5.8.) A similar relation holds for the linear momentum of a

system of particles. (See Section 6.6.) As we now demonstrate, the
angular momentum can also be expressed as the angular momentum
of the center of mass, plus the angular momentum about the center of
mass.
The total angular momentum of a system of particles is
X
L= ri × mi ṙi .
i

The position vector of particle i can be expressed as

ri = rc + r0i ,
where rc is the vector to the center of mass and r0i is the position of
particle i with respect to the center of mass (see Figure 5.13). Using
this relationship, the angular momentum is
X
L = mi (rc + r0i ) × (ṙc + ṙ0i )
i
X X
= mi rc × (ṙc + ṙ0i ) + mi r0i × (ṙc + ṙ0i )
i i
X X X X
= mi rc × ṙc + mi rc × ṙ0i + mi r0i × ṙc + mi r0i × ṙ0i .
i i i i

The quantities rc and ṙc do not depend

P on i and can be taken out of
the summations. Also note that i mi = M (the total mass). Conse-
quently,
! !
X X X
L =M (rc × ṙc ) + rc × mi ṙ0i − ṙc × mi r0i + mi (r0i × ṙ0i ) .
i i i
0
= 0 and i mi ṙ0i =
P P
But the two middle terms are zero because i mi ri
0 by the definition of the center of mass. (The first of these is propor-
tional to the position of the center of mass with respect to itself and
the second is proportional to the velocity of the center of mass with
respect to itself.) The expression for L then reduces to
X
L =M (rc × ṙc ) + mi (r0i × ṙ0i ) ,
i

which can be written

L = Lc + L0 .
Here Lc = M (rc × ṙc ) is the angular momentum of a particle of mass
M locatedPat the center of mass and moving with the center of mass,
and L = i mi r0i × ṙ0i is the sum of the angular momenta of all the
0
244 7. CONSERVATION OF ANGULAR MOMENTUM

particles relative to the center of mass. That is what we set out to

prove.

mi r0i = 0.
P
Exercise 7.7. Prove that i

7.4. Rotation of a Rigid Body about a Fixed Axis

A rigid body is a system of particles in which all the particles are
constrained to remain at constant distances from one another. If the
body is rotating about a fixed axis of rotation, the particles move in
circular paths centered on the axis. These paths lie in planes perpen-
dicular to the axis of rotation. All the particles have the same angular
velocity. The angular velocity ω is a vector directed along the rota-
tion axis whose magnitude is equal to the rotation rate. The sense of
ω is given by the right hand rule, as illustrated in Figure 7.4. The
magnitude of ω is
2π
|ω| = ω = = 2πf,
T
where T is the period of revolution and f is the frequency.

Figure 7.4. The direction of the vector angular veloc-

ity ω is given by the right hand rule. If the fingers of
your right hand curl in the direction of the motion, your
thumb points in the direction of ω.

As noted in Equation (1.23), the linear velocity vi of particle i is

related to the angular velocity ω by
vi = ω × ri⊥ ,
where ri⊥ is a vector in the plane of the circular path from the axis of
rotation to the particle. Since the angle between ω and ri⊥ is 90◦ , the
 7.4. ROTATION OF A RIGID BODY ABOUT A FIXED AXIS 245

magnitude of vi is ri⊥ ω, as expected, and the angular momentum of

2
particle mi is li =mi ri⊥ ω. I wrote the distance from the axis of rotation
as ri⊥ to emphasize that it is the perpendicular distance from the axis
and for future convenience.
Now recall that angular momentum depends on the choice of origin.
If we place the origin at the center of the circular path of some particle,
say mi , then it will not be at the center of the path of another particle,
say mj , that is displaced somewhat along the axis of rotation. We need
an expression for the angular momentum of a particle relative to an
arbitrary origin. However, to keep our analysis simple, let us restrict
the location of the origin to a point on the axis of rotation. This is not
the most general position, but it is sufficient for our purposes at this
time.
Our generalization of the origin to an arbitrary location of the axis
of rotation is illustrated in Figure 7.5.
The angular momentum of particle i can be expressed in terms of
the angular velocity as follows:
li = ri × pi = mi (ri × vi ) = mi ri × (ω × ri ) . (7.7)
Using the “BAC-CAB” rule for the vector triple product,4 this becomes

li =mi ri2 ω − (ri · ω) ri .

 

From Figure 7.5, we appreciate that if the z axis is along ω, then

ri = ri cos θi k̂ + ri sin θi êi
where θi is the angle between ri and ω, and êi is a unit vector in the
plane of the circular path and pointing (instantaneously) towards mi .
Using this expression for ri and doing a little bit of algebra, you can
easily show that

li = mi ri2 ω sin2 θi k̂−mi ri2 ω sin θi cos θi êi . (7.8)

Now I am going to place an important restriction on the analysis.
I will only consider bodies that are symmetrical about the rotation
axis. (When this restriction is lifted in Chapter 16 you will need to
use tensors to describe the motion. I don’t want to talk about tensors
just yet, so let’s assume the rotating body is symmetrical about the
axis of rotation.) In a symmetrical body, for every particle i there is
an identical particle j diametrically opposite to it at the same distance
from the axis of symmetry. Consider the ith particle which has mass
mi and angular momentum li , as shown in Figure 7.6. The jth particle
4The “BAC - CAB” rule is A × (B × C) = B(A · C) − C(A · B).
246 7. CONSERVATION OF ANGULAR MOMENTUM

^
ei
ri vi
mi

ri
i
k

Figure 7.5. The origin is located on the axis of rota-

tion. The vector ri specifies the location of particle mi
relative to the origin. ri is resolved into a component
along the rotation axis (ri cos θi k̂), and a component per-
pendicular to the rotation axis (ri sin θi êi ). The unit vec-
tor êi points towards the instantaneous position of the
particle.

is identical, except for the direction of its angular momentum, lj . The

components of li and lj parallel to the rotation axis add together, but
the components perpendicular to the rotation axis cancel. So adding
equations of the form (7.8) to get the total angular momentum of a
symmetrical body yields
X
L= mi ri2 ω sin2 θi k̂.
i

The quantity ri sin θi is the perpendicular distance from the axis of

rotation to the ith particle. If this distance is denoted by ri⊥ we can
write !
X
2
L= mi ri⊥ ω k̂. (7.9)
i

Exercise 7.8. Fill in the missing steps between Equations (7.7)

and (7.8).
Exercise 7.9. A spinning disk of radius 10 cm and mass 15 gr is
rotating at 50 rad/s. A sticky piece of chewing gum (mass = 3 gr)
falls on the rim of the disk. Determine the final angular velocity of the
system. Answer: 35.7 rad/sec.
 7.5. THE MOMENT OF INERTIA 247

Figure 7.6. The angular momentum of a particle (rela-

tive to the given origin) has components in the direction
of the axis of rotation and perpendicular to it, as shown
in the panel on the left. The angular momentum of a
symmetric particle has an equal but opposite component
perpendicular to the rotation axis.

7.5. The Moment of Inertia

2
P
The quantity i mi ri⊥ in Equation (7.9) depends only on the mass
of the body and how that mass is distributed with respect to the axis
of rotation. For any given body rotating about some specific axis, this
quantity is a characteristic constant. We call it the moment of inertia
of the body5 and denote it by the letter I.
X
2
I= mi ri⊥ . (7.10)
i

We can generalize Equation (7.10) to a continuous body of density ρ,

and write Z Z Z Z Z Z
2 2
I= r⊥ dm = ρr⊥ dτ. (7.11)
body body

5Although you studied the moment of inertia in your introductory mechanics

course, you will find out that this topic is more complicated than you might expect.
In discussing the general motion of a rigid body we need to introduce a quantity
called the inertia tensor. For the moment, however, we will restrict ourselves to
the simple situation of a symmetrical rigid body rotating about a fixed axis and
the scalar quantity I is all we need.
248 7. CONSERVATION OF ANGULAR MOMENTUM

where dm = ρdτ is a mass element of the body and r⊥ is the perpen-

dicular distance from mass element dm to the axis of rotation.
Using the definition for moment of inertia (Equation 7.10), we can
write the angular momentum (Equation 7.9) in the form
L =Iω.
Keep in mind that this relationship is based on the body being sym-
metrical about the rotation axis. Only then will the angular momentum
be parallel to the angular velocity. If the body is not symmetric about
the axis of rotation, the angular momentum vector will not be parallel
to the angular velocity vector. (Such a body will tend to “wobble”
about the axis.)
We now return to the relationship between the torque and the rate
of change of angular momentum. If a torque N acts on a body there
will be a change in its angular momentum given by
dL
N= .
dt
Replacing L by Iω,
d
N = (Iω) .
dt
For a body with constant moment of inertia, this becomes
dω
N=I
dt
or
N = Iα. (7.12)
This has the same form as F = ma with I replacing m, α replacing a,
and N replacing F.
The conservation of angular momentum for a rigid body when no
external torques are acting on it is just Li = Lf . For a symmetrical
body this can be written as
Ii ωi =If ωf .
Observe the analogy between the expression for linear momentum,
p =mv, and the expression for angular momentum, L =Iω.
7.5.1. Two Theorems. Two useful theorems for calculating the
moment of inertia are the parallel axis theorem and the perpendicular
axis theorem.
The parallel axis theorem states that the moment of inertia of a
body about a given axis is equal to the moment of inertia about a
parallel axis through the center of mass plus M d2 where M is the mass
of the body and d is the distance between the axes. The parallel axis
 7.5. THE MOMENT OF INERTIA 249

theorem is a very valuable labor saving device. If we write Ik for the

moment of inertia about the parallel axis and Ic for the moment of
inertia about an axis that passes through the center of mass, we have
Ik = Ic + M d2 .
The perpendicular axis theorem can be applied to plane laminar
bodies. Consider two mutually perpendicular axes in the plane of the
lamina. Let the moments of inertia about these two axes be I1 and I2 .
Then the moment of inertia about an axis perpendicular to the lamina
and passing through the point of intersection of the first two axes, is
I3 = I1 + I2 .

Figure 7.7. A rod of length l with axis passing through

the center of mass. The mass element dm = λdx is a
distance x from the axis.

Worked Example 7.3. (a) Determine the moment of inertia

of a cylindrical rod of mass M and length l about an axis passing
through the center of mass of the rod (line AA), as illustrated in
figure 7.7. (b) Determine the moment of inertia about line BB at
the end of the rod.
Solution: (a) The mass per unit length (linear mass density)
is λ = M/l. The moment of inertia about the axis (AA) is
Z Z Z Z Z l/2  3 l/2
2 2 2 x
I = r⊥ dm = r⊥ λdx = λ x dx = λ
−l/2 3 −l/2
λ 3  1
= 2l /8 = M l2 .
3 12
250 7. CONSERVATION OF ANGULAR MOMENTUM

(b) The moment of inertia about BB can be obtained using the

parallel axis theorem
Ik = Ic + M d2
 2
1 l 1
= M l2 + M = M l2 .
12 2 3

Exercise 7.10. Determine the moment of inertia of a rod of length

l and mass M about an axis perpendicular to the axis of the rod and
a distance a beyond one end of the rod. Answer: M a2 + M al + 13 M l2 .
Exercise 7.11. A force of 2 N is applied tangentially on the rim
of a disk of mass 15 kg and radius 20 cm. The disk is mounted on a
frictionless fixed axis perpendicular to the plane of the disk and passing
through its center. Determine the angular acceleration of the disk.
Answer: 4/3 rad/sec2 .
Exercise 7.12. The moment of inertia of a uniform square plate
of side a about a perpendicular axis through its center is (1/6)M a2 .
Determine the moment of inertia about an axis in the plane of the
plate. Assume this axis passes through the center of the plate. Answer:
(1/12)M a2 .
RRR 2
Exercise 7.13. From the definition I = r⊥ dm, obtain the
moment of inertia for a ring or hoop of mass M and radius R, relative
to an axis through its center and perpendicular to the plane of the ring.
Answer: M R2 .
Exercise 7.14. A meter stick of mass 0.25 kg is mounted horizon-
tally on a perpendicular axis at the 20 cm mark. A downward force of
3 N is applied at the 80 cm mark. Determine the angular acceleration
of the ruler. Answer: 44.12 rad/s2 .

7.6. The Gyroscope

Figure 7.8 illustrates a simple gyroscope. For convenience let us
place the origin of coordinates at the base (point Q in the figure). The
gyroscope is symmetrical, so the angular momentum vector and the
angular velocity vector are parallel and point along the axis of rotation.
The gravitational force acts at the center of mass, which is also on the
 7.6. THE GYROSCOPE 251

axis of rotation, a distance R from Q. The torque about Q due to this

force is
NQ = R×M g.
dL
Since dt = N we have
dL
= R×M g =RM g sin θφ̂.
dt
The torque points in a direction perpendicular to both R and M g (as
well as L). In spherical coordinates this torque points in the direction
of the unit vector φ̂. Hence, N = N φ̂. Further, note that L =Iω and
I is a constant, so
dL dω dω
N= =I = I φ̂.
dt dt dt
This tells us that the angular momentum vector (as well as the angular
velocity vector) is changing. It is not changing in magnitude because
the direction of N is perpendicular to the direction of the angular
momentum. Consequently, the angular momentum vector must be
changing in direction, as indicated in the rightmost sketch in the figure.

Figure 7.8. A gyroscope. The middle sketch shows

the vectors involved. Note that the torque (R×M g) is
perpendicular to both R and M g. The sketch to the right
illustrates that the tip of the angular momentum vector
traces out a circle.
The gyroscope precesses, that is, the axis of rotation changes in
direction in such a way that the tip of the angular momentum vector
traces out a circle, as indicated in the figure. The precession rate, Ωp ,
is the speed with which the tip of the angular momentum vector goes
around this circle. That is,
dφ
Ωp = .
dt
252 7. CONSERVATION OF ANGULAR MOMENTUM

Figure 7.9 gives three different views of the precession of the angu-
lar momentum vector around the vertical. The leftmost figure shows
the angular momentum vector and the circle traced out by the tip of
this vector. Note that as time passes, the angular momentum goes
from L to L + ∆L. The magnitude of the angular momentum does not
change, only its direction, so ∆L lies in the plane of the circle (and
is perpendicular to both L and L + ∆L). The middle figure is a side
view, indicating that θ is the angle between L and the vertical, and
that the horizontal component of L is L sin θ. The rightmost figure is a
top view, looking down on the circular path, indicating that ∆L is the
chord of the arc subtended by the angle ∆φ. Setting the chord equal
to the arc we have
∆L = L sin θ∆φ.
Divide by ∆t and let ∆t → dt, to obtain
dL dφ
= L sin θ .
dt dt
dL dφ
Now recall that dt
= RM g sin θ, and dt
= Ωp so
RM g sin θ = L sin θΩp ,
and consequently
RM g RM g
Ωp = = .
L Iω
It is interesting to note that the precession rate increases as the top
slows down.
Since the angular momentum vector and the angular velocity vector
(ω) lie along the same line, they vary together. The angular velocity
vector is precessing at the same rate as the angular momentum vector.
A gyroscope also “nods” as it precesses. This is called nutation.
We shall consider it in Chapter 16.

Exercise 7.15. A gyroscope is made of a very light axle and a

heavy ring. The ring (of mass M and radius a) is connected to the
axle by nearly massless spokes. The ring is a distance d from the point
of the axis in contact with the floor. Determine the precession rate of
this gyroscope. (Answer: Ωp = gd/a2 ω)
 7.7. ANGULAR MOMENTUM IS AN AXIAL VECTOR 253

z z
Lsinθ
ΔL ΔL
ΔΦ
L L+ΔL
L Lsinθ
θ

Precessing Angular Side View Top View

Momentum Vector

Figure 7.9. Left figure: The tip of the precessing angu-

lar momentum vector traces out a circle. Middle figure:
Side view showing the angle θ between the angular mo-
mentum vector and the vertical. The horizontal compo-
nent of angular momentum is L sin θ. Right figure: Top
view showing that ∆L = L sin θ∆φ.

7.7. Angular Momentum is an Axial Vector

Angular momentum is a vector, but it is a somewhat different type
of vector than familiar “ordinary” vectors such as the velocity vector
and the position vector. The difference between the angular momentum
vector and an ordinary vector is related to its behavior under a coordi-
nate transformation. But before describing this difference, I have to tell
you that there are two kinds of coordinate transformations. They are
called “proper” transformations and (of course) “improper” transfor-
mations, depending on whether or not the coordinate axes retain their
cyclic order. We usually draw a Cartesian coordinate system such that
ı̂ × ̂ = k̂ Such a coordinate system is called “right handed.” Using
your right hand, if you point your fingers along the x-axis and bend
them into the y-axis, your thumb will be pointing in the direction of
the z-axis. If you rotate a right handed coordinate system, you still get
a right handed coordinate system, even though the axes will now be
pointing in different directions. However, if you reflect a right handed
coordinate system, you end up with a left handed coordinate system.
This is illustrated in Figure 7.10 where the right handed coordinate
system reflected in the xz-plane is transformed into a left handed coor-
dinate system. This is, of course, an example of an improper coordinate
transformation.
Consider the behavior of a vector under a transformation of the
coordinate system. To begin, consider the behavior of the position
vector r. The components of r may or may not change sign during
a coordinate transformation. That detail is not important. However,
254 7. CONSERVATION OF ANGULAR MOMENTUM

Figure 7.10. The reflection of a Cartesian coordinate

system. The reflection is in the xz-plane. Note that the
axis perpendicular to the reflecting plane changes direc-
tion.

what is important is whether or not the components of other vectors

have the same sign changes as the components of r under the same
transformation.
We call r a “polar” or “true” vector. Feynman6 called it an “honest”
vector. Many other vectors, such as velocity, momentum and force, are
polar vectors. They behave the same way as r under a coordinate
transformation such as a rotation or a reflection.
For example, if you reflect the vector r it changes direction. (More
precisely, the component of r perpendicular to the plane of reflection,
changes sign.) See Figure 7.11. Under a similar reflection, velocity and
force behave the same way.
There are other vectors, particularly those related to rotations, that
do not transform like r. These are called “axial” or “pseudo” vectors.
Recall that angular velocity ω was defined to be a vector lying along
the axis of rotation. The sense or direction of ω was given by the right
hand rule (see Figure 7.3). If you observe a rotating wheel in a mirror,
you see that the angular velocity vector does not change direction as
shown in Figure 7.12. Angular momentum is another axial vector. The
component perpendicular to the plane of reflection does not change
direction. Therefore, angular velocity and angular momentum do not
transform like the “honest” vector r. This means that ω and L are
axial vectors. Axial vectors behave differently than r under improper
transformations.
6Richard
P. Feynman, The Feynman Lectures on Physics, Addison-Wesley,
Reading, Mass., 1963.
 7.8. SUMMARY 255

Figure 7.11. The reflection of the position vector r in a mirror.

You will study the transformation properties of vectors in much

more detail in a course in mathematical methods. At this stage, it is
sufficient for you to realize that there are different kinds of vectors and
that they are distinguished by their transformation properties.

Figure 7.12. The angular velocity vector does not

change directions under reflection.

7.8. Summary
The angular momentum of a particle is defined as
l=r×p
and, consequently, it depends on the choice of origin. Since
dl d
= (r × p) = r × F = N
dt dt
256 7. CONSERVATION OF ANGULAR MOMENTUM

we appreciate that if N = 0 then l = constant. This relation summa-

rizes the law of conservation of angular momentum.
The angular momentum of a system of particles (or an extended
body) is
L =Σli
and the law of conservation of angular momentum can be written
dL (e)
= Ntot
dt
(e)
where Ntot is the net torque due to all external forces. Internal forces
(assuming they obey the third law in the strong form) do not contribute
to the rate of change of angular momentum.
The angular momentum of a system of particles can be written
as the sum of the angular momentum of the center of mass and the
angular momentum relative to the center of mass, that is,
L = Lc +L0 ,
where
Lc = M (rc ×ṙc ),
and
L0 = Σmi (r0i ×ṙ0i ).
The angular momentum of a symmetrical body rotating about an axis
of symmetry is
 
2
L = Σmi ri⊥ ω.
i
The quantity in parenthesis is the moment of inertia of the body and
is denoted I. Therefore,
L =Iω,
and we can write
(e) dL d d
Ntot = = Iω = Iω = Iα.
dt dt dt
(e)
If Ntot = 0, the angular momentum is constant and consequently
Lf = Li =⇒ If ωf = Ii ωi .
The motion of a gyroscope illustrates the concepts presented in
this chapter. The torque due to gravity causes the angular momentum
vector to precess at a rate
RM g
Ωp = .
Iω
 7.9. PROBLEMS 257

7.9. Problems
Problem 7.1. Two particles of equal charge and equal mass travel
at the same speed v but in opposite directions. Their initial trajecto-
ries are straight lines separated by a distance b. Due to their mutual
repulsion, they will reach a point of closest approach and then move
away from each other. The electrical force between the charges is given
by F = (q 2 /4π0 r2 )r̂. The vector rr̂ is drawn from one particle to the
other. Although it is unrealistic to do so, ignore the magnetic force
between the moving charges. (a) Draw the trajectories of the two
particles, indicating the electric forces acting on them. (b) Explain
whether or not the angular momentum of the system is constant. (c)
Do your answers depend on your choice of coordinate origin?
Problem 7.2. A uniform rod of length d and mass M is moving
at a constant velocity v in the direction of its length. (a) Obtain its
angular momentum relative to an arbitrary origin and show that this
is the same as the angular momentum of a particle of mass M moving
along the trajectory of the rod. (b) If the rod is inclined at an angle α
to its velocity vector, what is the position of the equivalent particle?
Problem 7.3. Prove that the total angular momentum of a system
of two particles is independent of the displacement of the origin only if
the total linear momentum is zero or if the displacement of the origin
is parallel to the total linear momentum.
Problem 7.4. In proving the law of conservation of angular mo-
mentum for an extended body (or system of particles) we assumed
that Newton’s third law is obeyed in the strong form. In this problem
you will investigate this assumption for the case of particles interacting
through magnetic forces. (The magnetic force between moving charged
particles is not directed along the line joining the particles.) The or-
bital motion of an electron around the nucleus is equivalent to a small
current loop, which we usually call a “magnetic moment.” A magnetic
moment generates a magnetic field. Another electron orbiting around
another nucleus will feel a force due to this magnetic field. Let us gen-
eralize the problem to two current loops of arbitrary shape that carry
currents I1 and I2 . Considering small elements of the current loops,
I1 dl1 and I2 dl2 , the force on I1 dl1 due to I2 dl2 is
dF12 = I1 dl1 × B
where
I2 dl2 × r12
I
µ0
B= .
4π |r12 |3
2
258 7. CONSERVATION OF ANGULAR MOMENTUM

Show that the force between the two current loops obeys Newton’s third
law in the strong form. (Hint: this involves showing that dl1 · r12 / |r12 |3
is an exact differential.)
Problem 7.5. An electron is fired horizontally with speed voy be-
tween parallel charged electrical plates. Assume the plates are oriented
horizontally and that the top plate is negatively charged and the bot-
tom plate is positively charged, so the electron feels a uniform force
downward. Denote this force by Fe . Using Cartesian coordinates, cal-
culate the angular momentum as a function of time and show that the
time rate of change of angular momentum is equal to the torque exerted
on the electron by the electric field. Evaluate the angular momentum
(and torque) relative to an origin located at the position of the elec-
tron when it first enters the electric field. See Figure 7.13. Answer:
l = −(1/2)v0y Fe t2 ı̂.

Figure 7.13. An electron between charged plates.

Problem 7.6. In analyzing the collision of two particles we stated

that the conservation of linear momentum (p =constant) led to three
equations describing the motion. Since there are no external torques
acting on the system, it might seem that conservation of angular mo-
mentum (l =constant) would lead to three additional equations. Show
that this is not true, that is, that conservation of angular momentum
does not lead to any new relations.
Problem 7.7. A roll of rather heavy gift wrapping paper is placed
on a frictionless roller and suspended by a bracket of length b as shown
in Figure 7.13. The roll is in contact with the wall. The coefficient of
static friction between wall and paper is µ. A portion d of the paper
is hanging down. Assume the total length of paper on the roll is D
and that it has a linear mass density λ. Determine the length d such
that the paper will unroll on its own. At the instant the roll begins to
unwind on its own, it has a radius R.
 7.9. PROBLEMS 259

Figure 7.14. A roll of paper that will unwind under its

own weight.

Problem 7.8. Professor Ptolemy of the Astronomy Department

claims that there is a planet with the same mass as Earth that is
directly on the other side of the Sun in an orbit of the same radius as
Earth’s. This planet, he says, has the same mass as Earth, but it is
cylindrical in shape. The axis of the cylinder lies in the orbit plane
and its length is one Earth diameter. Determine the torque the Sun
exerts on this planet, as a function of the angle between the axis of
the cylinder and a line from the planet to the Sun. You may assume
the angle between the axis of the cylinder and the line to the Sun is
small. (Hint: Determine the gravitational force on the planet, then
find the center of gravity and finally calculate the torque. By the way,
if there were a planet on the other side of the Sun from us, we would
be able to see it at times, due to the eccentricity of Earth’s orbit. But
in this problem you may assume circular orbits. Hint #2: The center
of gravity is the point in the body where the gravitational force acts.
In this problem it is not at the same point as the center of mass.)
Problem 7.9. A neutron star whose mass is 1.5 times the mass
of the Sun has collapsed to a sphere of radius 10 km. If the original
angular velocity was one revolution per month (before the collapse),
what is its final angular velocity? Assume the initial density of the
star was equal to the mean density of the Sun.
Problem 7.10. Prove the parallel axis theorem.
Problem 7.11. Prove the perpendicular axis theorem.
Problem 7.12. (a) Derive an expression for the moment of inertia
of a disk of mass M and radius R about an axis perpendicular to its
plane and passing through its center of mass. (b) Obtain an expression
260 7. CONSERVATION OF ANGULAR MOMENTUM

for the moment of inertia about a parallel axis that is tangent to the
edge of the disk.
Problem 7.13. Determine the moment of inertia of an annular
cylinder (or ring) about an axis perpendicular to the plane of the ring
and passing through its center. The ring has uniform density, mass M,
inner radius R1 and outer radius R2 .
Problem 7.14. Derive an expression for the moment of inertia of
a uniform, solid sphere about a diameter.
Problem 7.15. Determine the moment of inertia of a flat disk of
mass M and radius R about an axis tangent to the edge of the disk
and lying in the plane of the disk.
Problem 7.16. Assume the Earth is a sphere of mass M and radius
R. Remove the Southern Hemisphere. By direct integration determine
the moment of inertia of the remaining hemisphere about an axis per-
pendicular to the equatorial plane and passing through the North Pole.
Problem 7.17. A disk of mass M and moment of inertia I has a
hole in its center. An axle, which is slightly smaller in diameter than
the hole, passes through the hole. The axle is horizontal and the plane
of the disk is vertical. The disk spins smoothly on the axle. There is
friction between the disk and the axle, and the coefficient of friction
is µ. Assume the disk has an initial angular speed ω0 . Determine the
number of turns and the time for the disk to stop. (Hint: the point of
contact is not over the center of mass.)
Problem 7.18. A dumbbell is made up of a rod of length l and
mass m connected on either end to spheres of mass M and radius
R. (a) Determine the moment of inertia of the dumbbell about an axis
perpendicular to the rod and going through its center. (b) You suspend
the dumbbell from its center of mass and apply a force of 1.5 N at the
center of mass of one of the spheres. The force is in the horizontal plane
and perpendicular to the rod. Determine the angular acceleration of
the system. Assume M = 1 kg, R = 10 cm, l = 20 cm and m = 0.1
kg. Answer: (b) 0.1 rad/s2 .
Problem 7.19. The gyroscope of Figure 7.6 is made up of a disk
of mass M and radius a and an axis of negligible mass that is free to
swivel in any manner about the bottom end. It is inclined at an angle
θ = 90◦ so the axis is horizontal and the gyroscope is precessing. The
disk spins about the axis with angular speed ω. Write an expression
for the total angular velocity vector in Cartesian coordinates in terms
of M, a, R, ω, the time, and any other appropriate parameters.
 7.9. PROBLEMS 261

Problem 7.20. A gyroscope that is free to move about the support

point is in the upright position and is spinning at an angular velocity ω k̂
(see Figure 7.6). Assume the axis has a length d and negligible mass.
The disk of mass M and radius a is mounted at d/2. A frictionless ball
bearing device allows you to grasp the top of the axis and pull it to a
horizontal orientation. After pulling the gyroscope down, you hold it
stationary with the axis horizontal. (a) How much work did you do to
rotate the gyroscope axis? (b) What is the final angular velocity of the
gyroscope? (c) What force do you need to exert on the tip of the axis
to keep it stationary? You may assume that ω is large so Ωp is small
enough that the energy associated with precession can be neglected.

Figure 7.15. The speed of the particle m does not in-

crease as the string winds up on the cylindrical rod.

Problem 7.21. A string is partially wound around a cylindrical

rod of radius a, as shown in Figure 7.15. A particle of mass m is tied
to the end of the string. The mass m is given an initial velocity v0 and
the string winds itself onto the rod. Make the unrealistic assumption
that there is no gravitational force. Show that the the linear velocity
of the mass is constant. Do this is in the following two ways: (a)
Using the conservation of energy, and (b) Using the fact that the rate
of change of angular momentum is equal to the torque. (Hint: there is
a torque on the particle because the rod does not have zero radius and
the instantaneous velocity of the particle is perpendicular to the string
at all times.}
Problem 7.22. A pendulum consists of a bob of mass M and a
massless string of length l. The pendulum is initially at rest. It is
struck by a bullet of mass m and velocity v. Let m = 0.1M. Assume
the pendulum hangs vertically and the bullet travels horizontally. (a)
Determine the initial angular velocity of the pendulum if the bullet be-
comes embedded in the bob. (b) Determine the initial angular velocity
of the pendulum bob if the bullet bounces back off the bob elastically.
262 7. CONSERVATION OF ANGULAR MOMENTUM

Problem 7.23. Generalize the derivation of dL dt

= N to the case of
a moving origin. If the origin (Q) is at rQ , and LQ and NQ are the
total angular momentum and torque with respect to Q show that
dLQ
= NQ
dt
as long as the point Q has zero acceleration or is accelerating along
the line from Q to the center of mass. You may assume the internal
forces obey the third law in the strong form so the total internal torque
vanishes.
Problem 7.24. Consider a symmetrical rigid body to be composed
of N particles of masses mi (i = 1, N ). Assume the body is rotating
about a fixed axis with angular speed ω. Prove that the kinetic energy
is 21 Iω 2 where I is the moment of inertia.
Problem 7.25. A particle moves in the field of an attractive central
force whose potential is given by V = −k/r. Show that the angular
momentum does not change.

Figure 7.16. Two disks collide and stick together. See

Problem 7.26.

Problem 7.26. Figure 7.16 is a view from above, showing two

identical disks of mass M and radius R on a frictionless surface. One
disk is at rest, the other is rotating counterclockwise with angular ve-
locity ω and is moving with a linear velocity v = 21 ωR. It makes a
grazing collision with the second disk at point P. After the collision
the two disks stick together. Determine the final angular momentum
of the system with respect to P.
Computational Projects
Computational Project 7.1. A child’s playground merry-go-
round has been fitted with frictionless bearings. If air resistance can be
 7.9. PROBLEMS 263

ignored, this “ideal merry-go-round” will rotate indefinitely. Suppose

that it is spinning at 3 rad/sec when it starts to snow. The rate at
which snow accumulates on the disk is 1 gram/second. The disk has a
radius of 1.5 m and a mass of 40 kg. Determine the angular speed of
the merry-go-round after 30 minutes of steady snowfall. Plot ω vs t.
Chapter 8
Conservation Laws and Symmetries

8.1. Symmetry
Nature exhibits many symmetries, from the the structure of atoms
to the patterns of snowflakes. As physicists, we exploit the concept
of symmetry to understand the nature of physical reality. On a more
mundane level, we find symmetry concepts to be a valuable aid in
solving physics problems.
All of us have an intuitive concept of what is meant by symmetry.
We recognize spatial symmetry in flowers and snowflakes and in the
bodies of most living creatures. We say these things “look the same”
when viewed from different points of view. For example, a snowflake
looks the same when rotated through 60 degrees (known as hexagonal
symmetry). A human body looks the same when viewed in a mirror.
A cylindrical vase looks the same when rotated through an arbitrary
angle. Although its mirror image may be indistinguishable from the
vase itself, it will probably look different when it is turned upside down.
A physicist would say the vase has rotational symmetry and symmetry
under reflection, but it does not have “right side up - upside down”
(axial) symmetry.
From the examples above, you can appreciate that a symmetry is
related to an invariance under some particular operation. In simpler
words, a symmetry means that a system is not changed when something
is done to it. If there is no way we can tell whether or not a vase has
been rotated through some angle, then it has rotational symmetry. If a
pendulum oscillates in exactly the same way on one side of the room as
on the other, then it has translational symmetry. A mechanical system
that does not change in time has temporal symmetry.

265
266 8. CONSERVATION LAWS AND SYMMETRIES

Among the simpler symmetry operations are spatial translations,

rotations, reflections, and translations in time. Other symmetry oper-
ations are not as familiar to you. For example, “charge conjugation”
symmetry refers to interchanging positive and negative charges. Does
an “antihydrogen atom” with a negative proton and a positive electron
behave the same as an ordinary hydrogen atom? If so, it has charge con-
jugation symmetry. Another way of expressing invariance under charge
conjugation is to call it matter-antimatter symmetry. This means that
matter and antimatter particles are attracted to one another by the
gravitational and electromagnetic forces and appear to be very much
the same except for the fact that when they come into contact they
annihilate one another!
An interesting symmetry is called “chirality”1 This is the symme-
try between your left hand and the mirror image of your right hand.
There is no combination of translations or rotations that will transform
one hand into the image of the other, but a mirror reflection does al-
low such a transformation. This is important in chemistry since some
molecules are chiral, that is, they exist in both “left hand” and “right
hand” forms, which may have different properties, such as their effect
on polarized light. It is worth noting that proteins in sugars contain
the “right-handed” form of the the alanine molecule. This molecule
also comes in a mirror imaged or “left-handed” form. Biological sys-
tems (such as you and me) can only digest the right-handed alanine as
a result of some evolutionary quirk. However, it is possible to manu-
facture “left-handed” sugar that tastes the same as “ordinary” sugar,
but it is not digested. This product is useful for diabetics and people
who are avoiding sugar for medical reasons. It has the drawback of
being expensive to produce.
For a simple example, consider a translation operation. Suppose
a particle is lying on a smooth, infinite, featureless table. If we slide
the particle to some other point on the table, nothing has changed.
The system has symmetry under horizontal translations. On the other
hand, if we move the particle vertically, its potential energy changes.
The state of the system is different. The system exhibits symmetry
for translations along the two horizontal axes but not for translations
along the vertical axis.
If a system that is moved forward in time behaves the same as
before, we say it is symmetric with respect to a translation in time.
(Invariance with respect to time is an important property of a clock.)

1The
name comes from the Greek word for “hand.” The word “chiral” is
pronounced “kai-ral” and rhymes with “spiral.”
 8.2. SYMMETRY AND THE LAWS OF PHYSICS 267

There is an even more complex symmetry operation involving time

called “time reversal.” Time reversal is analogous to spatial reflection.
Imagine a box full of Mexican jumping beans. Suppose you videotape
the beans while they are jumping wildly all over the place inside the
box. If you show the tape to your family, they will not be able to
tell whether it is running forward or backward. A box full of Mexican
jumping beans is symmetric under time reversal. But if you videotape
your friend jumping off a diving board into a pool, you will certainly
know if the tape is played in reverse.

8.2. Symmetry and the Laws of Physics

The laws of physics are a set of relations between physical quan-
tities (usually expressed in mathematical form) that describe how the
universe behaves. For example, F = dp dt
and the conservation of energy
are laws of physics.
Do the laws of physics exhibit symmetry? That is, do the laws
of physics remain constant under symmetry operations? Suppose I use
some equipment and verify that for a constant mass system the relation
F = ma is true in my laboratory. If I perform a spatial translation by
taking my equipment to your laboratory, will F = ma still be true? I
am sure you will agree that Newton’s second law is invariant under a
space translation. But is F = ma invariant under a reflection? If you
look at a physical system in a mirror, will Newton’s law be F = ma or
will it be F = −ma? If you recall that both F and a are polar (“honest”)
vectors, you can easily convince yourself that in the reflected system the
law is still F = ma. Would F = ma hold in a time reversed system?
(The answer is yes.) You see that this question of the symmetry of
physical laws is rapidly becoming complicated!
Perhaps the most surprising thing about symmetry and the laws of
physics is the relationship between symmetries and conservation laws.
This relationship is expressed in Noether’s theorem.2 The theorem tells
us that:
For every symmetry there is a
corresponding constant of the motion.
In the next section we shall explore this relationship using La-
grange’s equations.

2Emmy Noether (1882-1932) was one of the most important mathematical

physicists of the 20th century. She proved the theorem bearing her name in 1918.
268 8. CONSERVATION LAWS AND SYMMETRIES

8.3. Symmetries and Conserved Physical Quantities

Consider a physical system. If it is symmetrical with respect to a
rotation, then it is unchanged by the rotation. The system is the same
after it was rotated as it was before.
A physical system can be described in many ways; for example, you
could generate a table giving all of the physical properties of the system
(velocity, position, etc.) as functions of time. However, a much better
and simpler way to describe a physical system is to give its Lagrangian.
If you can write down the Lagrangian (and a set of initial conditions)
you know all the essential mechanical properties of a physical system.
You can use Lagrange’s equations to determine the equations of motion
and you can solve them to evaluate how the system will evolve in time.
Suppose the Lagrangian of some system does not contain a particu-
lar coordinate. Specifically, suppose that the angle φ does not appear in
the Lagrangian. Remember that we call such a coordinate “ignorable”
and as we showed in Chapter 4, the generalized momentum conjugate
to an ignorable coordinate is constant. The generalized momentum
conjugate to the angle φ is the angular momentum, so for this system,
the angular momentum is constant.
That is, if a system is symmetrical with respect to a rotation, then
the Lagrangian of the system does not depend in any way on φ and φ
will not appear in the Lagrangian. Consequently, the angular momen-
tum associated with rotations through φ will be conserved. Similarly,
if a system is symmetrical with respect to a translation, then its linear
momentum will be conserved.
Let us consider these concepts in more detail.

8.3.1. Conservation of Linear Momentum. Imagine a particle

in a homogeneous region of space. By spatial homogeneity we mean
that the space is the same at all points. Obviously, a particle in this
space will exhibit translational symmetry. We now show analytically
that its momentum must remain constant.
The Lagrangian for the particle cannot depend on position, so in-
stead of L = L(q, q̇, t) the Lagrangian is just L = L(q̇, t). (I am consid-
ering a single generalized coordinate for simplicity.) Lagrange’s equa-
tion is
d ∂L ∂L
− = 0.
dt ∂ q̇ ∂q
Since L does not depend on q, this relation reduces to
d ∂L
= 0.
dt ∂ q̇
 8.3. SYMMETRIES AND CONSERVED PHYSICAL QUANTITIES 269

Therefore,
∂L
= constant.
∂ q̇
But recall that
∂L
= p = generalized momentum.
∂ q̇
Therefore, p = constant and the proposition is proved.

Exercise 8.1. Show that if a system is symmetric under rotations

through some angle φ then the angular momentum conjugate to φ is
constant.

8.3.2. Conservation of Energy. Perhaps the most surprising

of the symmetry/conservation relationships involves temporal symme-
tries. A system that is symmetrical with respect to a translation in
time exhibits conservation of energy. If the system does not depend on
time in any way, the Lagrangian is not a function of time. Therefore,
instead of L(q, q̇, t) the Lagrangian will be L = L(q, q̇). Consequently,
∂L/∂t = 0. Recall that if the constraints and the transformation equa-
tions and the potential energy are all time independent, the Hamilton-
ian is equal to the total energy. If a system does not depend explicitly
on time, we expect these conditions to be met. Therefore, it is suffi-
cient to prove that the Hamiltonian is constant if the Lagrangian is not
a function of time.
But if L does not depend on t, its total time derivative will be
dL ∂L dq̇ ∂L dq
= + .
dt ∂ q̇ dt ∂q dt
By Lagrange’s equation,
∂L d ∂L
= ,
∂q dt ∂ q̇
so
dL ∂L dq̇ dq d ∂L
= + .
dt ∂ q̇ dt dt dt ∂ q̇
Now note that
 
d ∂L ∂L dq̇ dq d ∂L
q̇ = + ,
dt ∂ q̇ ∂ q̇ dt dt dt ∂ q̇
270 8. CONSERVATION LAWS AND SYMMETRIES

so  
dL ∂L dq̇ ∂L dq̇ d ∂L
= − + q̇ .
dt ∂ q̇ dt ∂ q̇ dt dt ∂ q̇
Therefore,
 
dL d ∂L
= q̇ ,
dt dt ∂ q̇
or  
d ∂L
L − q̇ = 0.
dt ∂ q̇
The definition of the Hamiltonian is
H = pq̇ − L,
∂L
where p = ∂ q̇
. Therefore the quantity in brackets is −H and
d
[−H] = 0.
dt
Consequently,
H = constant.
That is, for a system in which L is independent of the time, the Hamil-
tonian (and hence the energy) is constant.

8.4. Are the Laws of Physics Symmetrical?

So far we have seen that if a physical system exhibits some sort of
symmetry, it will be characterized by having a corresponding conserved
quantity. (Thus, we showed that if the system is symmetrical with
respect to translations, then its linear momentum is constant, and if it
is symmetrical with respect to time, its energy is constant.)
However, we have not shown that the laws of physics are invari-
ant under symmetry operations, such as translations, reflections, in-
versions, and so on.
The discussion so far could lead you to believe that the laws of
physics are invariant under any kind of transformation, but this is not
true. For example, Galileo noted that the laws of physics are not invari-
ant under a change of scale. To use his simple example, the bones of
a dog would be shaped differently if dogs were as big as houses. Feyn-
man pointed out that a model cathedral built out of toothpicks has
very different properties from a full sized cathedral. There are trans-
formations (such as change in scale) under which physical relationships
do not remain invariant. Aeronautical engineers are well aware that a
scale model airplane in a wind tunnel does not behave the same as the
full scale airplane. However, you probably expect the laws of physics
 8.4. ARE THE LAWS OF PHYSICS SYMMETRICAL? 271

to be invariant under a reflection or a translation. In fact, the force

of gravity, the electromagnetic force, and the strong nuclear force all
exhibit symmetry under reflections and translations. We have seen in
Section 3.5 that a rotating water filled bucket exhibits different behav-
ior than a bucket full of water at rest. In Chapter 13 we will see that
F = ma takes on a more complicated form in a rotating coordinate
frame.
A less well known lack of symmetry is related to the parity opera-
tion.
8.4.1. Non-conservation of Parity. The parity operation is es-
sentially a quantum mechanical operation but in classical terms we can
think of it as a reflection followed by a 180o rotation about the reflected
axis is called the parity operation. The effect of this inversion opera-
tion is to reverse each coordinate axis; in a parity operation all three
coordinates change sign:
(x, y, z) → (−x, −y, −z).
A physical system can be assigned a value for parity, either even or
odd, depending on how it behaves under a parity operation.3
This can be more easily understood in terms of a specific example.
Consider a classical system described by a position vector r and an an-
gular momentum L. The position vector has odd parity because under
the parity operation r → −r. Likewise, the linear momentum vector
p = mv has odd parity. The angular momentum vector, however, has
even parity because it is given by L = r × p. (The spin vector S also
has even parity.)
It seems reasonable that the parity of a physical system should
remain constant. The invariance of the parity of a physical system is
referred to as “conservation of parity.” For example, in the reaction
π − + d → n + n,
the total parity of the products (the two neutrons) is the same as the
total parity of the pion and deuteron. (When we speak of the parity of a
particle, we mean the parity of the quantum mechanical wave function
that describes the particle.)
In the 1950’s there was a great deal of interest in interactions in-
volving elementary particles. Physicists were surprised to discover that
3In quantum mechanics the wave function describing a system is characterized
by its parity; for example, the parity of a system with orbital angular momentum
number l is (−1)l . If the wave function is unchanged under an inversion, we say the
system has “even” parity, but if it is transformed into the negative of the original
wave function, we say it has “odd” parity.
272 8. CONSERVATION LAWS AND SYMMETRIES

parity is not conserved for particles undergoing a particular decay pro-

cess called beta decay. This discovery was related to a problem that
high energy physicists called the tau-theta puzzle. The particle known
as the τ meson was observed to decay into three pions, and the final
state had odd parity. The particle known as the θ meson decayed into
two pions and the final state had even parity. But the tau and theta
mesons were identical in all respects except for their decay products.
They had to be the same particle! The tau-theta puzzle was solved in
1956 by T. D. Lee and C. N. Yang who suggested that the tau and
theta mesons were indeed the same particle but that parity was not
conserved in the decay process. Their idea was verified by a series of
clever experiments carried out by C. S. Wu.
High energy physicists began to question some other conservation
laws. They denoted the parity operation by P. The operation called
charge conjugation in which a particle is transformed into its antipar-
ticle was denoted by C. It was found that in nature there are certain
reactions involving elementary particles in which CP is found to be
violated. (The CP operation involves changing all the particles into
antiparticles and then reflecting all the axes.)4

8.5. Strangeness
Conservation laws are used in an unusual but interesting way by
high energy physicists. In studying the elementary particles, it is found
that some reactions never occur. There is no apparent reason why a
reaction such as
π− + p → π◦ + Λ
is never observed. It does not violate any of the everyday conserva-
tion laws. But it does not happen. Using the principle that “what is
not forbidden is required,” high energy physicists decided that there
must be a conservation principle at work. They called it “conserva-
tion of strangeness.” The way it works is this: Each particle is given
a “strangeness quantum number.” For example, the strangeness of a
π particle is 0, that of a proton is also 0, and the strangeness of a Λ
particle is -1. Note that the total strangeness on the right hand side
4However, relativistic quantum mechanics shows that CPT must be invariant
in any physical process. Here the letter T stands for the time reversal process. It is
interesting to note that CP violation was confirmed experimentally at the Stanford
Linear Accelerator Center (SLAC) in 2001 based on the observation of 32 million
decay events by a 1,200 ton detector. The violation of CP explains why the universe
is predominantly made of ordinary matter rather than being 50% anti-matter. (See
Colin Macilwain, “Physicists show what really matters,” Nature, 412, 105, 12 July
2001.)
 8.5. STRANGENESS 273

of the reaction is not equal to the total strangeness on the left hand
side. Therefore, in this reaction, strangeness is not conserved and the
reaction is “forbidden.” This is not any weirder than requiring that
linear momentum be conserved during a collision. However, we feel at
home with momentum conservation because we have an analytical ex-
pression for momentum and we know that conservation of momentum
implies that no net external forces are acting on the system. We do
not have an analytical expression for strangeness nor do we know the
symmetry implied by strangeness conservation. We can be sure, how-
ever, that there is some symmetry in nature that requires strangeness
to be conserved.5
As an example of the importance of strangeness, Figure 8.1 is a
plot of strangeness vs charge for the eight spin 1/2 particles called
baryons. Note the neutron and proton on the top of the plot, both
having strangeness 0, but charge 0 for the neutron and +1 for the pro-
ton. (The lines of constant charge are skewed.) A plot just like this one
can be generated for the eight spin zero particles called mesons. These
plots (known as “The Eightfold Way”) are the basis of the standard
model of particle physics.6

n p
s=0

s=-1 Λ
Σ- Σ0 Σ+

s=-2
Ξ- Ξ0

q=-1 q=0 q=+1

Figure 8.1. The Eightfold Way: a plot of the strange-

ness and charge of the eight baryons. Note that the
charge axis is skewed.

5Ifyou wish to delve further into these questions, a good source is, Brehm, J.
and Mullin, W. J., Introduction to the Structure of Matter. John Wiley and Sons,
1989, New York, NY. Chapter 16.
6An interesting article on the history of the standard model by one of its
founders is Gerard ’t Hooft, “The making of the standard model”, Nature, 448,
271-273 (19 July 2007).
274 8. CONSERVATION LAWS AND SYMMETRIES

8.6. Symmetry Breaking

You may have heard the expression “symmetry breaking.” This is
used to describe a situation in which a symmetry ceases to exist. Con-
sider, for example, the freezing of a liquid. An atom in a liquid is as
likely to move in one direction as in another. There are no preferred
direction for the coordinate axes. There is isotropy and spatial ho-
mogeneity so the system exhibits symmetry under translation. If the
liquid freezes, the situation changes drastically because now the crystal
does have preferred directions. The space is no longer isotropic. The
symmetry between different directions has been “broken.” Theoreti-
cians interested in the evolution of the early universe believe that at
the high temperatures shortly after the big bang, the electromagnetic
interaction and the weak interaction were indistinguishable. But as the
universe cooled down there was some sort of “phase transition” caus-
ing these two forces to have vastly different magnitudes and to act on
vastly different spatial scales. This symmetry breaking is sometimes
referred to as a “freezing out” of the different kinds of forces.7

8.7. Summary
The point of this chapter was to introduce you to the idea that con-
servation laws are a consequence of symmetries in the physical universe.
We showed that momentum conservation is related to the homogeneity
of space (symmetry under translation) and energy conservation is a
consequence of temporal symmetry. The technique we used to demon-
strate this was to consider Lagrange’s equation when some coordinate
is ignorable. An ignorable coordinate in the Lagrangian implies a sym-
metry with respect to that variable.
Finally, we considered examples from elementary particle physics in
which the consequences of conservation laws are evident even if we do
not have a mathematical expression for the conservation law or even
know the form of the conserved quantity.

8.8. Problems
Problem 8.1. Consider a double pendulum (as in Chapter 4).
What physical quantities are conserved?
Problem 8.2. (a) Prove that if a certain coordinate qi does not
appear in the Lagrangian, it is also ignorable in the Hamiltonian. (b)
7For
a good discussion see Whitten, E., “When Symmetry Breaks Down,”
Nature, 429, 507-508, (3 June 2004).
 8.8. PROBLEMS 275

Show that
∂H ∂L
=− .
∂qi ∂qi
Problem 8.3. Does the law
dL
N=
dt
exhibit symmetry under reflections? Explain.
Problem 8.4. Are your footprints chiral? Does the pattern formed
by your footprints when you are walking exhibit chirality? Would chi-
rality be exhibited if the pattern were of infinite length? (Hint: What
is a “glide reflection”?)
Problem 8.5. The “Baryon Decuplet” consists of four ∆ parti-
cles with charges from -1 to +2 and strangeness (s)=0, three Σ∗ with
charges from -1 to +1 and s=-1, two Ξ∗ with charges -1, 0 and s=-2,
and one Ω with charge -1 and s=-3. Plot the decuplet on a graph like
that of Figure 8.1.
Problem 8.6. The position vector r has components (1,2,3). Ex-
press the components of this vector after (a) an inversion and (b) a
reflection in the xy plane. (c) Let r0 be the vector obtained by re-
flecting r in a plane perpendicular to the unit vector n̂. Show that
r0 = r − 2(r · n̂)n̂.
Problem 8.7. An electrmagnet consists of a current carrying coil
wrapped around an iron core, as illustrated in Figure ??. (a) What
is the direction of B in the mirror image? (b) If a positive charge is
moving towards the mirror, it will feel a force F = qv × B directed out
of the page. What is the direction of the force in the reflected system?
276 8. CONSERVATION LAWS AND SYMMETRIES

B
q
v

Figure 8.2.
 Part 2

The Gravitational Field

Chapter 9
The Gravitational Field

Classical field theory is primarily a study of electromagnetic and

gravitational fields. This chapter is an elementary introduction to field
theory and limited to a few aspects of the gravitational field.
Perhaps the most thoroughly studied field is the electromagnetic
field. You will learn about it in your E&M course. However, field the-
ory is used to study many other physical phenomena. For example,
field theory is used in fluid dynamics and in studies of elasticity, not
to mention such important reasearch areas in physics as the unified
field theory (which hopes to unite gravity and electromagnetism) and
quantum field theory. Quantum field theory arose from studies of the
power radiated by an atom when it transitions to a lower energy state.
An interesting aspect of quantum field theory is that it associates par-
ticles with the fields. Relativistic quantum field theory has become an
important tool in high energy physics. As you might imagine, field
theory is quite complex, both conceptually and mathematically. You
may be relieved to know that in this chapter we will limit ourselves to
a few fairly simple ideas.1
An interesting and important aspect of field theory is the forma-
tion and propagation of waves. When you study the electromagnetic
field you will spend a significant amount of time and mental energy
learning about electromagnetic waves. These waves are of great prac-
tical importance, particularly since X-rays, visible light, radio waves,
1To learn more about field theory and the closely related subject called potential
theory, see L. D. Landau and E. M. Lifshitz, The Classical Theory of Fields: Course
of Theoretical Physics. Vol 2.Pergammon Press, 1975. A very old but very good
book that was first published in 1929 is Oliver Dimon Kellogg, Foundations of
Potential Theory. J. Springer, 1929. (Available as a republication from Dover
Publications.)
279
280 9. THE GRAVITATIONAL FIELD

and infrared radiation are all electromagnetic waves. The gravitational

field also gives rise to waves. Einstein’s general theory of relativity pre-
dicts that gravity waves are generated by the acceleration of massive
bodies. Thus, gravity waves are expected to be produced by super-
nova explosions and by orbiting pulsars. Recently, direct detection of
gravitational waves was made with a very large (and complex) system
called LIGO.2 We will not be considering gravitational waves here.

9.1. Newton’s Law of Universal Gravitation

The concept of gravitation was put on a mathematical basis by
Isaac Newton when he formulated the law of universal gravitation.
He postulated that every particle in the universe attracts every other
particle with a force that is proportional to the product of their masses
and inversely proportional to the square of the distance between them.
In equation form, Newton’s law of universal gravitation states that the
force on a particle of mass m2 due to a particle of mass m1 is
m1 m2
F21 = − G r̂, (9.1)
|r2 − r1 |2
where G is a constant of proportionality, known as the universal gravi-
tational constant and found experimentally3 to be approximately equal
to 6.67×10−11 Nm2 /kg2 . As shown in Figure 9.1, the vectors r1 and
r2 are the position vectors of the two particles, and r̂ is a unit vector
pointing from m1 to m2 . This unit vector is given by
r2 − r1
r̂ = .
|r2 − r1 |
A slightly different way to express the law of gravitation is
r2 − r1
F21 = − Gm1 m2 . (9.2)
|r2 − r1 |3
Although this is more complicated, it avoids confusion as to the direc-
tion of the force. Recall that F21 is the force on m2 due to m1 and that
2LIGO stands for Laser Interferometer Gravitational-Wave Observatory. On
September 14, 2015, LIGO observed gravitational waves from the merger of
two black holes. More details are found in the article by B. P. Abbott
and many co-authors in Physical Review Letters, Volume 116, February, 2016,
DOI:10.1103/PhysRevLett.116.061102, and in the article by Adrian Cho in Science
Magazine, Feb 11, 2016 (DOI: 10.1126/science.aaf4041)
3A recent measurement of G by Guglielmo Tino and collaborators at the Uni-
versity of Florence, Italy, yielded a value of G = 6.67191(99) × 10−11 m3 kg−1 s−2 .
See the article “Precision measurement of Newtonian gravitational constant using
cold atoms,” by G. Rosi et al., Nature 510, 518-521, 2015.
 9.1. NEWTON’S LAW OF UNIVERSAL GRAVITATION 281

m1 ^
r
r2- r1
r1
m2
r2

Figure 9.1. A body of mass m1 is located at r1 and a

body of mass m2 is located at r2 . The relative position
of m2 with respect to m1 is r2 − r1 .

the negative sign in the equation indicates that the force is attractive.
By Newton’s third law, the force on particle m1 due to particle m2 is
r1 − r2
F12 = − Gm2 m1 .
|r2 − r1 |3
What is the force law for extended bodies? If the distance between
two bodies is much larger than the dimensions of the bodies (as for
astronomical objects), it is usually safe to assume that the bodies are
particles and the quantity |r2 − r1 | is the distance between the centers
of the two bodies. If an extended body cannot be treated as a particle,
you will need to determine the gravitational field of the extended body,
as discussed below in Section 9.3. (For example, if a satellite is in a
near-Earth orbit, the Earth cannot be considered a point mass.)

9.1.1. Universality of the Law of Gravitation. According to

Newton’s Law of Gravitation, every object (every atom) in universe
is exerting a force on every other object in the universe. It is true
that the force decreases as 1/r2 (inverse square force law), so the force
exerted on us by the distant stars is entirely negligible. Nevertheless,
a person on Earth is exerting a tiny force on all the planets and stars
and galaxies.4

9.1.2. Action at a Distance. Newton’s law of universal gravita-

tion is a prime example of the concept of “action at a distance” which
means that body A exerts a force on B, no matter how far apart the
bodies happen to be. Naturally, the question arises, “How can bodies
that are separated by great distances exert forces on one another?”
Newton stated, “Hypothesis non fingo,” which means, “I make no hy-
potheses.” This is a perfect example of physics telling us how nature
4The British physicist P.A.M. Dirac is reputed to have expressed this aspect
of the law of gravitation with the poetic sentiment, “When you pluck a flower, you
move a distant star.”
282 9. THE GRAVITATIONAL FIELD

works but making no assumptions (or hypotheses) as to why it works

that way.
The biggest problem with the action at a distance concept is that
it implies that the force between two bodies is propagated instanta-
neously. That is, if body A is moved to another position, body B will
instantaneously notice a change in the force acting on it, even if B is
halfway across the universe. This is obviously impossible. We believe
(but cannot yet prove) that the gravitational force is transmitted at
the speed of light. The conceptual difficulties inherent in Newton’s law
of universal gravitation expressed in terms of action at a distance are
effectively dealt with when one considers the gravitational field, rather
than the gravitational force.

Exercise 9.1. Determine the force exerted by Mars on the Sun.

What is the acceleration of Mars? What is the acceleration of the
Sun?
Exercise 9.2. A body falls in the gravitational field of the Earth.
Show that its acceleration is independent of its mass.
Exercise 9.3. A particle of mass m is located at (1,2), a particle
of mass 2m is located at (3, 4), and a particle of mass 3m is located
at (−2, −2). Determine the gravitational force acting on the particle of
mass 2m. Answer: F = −Gm2 (0.24ı̂ + 0.26̂).
Exercise 9.4. Prove that the gravitational force is conservative by
evaluating the curl of Newton’s Law of Universal Gravitation.

9.2. The Gravitational Field

By definition, a field is a physical quantity that is defined at ev-
ery point in some region of space. For example, the temperature is a
physical quantity and using a thermometer you could determine the
temperature at every point in this room. This scalar field might be
described by an equation T = T (x, y, z), or by a table of values, or by
a drawing showing the temperature isopleths of the room. Similarly,
the water in a river has a velocity at every point. The region of space is
the river and the physical quantity is velocity. This vector field might
be represented graphically or perhaps by an equation for the velocity as
a function of position. Suppose that somehow you could determine the
 9.2. THE GRAVITATIONAL FIELD 283

gravitational force on a mass at all points near a planet. This would

define the gravitational force field of the planet. Note that in all of
these examples, the value of the field depends on position.
To define the gravitational field more explicitly, consider a region
of space that is empty except for two particles. One particle has mass
M and it is located at a fixed point in space. At some other point you
place a particle of much smaller mass m. Assume you have a way to
determine the force acting on the small mass, wherever it may happen
to be. The small mass will be the “test body.”
Now imagine you move this test body from one point to another
and you measure the magnitude and the direction of the gravitational
force acting on it at every location. If you represent the force by a
vector, you will end up with a lot of force vectors all pointing towards
M. Unfortunately, as indicated by Equation (9.1), these force vectors
depend on the mass of the test body so they are not a good measure of
the gravitational field of M. To avoid any dependence on the test body,
you can define the gravitational field of M as the force per unit test
mass at all points in space around M. Furthermore, since you don’t
want m to have any influence on the field, you take the limit as m → 0.
These concepts are expressed more simply using mathematics. Let
F be the force on m due to M and let g be the gravitational field of
M. Then, by definition, if m is at r and M is at r0 ,
r − r0
 
F
g(r)= lim = −GM . (9.3)
m→0 m |r − r0 |3
The notation g = g(r) reminds us that g is a function of position.

M = source point
r-r'
r'
P = field point
r

Figure 9.2. Particle M at the source point r0 generates

a gravitational field everywhere. The gravitational field
at point P (r) (the “field point”) is given by Equation
9.3.

The location of body M is called the “source point,” and its position
is denoted by r0 . The point specified by r is called the “field point.” It
is the place where the field is evaluated. See Figure 9.2. Note that the
test mass does not enter into Equation (9.3), the expression for g(r).
284 9. THE GRAVITATIONAL FIELD

In introductory physics books it is usual to place the origin of co-

ordinates at the mass M. This gives a simpler but less descriptive ex-
pression for the field:
r r̂
g(r) = − GM 3 = −GM 2 , (9.4)
|r| r
where r is the location of the field point.
Let me be very clear on the difference between the force and the
field. The force F depends on the masses of both of the particles and
their positions at some instant of time. The field g depends only on
the mass of the source particle M and the relative position of the field
point. The force F describes an interaction between two objects, but
the field g is a property of a point in space.5
The point mass M is the source of the gravitational field. The field
extends throughout all space. It is not easy to grasp the field concept;
it is particularly difficult to describe exactly what is filling all of space.
We know that if we place a test mass at any point in the field it will
feel a force, so we think of the mass M as producing something which
exerts a force on any other mass in this space. The field is everywhere,
but it exerts a force only when a material body is placed in the field.
At the surface of Earth the gravitational field is approximately given
by g = − 9.8 m/s2 k̂. Here k̂ is a unit vector pointing upward at the
surface. For points far above the surface, the field is approximately
g(r) = −(GME /r2 )r̂ where the origin is at the center of the Earth and
ME is the mass of the Earth. (Do not confuse the field vector g(r)
with the constant value 9.8 m/s2 which is usually abbreviated by g.)
The field approach to gravity is quite different from the action at
a distance approach. In the action at a distance approach we consider
the force one body exerts directly on another. When using the field
concept, we think of the interaction of two bodies (M and m) as a
two step process in which mass M generates a field and then mass m
interacts with the field, rather than interacting with M directly. Thus
the field approach decouples the sources from the test body used to
determine the field. If two different source arrangements generate the
same field at some point, a test body at that point will feel the same
force.
It is tempting to consider the field to be nothing but a convenient
way of expressing the force, but it turns out that the field has an
5Sometimes,I like to change the definition of field slightly and state, “A field
is a region of space in which some physical quantity is defined at every point.”
This focuses the mind on the space rather than on the physical quantity. It is not,
however, the standard definition of field.
 9.2. THE GRAVITATIONAL FIELD 285

existence of its own, independent of the force. A force is just a force,

but a field has energy, momentum, and angular momentum as well as
the ability to exert a force. A field can propagate through space as a
wave and can exist independent of the source. (The field can persist
even after the source has ceased to exist.)
Although the field approach seems more complicated because it
introduces a new concept (“the field”), it does resolve a number of
difficulties with the action at a distance concept. For example, in the
action at a distance scenario, forces are transmitted instantaneously.
In the field concept, the response of the field to changes in the source
propagate at a finite speed. Since the source (M ) generates the field, it
is obvious that if the source is moved, the field will change. However,
this change need not occur instantaneously at all points in space. As
mentioned before, we believe that changes in the gravitational field are
propagated at the speed of light.
Furthermore, the field approach leads to a conceptual framework for
understanding how the force is transmitted. In quantum field theory
the interaction between particles can be described as an interchange of
“virtual” particles. The electromagnetic force, for example is believed
to be transported by virtual photons. This is often represented graphi-
cally by “Feynman diagrams.” For example, the electromagnetic inter-
action between two electrons (“electron-electron scattering”) is repre-
sented by the Feynman diagram of Figure 9.3. In the figure, the two
electrons are represented by the straight lines and the wavy line repre-
sents a virtual photon where the word “virtual” is used because these
photons are not (and cannot be) observed. The photon transmits the
electromagnetic force. It is emitted by one of the electrons and ab-
sorbed by the other. Both particles are deflected from their original
paths. Similarly, the gravitational force is assumed to be transmitted
by undetected particles called “gravitons.” Since we have our hands full
with just Newton’s theory of gravitation, we will not consider any of
these advanced concepts in this book. However, if you are interested in
delving deeper into quantum field theory, I recommend you read a very
short but very profound book by Richard Feynman called “QED.”6

Exercise 9.5. Two particles of mass M are separated by a dis-

tance a. Determine the gravitational field at a distance z along the

6Richard Feynman, QED: The Strange Theory of Light and Matter. Princeton
University Press, Princeton, NJ, 1985.
286 9. THE GRAVITATIONAL FIELD

e e

Figure 9.3. A Feynman diagram for electron-electron

scattering. The wavy line represents the virtual photon.

perpendicular bisector of the line joining the masses. Answer: g =

3/2
−2GM zẑ/ [(a/2)2 + z 2 ]
Exercise 9.6. A point mass M is located at the origin. Another
point mass M is located at (0,4). Determine the gravitational field at
(0,3) and at (3,0). Answer: Field at (3,0) is g = GM [−0.14ı̂ + 0.03̂] .
Exercise 9.7. Determine the gravitational field due to the Earth
and the Sun at the position of the Moon, (a) at the time of a lunar
eclipse, (b) at the time of a solar eclipse. (c) Determine the force on
the Moon at these two times.

9.3. The Gravitational Field of an Extended Body

The gravitational field produced by an extended body can be de-
termined by treating the body as a collection of point masses and sum-
ming the fields due to each individual particle. (Gravitational fields
obey the principle of superposition and add vectorially.) Although this
is a perfectly reasonable thing to do, it is not a practical thing to do.
So we assume that the material of the extended body is continuous.
This assumption breaks down on the atomic level where there are huge
differences in mass density between the small, heavy nuclei and the
empty space that comprises most of the volume of atoms. However, a
volume element that contains billions of atoms is still very small on the
macroscopic scale and we are justified in thinking of it as containing a
continuous distribution of matter with a well defined density.
Figure 9.4 shows an infinitesimal volume element (dτ 0 ) in a contin-
uous body. The mass contained in this volume element is dm = ρdτ 0
where ρ = ρ(r0 ) is the density. (For a surface, the element of mass is
σdA, for a line it is λds and for an extended body it is ρdτ, where σ, λ, ρ
are the mass per unit area, per unit length, and per unit volume.) The
 9.3. THE GRAVITATIONAL FIELD OF AN EXTENDED BODY 287

density may vary from one part of the body to another so it is written
as ρ = ρ(r0 ) to remind us that ρ is a function of position.
In Figure 9.4 the point P is the field point. Note that the source
point is at r0 and the field point is at r. The infinitesimal portion of
the gravitational field at r due to an infinitesimal mass element dm at
r0 is
r − r0
dg(r) = − Gdm .
|r − r0 |3
or
r − r0
dg(r) = − Gρ(r0 )dτ 0 ,
|r − r0 |3
where dτ 0 is so small that ρ(r0 )dτ 0 can be considered a particle, yet
large enough to treat the mass as continuous. To obtain the field at P
due to the whole body, we integrate over the body. That is,
0
0 r−r
Z
0
g(r) = −G ρ(r ) 0 3 dτ . (9.5)
body |r − r |

dm' = ρdτ'
r - r'
dg
P
r'
r

Figure 9.4. An extended body. The mass element

dV 0 = ρdτ 0 is located at r0 (the source point) and it
produces an infinitesimal gravitational field dg at point
P located at r (the field point).

For an extended body with sufficient symmetry, it is usually not

too difficult to evaluate this integral. To find the field due to an arbi-
trarily shaped body, it is usually best to first determine the potential
and then obtain the field, using the technique described in Section 9.4
below. Therefore, I will not emphasize the direct integration of Equa-
tion (9.5). Nevertheless, you should have some experience with this
technique so I will work through two examples. You can get some
additional experience by working out the exercises below. It might
help you to review the section on the electric field of continuous charge
distribution in your introductory physics textbook, as the techniques
are exactly the same. It is usually easier to solve such problems by
288 9. THE GRAVITATIONAL FIELD

resolving the vector dg into its components before carrying out the
integration.

Worked Example 9.1. A long thin rod of length L lies along

the x axis. The end points of the rod are at x0 = 0 and x0 = L.
Assume the rod can be represented as a continuous mass distri-
bution with linear density λ mass per unit length. Determine the
gravitational field at the point x = 4L.
Solution: In this problem, the mass element is dm = λdx0
where the variable x0 represents the distance from the origin to the
mass element. The infinitesimal field at P due to this portion of
the rod is
r − r0 x − x0
dg = −Gdm = Gdm
|r − r0 |3 |x − x0 |3
(x − x0 )ı̂ dx0
= −Gλdx0 = −Gλı̂ .
|x − x0 |3 (x − x0 )2
The field point is x = 4L, so integrating we have
Z x0 =L L
dx0

1 1
g = −Gλı̂ 0 2
= −Gλ 0
ı̂ = − λGı̂.
x0 =0 (4L − x ) 4L − x 0 12L

Worked Example 9.2. Determine the gravitational field at

a point P a distance +z along the axis of a very thin disk of mass
M, uniform density, and radius R. See Figure 9.5.
Solution: Treat the disk as if it had zero thickness. Then
the appropriate mass element is σdA where σ = M/πR2 . As indi-
cated in Figure 9.5 the field at point P due to mass element dm is
the vector dg pointing towards dm. The distance from the source
point dm to the field point is denoted D. Consider all the mass
elements lying on a ring a distance r from the axis. The vectors
from P to these mass elements form a cone of vectors with vertex
at P. The horizontal components of these vectors will cancel. The
components along the axis will sum. The resultant field is given
by
Z Z Z
Gdm σdA
g = (dg) cos θk̂ = − 2
cos θk̂ = −G cos θk̂.
D D2
 9.4. THE GRAVITATIONAL POTENTIAL 289

But cos θ = z/D = z/(r2 + z 2 )1/2 , and dA = rdφdr, so

Z R Z 2π Z R
(dr)(rdφ) rdr
g = −σGk̂ z 2 2 3/2
= −σGk̂(2π)z 2 2 3/2
r=0 φ=0 (r + z ) 0 (r + z )
Z u=R2 +z2  
du 1 1
= −σGk̂(π)z = −2πσGk̂z −√
u=z 2 u3/2 z z 2 + R2
 
2GM z
= − 2 1− 2 k̂.
R (z + R2 )1/2

P
dg

θ d
dr
D
r
φr

Figure 9.5. The gravitational field of a thin disk is de-

termined by integration of the elementary field vectors
dg.

Exercise 9.8. Determine the gravitational field at a point z on the

axis of a ring of mass M and radius R. Answer: −GM z/(z 2 + R2 )3/2 .
Exercise 9.9. Show that a disk can be treated as a collection of
concentric rings. Write an integral expression for g using the result of
Exercise 9.8 and show that it yields the expression obtained in Worked
Example 9.2.

9.4. The Gravitational Potential

The gravitational potential is a scalar quantity related to the grav-
itational field which is a vector quantity. It is usually easier to work
with the potential than the field; if you know the potential you can
always determine the field.
Recall the discussion in Section 5.3.3 proving that if F is a conser-
vative force, then ∇ × F = 0. It is easy to show that the gravitational
290 9. THE GRAVITATIONAL FIELD

force is conservative. (See Exercise 9.4 above.) Using the fact that the
curl of the gradient of any scalar function is zero, we conclude that
since ∇ × F = 0, then F can be expressed as the gradient of some
scalar function. That is, we can define the potential energy V = V (r)
such that
F = −∇V.
Then
Z r2 Z r2 Z 2
F · dr = − ∇V · dr = − dV = V1 − V2 .
r1 r1 1

Thus, the difference in potential energy between two points is defined

by
Z r2
V2 − V1 = − F · dr.
r1

In working with the gravitational potential energy, it is often convenient

to define the zero of potential energy at infinity, so let r1 = ∞ and
V1 = 0. Writing r for r2 we have
Z r
V =− F · dr.
∞

If F is the gravitational force acting on test body m located a distance

r from point mass M, you can write
Z r
GM m Mm
V (r) = − − 2 r̂·dr = −G . (9.6)
∞ r r
This is the gravitational potential energy of a system of two point
masses.
When using the field approach, it is convenient to define a quantity
called the “gravitational potential” which we will denote by the sym-
bol Φ. The gravitational potential is obtained from the gravitational
potential energy by dividing by m, the mass of the test body. The
potential energy (V ) of two interacting masses is an example of action
at a distance and depends on the presence of two bodies. The potential
(Φ) is defined at every point in space and depends on the presence of
just one body. Thus, Φ = Φ(r) is a field. The potential is related to
the potential energy by
V
Φ(r) = lim ,
m→0 m

where m refers to an infinitesiml test mass located at r.

 9.4. THE GRAVITATIONAL POTENTIAL 291

Dividing Equation (9.6) by m shows that the gravitational potential

at r due to a particle of mass M located at the origin, is
GM
Φ(r) = − . (9.7)
r
The potential of an extended body is obtained from
ρ(r0 )dτ 0
Z
Φ(r) = −G 0
. (9.8)
body |r − r |

where r is the field point and r0 is the location of the source point
dm = ρ(r0 )dτ 0 .

Figure 9.6. The potential at P due to a shell of radius

a. The mass element dm is the mass of the element of
area dA = a2 sin θdθdφ. If σ is the surface mass density,
dm = σdA.

Worked Example 9.3. Determine the gravitational potential

at a point outside of a spherical shell of mass M and radius a. See
Figure 9.6
Solution: The mass per unit area (σ) of the (uniform) shell is
the total mass of the shell divided by its surface area, or
M
σ= .
4πa2
Selecting the mass element is more or less an “art” and you will
become proficient at making a good choice after you have worked
292 9. THE GRAVITATIONAL FIELD

out a number of problems. For a mass shell, pick a small element

of area on the surface whose mass is
dm = σdA = σa2 sin θdθdφ.
Integrating over φ generates a ring on the surface of the shell, as
indicated in the figure. (Integrating over φ is equivalent to letting
the line a sin θ rotate around r. Note that all points on this ring
are equidistant from the field point.) The potential at P due to
the entire shell is given by the double integral

2π π
σa2 sin θdθdφ
Z Z Z
σdA
Φ = −G = −G ,
surface |r − r0 | φ=0 θ=0 |r − r0 |
Z 2
σa sin θdθ
Φ = −2πG ,
|r − r0 |
where the factor 2π came from integrating over φ. Using the law
of cosines, the figure shows that since |r0 | = a,
√
|r − r0 | = R = a2 + r2 − 2ar cos θ,
and R2 = a2 + r2 − 2ar cos θ. Therefore 2RdR = 2ar sin θdθ and
the integral takes the form
Z π 2 Z R=r+a
σa sin θdθ 2 (R/ar)dR
Φ = −2πG = −2πGσa
0 R R=r−a R
Z r+a
−2πGσa
= −Gσa2 (2π/ar) dR = [(r + a) − (r − a)]
r−a r
4πa2 σ GM
= −G =− ,
r r
where we used the fact that as θ goes from 0 to π, R ranges from
r − a to r + a.
Our result shows that for points outside the shell, the potential
of a shell of mass M is the same as the potential of a point mass
M at the origin. As a problem you can show that the field inside
the shell is zero.

Exercise 9.10. Determine the potential on the axis of a disk of

thickness t. Assume that t << z so the thin disk approximation is still
valid. (Answer: Φ = −2πρtG[(z 2 + R2 )1/2 − z].)
 9.5. FIELD LINES AND EQUIPOTENTIAL SURFACES 293

Figure 9.7. The force vectors are represented on the

left. The strength of the force is represented by the
length of the arrows. On the right, the field lines give
the direction of the force and the density of field lines at
any point is proportional to the strength of the field.

9.5. Field Lines and Equipotential Surfaces

Going back to the gravitational field (g) due to a point mass M ,
recall that g is the force per unit mass on an infinitesimal test body (See
Equation 9.3). You could, in principle, transport the test body to every
point in space around M and measure the force on it at each location.
Then you could represent the field g in a number of different ways. For
example, you could make up a large table giving the magnitude and
direction of g at every point. Or you could represent g by an equation.
(It would be Equation 9.4.) Or you could represent g graphically. The
graphical representation would show many arrows pointing toward M.
The field at some given point might be represented by drawing a vector
of appropriate length at that point, as illustrated by the sketch on the
left in Figure 9.7. Note that the vectors are longer at points close to
the mass and shorter at points further away. But this is not a very
good way to represent the field. The sketch on the right in Figure
9.7 illustrates a much more convenient way. In this representation the
lines give the direction of the force and the areal density of the lines
at a point gives the magnitude of the force at that point. Imagine the
lines are drawn in such a way that the number of lines crossing a unit
area perpendicular to the field is numerically equal to the field at that
location. If the field at some point has a magnitude of 15 N/kg, it
could be represented by drawing 15 lines (per meter squared) at that
point. The same number of lines go through the surface of any sphere
centered on M. Since the area of a sphere increases as r2 , the density
294 9. THE GRAVITATIONAL FIELD

of lines per unit area falls off as 1/r2 , just like the gravitational field.
The lines representing the field are traditionally called “lines of force”
although most modern books refer to them with the better terminology
of “field lines.”
Another graphical representation of the field consists in sketching
the surfaces on which the potential is constant. For a point mass,
the equipotential surfaces are concentric spheres centered on the mass.
As we will show in a moment, the field and the potential are related
by g = −∇Φ, and you know from Chapter 5 that the gradient of
Φ is perpendicular to the surfaces of constant Φ. Therefore, the two
representations (field lines and equipotential surfaces) are equivalent.

9.6. The Newtonian Gravitational Field Equations

The relationship between potential energy and force is F = −∇V
(Equation 5.7). Dividing both sides of this equation by m yields the
relation between the gravitational field g(r) and the gravitational po-
tential Φ(r).
g = −∇Φ. (9.9)
(In many books - particularly older books - the gravitational potential
is defined as the negative of Φ and the relation equivalent to Equation
(9.9) does not have the minus sign.)
The curl of a gradient is zero. Therefore,
∇ × g = −∇ × (∇Φ) = 0. (9.10)
(I am sure you expected this result because you know that g is a con-
servative field.)
Having found the curl of g, it is reasonable to go on and determine
the divergence of g. The process is a bit involved, so please follow
carefully. Consider a mass point M. Draw a surface S around it, as
illustrated in Figure 9.8. Let dS be an infinitesimal patch of the surface.
Let n̂ be a unit vector perpendicular to dS and let r be the vector from
M to dS.
Let the origin of our coordinate system be located at the point mass
M . Then the gravitational field of M at the position of the patch of
surface dS can be written as
GM
g = − 2 r̂,
r
so
GM GM
g · n̂ = − 2 r̂ · n̂ = − 2 cos θ
r r
where θ is the angle between r̂ and n̂ (see Figure 9.8).
 9.6. THE NEWTONIAN GRAVITATIONAL FIELD EQUATIONS 295

^
n dS
S ^
n
dS
θ
r r
M dS

Figure 9.8. A closed surface S encloses a point mass M.

The unit vector n̂ is perpendicular to the surface at the
location of dS, an infinitesimal portion of the surface. Let
dS⊥ be the projection of dS perpendicular to r. Note that
dS ·r̂ =dSn̂ · r̂ =dS cos θ = dS⊥ . The sketch on the right
views the areas dS and dS⊥ edge-on. The angle θ is the
angle between the radial vector and n̂, or equivalently,
between dS and dS⊥ .

Note that the solid angle7 dΩ subtended by dS at M is

dS⊥ dS cos θ
dΩ = = , (9.11)
r2 r2
where dS⊥ is the perpendicular projection of dS on r. Next consider
the surface integral
I I
GM
I = g · n̂dS = − 2
cos θdS
S S r
I I
dS cos θ
= −GM = −GM dΩ
S r2 S
= −4πGM.
This result is very interesting. It tells us that the surface integral I is
proportional to the mass M inside the surface. This fact, in itself, is
not too surprising. However, the result also tells us that it does not
matter where the mass is, as long as it is inside the closed surface. If
there are several masses (m1 , m2 , · · · ) inside S, then, by the principle

7The solid angle is the three dimensional angle formed at the vertex of a cone.
Let point O be at the origin and a surface of area dA be a distance r from the
origin. Then, if dA is perpendicular to r, the solid angle subtended by dA at O is
dΩ = dA/r2 . The units of solid angle are steradians. The solid angle subtended by
a spherical shell at a point inside the shell is 4π steradians.
296 9. THE GRAVITATIONAL FIELD

of superposition,
I X X
g · n̂ dS = − 4πGmi = −4πG mi = −4πGMenc , (9.12)
S
where Menc is the total mass enclosed by S. The equation
I
g · n̂ dS = −4πGMenc
S
is called Gauss’s Law. It is particularly useful in the study of electro-
statics.
Assume mass is continuously distributed in a volume V bounded
by a closedR surface S. The element of mass is ρdτ , so the mass enclosed
is Menc = V ρdτ. Therefore,
I Z
g · n̂ dS = − 4πGρdτ. (9.13)
S V
You may recall from your vector analysis course that Gauss’s divergence
theorem states that for any vector A,
I Z
A · n̂ dS = ∇ · A dτ. (9.14)
S V
Therefore,
R the left hand side of equation (9.13) can be expressed as
V
∇ · gdτ, and consequently
Z Z
∇ · g dτ = − 4πGρdτ,
V V
or
∇ · g = − 4πGρ. (9.15)
Thus, the divergence of g is proportional to ρ, the mass density. An
important theorem called the Helmholtz theorem states that for any
vector field, if one knows the divergence and the curl of the field, then
one can determine the field itself. For this reason, ∇ · g and ∇ × g are
called the “sources” of g. Equations (9.10) and (9.15) indicate that the
source of the gravitational field is the mass density ρ. In other words,
masses generate gravitational fields.
Worked Example 9.4. Use Gauss’s law to determine the
gravitational field outside of an infinitely long cylinder of radius a
with constant linear mass density λ.
Solution: To solve we construct a Gaussian surface which is a
closed surface that has the same symmetry as the mass distribution.
We select this surface so that on the surface g · n̂ is either constant
or zero. The field point lies at an arbitrary point on this surface.
 9.6. THE NEWTONIAN GRAVITATIONAL FIELD EQUATIONS 297

P = field point

r
a

Figure 9.9. A Gaussian surface. The outer cylinder

is the Gaussian surface appropriate for determining the
field of the cylindrical mass. Note that the field point is
on the Gaussian surface.

For this problem the appropriate Gaussian surface is a cylinder of

length L and radius r, where r > a.
On the curved face of this “tin can” surface, the field is directed
towards the axis of the mass element, and g is constant. On the
two end faces, g and n̂ are perpendicular to each other. Therefore,
Gauss’s law gives
I Z Z
g · n̂ dS = g · n̂ dS + g · n̂ dS = −4πGMenc .
S side ends
Now Z Z
g · n̂ dS = −g dS = −g(2πr)L,
side
and Z
g · n̂ dS = 0.
ends
The mass enclosed by the Gaussian surface is
Menc = λL.
Therefore,
4πGλL 2Gλ
g=− =− .
2πrL r
By symmetry, this is pointing toward the axis of the cylinder so
2Gλ
g=− ρ̂
r
where ρ̂ is the unit vector in cylindrical coordinates.

Exercise 9.11. Show that the solid angle subtended by a spherical

shell at any interior point is 4π steradians.
298 9. THE GRAVITATIONAL FIELD

Exercise 9.12. Use Gauss’s law to determine the gravitational field

of a point mass.
Exercise 9.13. Maxwell’s equations for the electric field (E) and
the magnetic field (B) can be written
∂B
∇ · E = ρ/0 ∇×E=−
∂t
∂E
∇·B = 0 ∇ × B = µ0 J+0 µ0
∂t
where ρ is the charge density and J is the current density. The quanti-
ties 0 and µ0 are constants. What are the sources of the electric field?
What are the sources of the magnetic field?

9.7. The Equations of Poisson and Laplace

Recall that g = − ∇Φ. Taking the divergence of both sides,
∇ · g = −∇ · ∇Φ= −∇2 Φ.
Using the relation ∇ · g = −4πGρ leads to
∇2 Φ = 4πGρ. (9.16)
This is called “Poisson’s equation.”
Poisson’s equation is a second order partial differential equation
whose solution depends on the boundary conditions. In a region of
space where there is no mass, the density ρ is zero, and Poisson’s
equation reduces to
∇2 Φ = 0. (9.17)
This very important special case is called “Laplace’s equation.”
Most of the time you will want to determine the field in a region of
space where ρ = 0, for example, in empty space outside of a star, and
you will use Laplace’s equation. If, however, you want to determine Φ
inside the star then you have to solve Poisson’s equation.
Since Poisson’s equation is an inhomogeneous differential equation,
its solution can be expressed as the sum of two parts:
• The general solution of the homogeneous differential equation
(Laplace’s equation).
• A particular solution of the inhomogeneous differential equa-
tion (Poisson’s equation).
 9.8. EINSTEIN’S THEORY OF GRAVITATION (OPTIONAL) 299

In your undergraduate course in electromagnetic theory you will

learn techniques for solving the Laplace equation under a variety of
different conditions. In your graduate course in electromagnetic theory
you will learn a number of methods for solving the Poisson equation.
Poisson’s equation, ∇2 Φ = 4πGρ is the “field equation” of the New-
tonian gravitational field. If you solve for Φ you know the potential
everywhere and you can determine the gravitational field g everywhere.
In the next section we shall consider the field equations for the gravi-
tational field according to Einstein.

9.8. Einstein’s Theory of Gravitation (Optional)

We cannot leave this chapter on the gravitational field without a
word or two about Einstein’s theory of gravitation, which is referred to
as The General Theory of Relativity. As you know, this is one of the
greatest achievements in physics and supersedes Newton’s theory of
gravitation. For example, it was known that the orbit of Mercury pre-
cesses at the extremely precisely measured rate of 5601 seconds of arc
per century. Newton’s theory, including the effects of the other planets
which perturb the motion of Mercury, yielded a precession rate of 5558
seconds of arc per second. The discrepancy (43 seconds per century)
was explained with great precision by Einstein’s theory. Note that
we are talking about a difference of less than one degree per century!
There are numerous other examples of situations in which Einstein’s
theory gives the correct value whereas Newton’s theory is slightly off.
In that case, you might ask, why do we bother studying Newton’s
theory? The answer is that calculations in Einstein’s theory are ex-
tremely complicated, and that Newton’s theory gives results that are
extremely close to the correct values.
This section will give you a rough idea of Einstein’s theory.8 You
will probably not be using the theory of general relativity unless you
decide to specialize in cosmology - the study of the structure of the
universe. Nevertheless, as a physicist, you should have some idea about
the theory.
You may be aware that in 1905 Einstein developed the theory called
the “Special Theory of Relativity.” (We will consider this theory in
Chapter 19.) The special theory deals with the relationship between
two inertial reference frames. That is, it considers two reference frames
that are moving at constant velocities. Clearly, the laws of physics have

8Avery readable book on the topic is “Relativity, Gravitation and Cosmology”

by Robert J. A. Lambourne, Cambridge University Press, 2010.
300 9. THE GRAVITATIONAL FIELD

to be the same in the two reference frames. In other words, a trans-

formation from one reference frame to the other should leave physical
equations and relationships unchanged. The theory was extremely suc-
cessful and was particularly applicable to electromagnetism. However,
Einstein was unsatisfied with one aspect of the special theory, and that
was that Newton’s law of universal gravitation did not transform as it
should. Furthermore, Einstein was interested in developing a theory
that would be applicable to accelerated (that is, non-inertial) reference
frames. He finally developed the “general” theory of relativity that
solved both of these problems. To do so, however, he had to delve
deeply into very advanced concepts in geometry.
Einstein’s theories required understanding how to transform from
one reference frame to another, a procedure that has much in common
with transforming from one coordinate system to another. Recall that
in Section (5.3.4) we introduced the concept of the “metric tensor”
that allowed us to transform from Cartesian coordinates to spherical
or cylindrical coordinates. Letting hij represent the elements (or com-
ponents) of the metric tensor, we wrote the differential (infinitesimal)
displacement as X
ds2 = h2ij dqi dqj ,
i,j
(see Equation 5.8). In Sections 5.3.5 and 5.3.6 we used this relation
to obtain expressions for the volume element and for del in cylindrical
and spherical coordinates.
In special relativity, Einstein had to use a 4-dimensional coordinate
system with x, y, z being the usual Cartesian space coordinates and the
fourth coordinate being ct where c is the constant speed of light. Note
that ct has the dimensions of distance, but it depends only on t. Thus,
time is the fourth dimension. Including an extra dimension requires a
somewhat more complicated expression for the displacement as
3
X
2
ds = ηµν dxµ dxν .
µν=0

There are two things to note about this expression. First of all it is
customary to use upper and lower indices which are called “contravari-
ant” and “covariant”. The difference need not concern us at present,
but it is important to appreciate that dxµ does NOT mean dx raised to
the power µ! It is just a way to denote the component9. Secondly, note
9Superscript
indices are denoted contravariant and subscript indices are de-
noted covariant. The difference between covariant and contravariant vectors is the
way they transform. The components of a covariant vector transform according
 9.8. EINSTEIN’S THEORY OF GRAVITATION (OPTIONAL) 301

that the indices µ and ν range from 0 to 3. Here 1,2,3 represent the
spatial coordinates, dx, dy, dz and zero denotes the “time” coordinate
cdt. Thus, dx3 represents dz and dx0 represents cdt. The quantity ds is
the “line element” in 4-dimensional “spacetime.” In special relativity,
the metric ηµν is quite simple. It is called the “Minkowski metric10”
and in matrix form, it can be written as
 
1 0 0 0
 0 −1 0 0 
ηµν =  0 0 −1 0 

0 0 0 −1
Now Minkowski spacetime is described as “flat” because Euclidean
geometry is valid in this space. This means, for example, that parallel
lines never meet and that the angles in a triangle add up to 180 degrees.
Note that the geometry of the surface of a sphere is non-Euclidean.
Parallel lines that start out perpendicular to the equator, meet at the
pole. The angles of a triangle may not sum to 180 degrees. In 3-
dimensional space, the line element ds on a flat surface is
ds2 = dx2 + dy 2 + dz 2 ,
but on the surface of a sphere of radius R,
ds2 = R2 dθ2 + R2 sin2 θdφ2 .
Not all surfaces that we might think of as curved are curved in the
mathematical sense. For example, the curved side of a cylinder is
mathematically flat because (as you can easily show) parallel lines never
meet, triangle have 180 degrees, etc.
Bernhard Riemann11 used the concept of line element to generalize
geometry. He expressed the generalized line element in a curved 3-
dimensional space as
3
X
2
ds = gij dxi dxj ,
i,j=1

∂xj
to A0i = ∂x0j Aj , whereas the components of a contravariant vector transform as
∂x0j
A0i = ∂xj Aj .
10This metric was developed by Hermann Minkowski who was one of Einstein’s
mathematics professors. Minkowski is credited for having developed the concept of
spacetime.
11Bernhard Riemann (1826-1866), who was a student of Carl Friedrich Gauss, is
considered one of the worlds greatest mathematicians. Among other achievements,
he developed the geometry of curved surfaces.
302 9. THE GRAVITATIONAL FIELD

where the gij are the elements of the metric tensor (or “metric coef-
ficients”). The metric relates the coordinate differentials (the dxi ) to
a length ds in the space under consideration. Therefore, the metric is
related to the geometry of the space. Once the metric is known, the ge-
ometry of the space is entirely determined. As we shall see, the metric
can not only tell us whether the space is flat or curved, it can actually
give us the curvature of the space.12 In fact the curvature is given by
a complicated quantity called the Riemann tensor that is defined as

l ∂Γlik ∂Γlij X m l X
Rijk ≡ − + Γik Γmj − Γm l
ij Γmk
∂xj ∂xk m m

where
 
1 X il ∂glk ∂gjl ∂gjk
Γljk = g + k + .
2 l ∂xj ∂x ∂xl

The Riemann tensor is rank 4 (as you can tell by noting that there
are four indices) and Γlmk is a rank 3 tensor. Obviously, this is a very
complicated quantity. You will probably never have to evaluate it un-
less you become a theoretical physicist specializing in general relativity!
l
Nevertheless, it is worth noting that for a flat space, Rijk = 0 every-
where. Now so far we have been considering the geometry of ordinary
3 dimensional space (as indicated by our use of latin letters as indices).
Einstein’s job was to generalize to 4 dimensional spacetime. Thus, in
a curved Riemannian 3-D space,
3
X
ds2 = gij dxi dxj ,
i,j=1

where ds2 is positive and the gij are functions of the coordinates,
whereas in a flat 4-D Minkowski spacetime
3
X
2
ds = ηµν dxµ dxν
µ,ν=0

where the ηµν are constants equal to (1, −1, −1, −1) and ds2 may be
positive, negative or zero. In order to treat curved spacetime, the
constant ηµν are replaced by the functional quantities gµν . That is, in

12It
might be mentioned that most of the metrics that are encountered in
physics are orthogonal, and as we have seen in Chapter 5, this means the metric
tensor can be expressed as a diagonal matrix.
 9.8. EINSTEIN’S THEORY OF GRAVITATION (OPTIONAL) 303

a curved 4 dimensional spacetime the line element is

3
X
2
ds = gµν dxµ dxν
µ,ν=0

In a flat region of spacetime, the metric reduces to the Minkowski met-

ric. It turns out that even if spacetime is curved overall, it is possible
to define a small neighborhood about any point in which the spacetime
can be considered to be flat. (This is similar to our experience on the
curved surface of the Earth; in a small neighborhood of your location,
the Earth can be treated as if it were flat.) Thus, in a small enough
neighborhood of any point, special relativity is valid. Another way of
saying this is that special relativity applies locally but not globally.
We now consider some consequences of assuming that spacetime
is curved. Einstein was aware of two problems with Newton’s law of
universal gravitation with respect to special relativity. First, as we
mentioned before, the gravitational force law does not transform as it
should under a change in inertial reference frame, and secondly the
concept of action at a distance and the instantaneous transmission of
gravitational forces is not consistent with special relativity which is
based on the postulate that nothing can travel faster than the speed of
light.
Einstein’s new theory required that the laws of physics be expressed
as tensors that are invariant under transformations from one acceler-
ated reference frame to another. Newton’s law of universal gravitation
does not meet this condition. Another property that must be met is
that the laws must reduce to the well known forms of classical physics
in the appropriate limit, just as special relativity reduces to classical
physics when the speeds of the reference frames are much small than
c.
According to Newtonian gravity, the source of the gravitational field
is mass. Recall the Poisson equation states that
∇2 Φ = 4πGρ,
where ρ is the mass per unit volume. Furthermore, in Newtonian me-
chanics, mass is a conserved quantity. But according to relativity the-
ory, mass and energy are interchangeable quantities. The problem is
to replace the Poisson equation with tensors and to find a substitute
for mass on the right hand side. In special relativity it is shown that
the mass-energy-momentum relationship is
E 2 = p 2 c2 + m 2 c4 ,
304 9. THE GRAVITATIONAL FIELD

where p is the momentum. Einstein decided that this quantity could

serve in the role of the mass of Newtonian mechanics. The expression
had the added benefit that it could be written in tensor form as the
Energy-Momentum tensor T µν . This is a 4×4 symmetric tensor which
has 16 components, although only 10 of them are independent. These
components are:
T 00 = local energy density (including mass energy),
T 0i = momentum density in i direction times the speed of light,
T ij = rate of flow per unit area of ith component of momentum
perpendicular to the j direction.
It can be shown that conservation of energy and momentum requires
that X
∇µ T µν = 0.
µ
Examples of the energy-momentum tensor often involve electro-
magnetic fields and are beyond the scope of our study. However, two
simple situations can be noted. For a region of space where the only
constituents are non-interacting particles of mass m we find
T µν = ρU µ U ν
where U is the four-dimensional analog of velocity. For a region of
space filled with an ideal fluid with pressure P we have
T µν = (ρ + P/c2 )U µ U ν − P g µν .
Finally, for a region of empty space (no particles and no electromagnetic
fields)
T µν = 0.
The tensor T µν will replace 4πGρ on the right hand side of Poisson’s
equation. Next we consider what to put on the left-hand side of the
δ
equation. Recall that the rank 4 Riemann tensor Rαβγ describes the
curvature of spacetime. Using a mathematical procedure known as
“contracting” we can define a rank 2 tensor Rαβ that has some of the
properties of the Riemann tensor. Then a third quantity, R, called the
Ricci scalar, can be obtained by evaluating
X
R= g αβ Rαβ .
α,β

Einstein put these together to form the following rank 2 tensor

1
Gµν = Rµν − g µν R,
2
referred to as the “Einstein tensor.” Note that all the terms on the
right hand side are functions of the metric g µν . It turns out that in the
 9.8. EINSTEIN’S THEORY OF GRAVITATION (OPTIONAL) 305

“Newtonian limit” this tensor reduces to the left hand side of Poisson’s
equation. It is, of course, a sign of Einstein’s genius that he realized
this correspondence and expressed the Einstein tensor as proportional
to the energy-momentum tensor. It turns out that the proportionality
constant is K = 8πG/c2 where G is the universal gravitational constant
in Newton’s law of gravity. Consequently,
8πG
Gµν = − 2 T µν . (9.18)
c
This is, essentially, the Einstein expression for the gravitational field. It
is a set of 10 coupled second order differential equations for the metric
g µν and is referred to as “Einstein’s field equations.”13
We still have not answered the question about how a mass particle
will move in curved spacetime. Recall our discussion of the calculus of
variations in Chapter 4 in which we showed that the shortest distance
between two points in a plane is a straight line and the shortest distance
between two points on the surface of a sphere is a section of a great
circle (or geodesic). It is an interesting consequence of the general rela-
tivity that a particle in a gravitational field will move along a geodesic.
At first Einstein thought that this would have to be incorporated into
his theory as a postulate, but some years after he had published his
theory he realized that geodesic motion follows from the condition
X
∇µ T µν = 0.
µ

So geodesic motion is implicitly included in the theory.

The motion of a particle in curved spacetime is often described
in terms of the analogy of a bowling ball placed on a soft mattress,
thus deforming the surface of the mattress. A marble rolling along the
mattress will roll into the depression, as if attracted to the bowling
ball.
John Wheeler summarized the rules of general relativity in the fol-
lowing way:
Matter tells space how to curve.
Space tells matter how to move.
Our discussion has been rather abstract, so it might be helpful to
present a specific metric tensor that was evaluated by Karl Schwarzschild
shortly after Einstein published his theory. Schwarzschild considered
13For the purists, I must admit that Equation 9.18 is a simplified version of
the Einstein Field equations because I have left off the “cosmological constant.”
You may have noted that there are 16 field equations, but only 10 of them are
independent.
306 9. THE GRAVITATIONAL FIELD

Figure 9.10. A bowling ball on a mattress as an analog

to curved spacetime.

the gravitational field in the empty space outside of a star. In empty

space the energy-momentum tensor components are all zero. Solving
the Einstein field equations for the components of the metric tensor,
Schwarzschild found
1 − 2GM 0 0 0
 
c2 r
1
 0 − 1− 2GM 0 0 
gµν =  c2 r .
0 0 −r2 0 
0 0 0 −r2 sin2 θ

The quantity 2GMc2

is a distance. It is called the Schwarzschild radius
(RS ). Note that the g11 element of the matrix becomes infinite at the
Schwarzschild radius. It turns out that if a star shrinks to a radius less
than RS , it becomes a black hole from which not even light can escape.
The analysis of Schwarzschild led to other important consequences,
such as predicting the correction for the precession of planetary orbits
that we discussed previously.
The theory of general relativity is, as we have seen, basically a field
theory of gravitation. It has been submitted to numerous experimental
tests and has been proved correct over and over. It is unfortunate that
it is so mathematically complex that it is only accessible to a small
community of physicists, astronomers, and mathematicians.

9.9. Summary
A field is defined as a physical quantity whose value can be deter-
mined at every point in some given region of space. The gravitational
field g(r) due to a particle of mass M located at point r0 can be defined
in terms of the force it exerts on an infinitesimal point mass m located
 9.9. SUMMARY 307

at r,
r − r0
 
F
g(r)= lim = −GM .
m→0 m |r − r0 |3
The field produced by an extended body can be determined by evalu-
ating the expression
r − r0
Z
g(r) = −G ρ(r0 ) 0 |3
dτ 0 .
body |r − r
To evaluate the integral it is usually necessary to resolve the vectors
into components, using the symmetry of the problem as a guide.
The gravitational potential Φ(r) is given by
ρ(r0 )dτ 0
Z
Φ(r) = −G 0
.
body |r − r |

A field can be represented graphically by drawing the field lines (as

illustrated in Figure 9.7) or by drawing equipotential surfaces.
The Helmholtz theorem tells us that we can determine a vector field
if we know its “sources” defined as the gradient and curl of the field.
For the gravitational field these are:
∇ × g = 0,
∇ · g = −4πGρ.
The first of these relations indicates that the gravitational field is con-
servative, thus allowing us to define a potential. The second equation
is Gauss’s law and in integral form it is expressed as:
I
g · n̂ dS = −4πGMenc
S

where Menc is the mass enclosed by the surface S. In terms of the

potential, this leads to
∇2 Φ = 4πGρ
which is called Poisson’s equation. In regions of space where ρ = 0 this
reduces to Laplace’s equation
∇2 Φ = 0.
Finally, although Newton’s theory of gravitation has been super-
seded by Einstein’s theory of general relativity, we still use Newtonian
mechanics for most practical problems because it is much easier to
apply and the differences are miniscule.
308 9. THE GRAVITATIONAL FIELD

9.10. Problems
Problem 9.1. A neutron star has a mass of 1030 kg and a radius
of 5 kilometers. A body is dropped from a height of 20 cm above the
surface. Determine the speed of the body when it hits the surface.
Problem 9.2. The Earth is suddenly brought to a standstill. Eval-
uate the time required for it to collide with the Sun. You can assume
the Sun does not move and that the center of mass of the system is at
the center of the Sun. How good is this approximation?
Problem 9.3. Determine the period of a surface skimming satellite
in a circular orbit about a uniform, perfectly spherical planet of radius
R and density ρ. Determine the period of such a satellite about a
different planet which has radius 2R but the same density as the first.
Explain your result.
Problem 9.4. Consider an infinitely long, straight string whose
linear mass density is λ (mass per unit length). By direct integration,
determine the gravitational field a distance r from the string.
Problem 9.5. By direct integration, find the field at a point on
the axis of symmetry of a cylinder of length L, radius R and uniform
density ρ. (Hint: Let the mass element be an infinitesimally thin disk
and use the result of Worked Example 9.2.)
Problem 9.6. By direct integration determine the gravitational
field a distance z above an infinite flat surface whose mass per unit
area is σ. Check your result by using the result of Worked Example
9.2,
Problem 9.7. Consider an infinite string with linear mass density
λ. A particle of mass m is a distance d from the string. Determine the
force on the particle.
Problem 9.8. Using the technique of Worked Example 9.3, de-
termine the potential and the field of a spherical shell of mass M and
radius a, at an interior point. Answer: g = 0 and Φ = −GM/a =
constant.
Problem 9.9. Assuming that the mass cylinder of Worked Exam-
ple 9.4 has a constant mass density ρ, determine the field at a point
inside the cylinder.
Problem 9.10. A sphere of radius R and constant mass density
ρ has a spherical cavity of radius r where r = R/2. The center of the
cavity is a distance R/4 from the center of the sphere. A particle of
 9.10. PROBLEMS 309

mass m is located at an outside point, a distance z from the center

of the larger sphere and along the line going through the centers of
the sphere and the cavity. Determine the force on the particle. (Note:
z > R.)
Problem 9.11. Consider a planet of mass M and radius R. Assume
the planet is spherical and has a constant density. By direct integration
determine the gravitational field and the gravitational potential at all
points inside and outside the planet. (Assume the potential is zero at
infinity and there are no other bodies in the universe.)
Problem 9.12. Suppose an intergalactic gas cloud were found to
have a density given by
M b2
ρ= .
2πr(r2 + b2 )2
Here M and b are constants related to the size and mass of the cloud.
Note that the density extends out to infinity, but the density is infinitely
small for very large values of r. Determine the gravitational field and
the gravitational potential as functions of r.
Problem 9.13. (a) Determine the gravitational potential at a point
on the axis of a ring of mass of radius a and mass m. (b) Determine the
potential for this ring at an off-axis point located a distance r from the
center of the ring and making an angle θ with the axis. You may assume
r >> a and you can disregard any terms of order (a/r)3 or smaller.
Hint: the angle γ between two lines whose directions are specified by
θ1 , φ1 and θ2 , φ2 is given by cos γ = cos θ1 cos θ2 + sin θ1 sin θ2 cos(φ1 −
φ2 ).
Problem 9.14. Determine the gravitational potential inside and
outside of a constant density sphere of radius R and mass M. Plot Φ
vs r. (Answer: Inside the sphere Φ = −GM r2 /R3 .)
Problem 9.15. Suppose a tunnel were dug through the Earth
along a diameter. Show that an object dropped into this tunnel would
undergo simple harmonic motion. Determine the period of the motion.
(Air resistance is, of course, neglected.) Answer: 84 minutes.
Problem 9.16. A particle is at an arbitrary point P inside a mass
shell. (a) Draw the line through P so that it passes through the center
of the shell (that is, the line is a diameter). Construct a “double cone”
with vertices at P as shown in Figure 9.11. Show that the resultant
gravitational force on the particle due to the two surface elements A1
and A2 is zero. (b) Generalize to any line through P and show that
310 9. THE GRAVITATIONAL FIELD

there is no net force acting on the particle at any interior point. (This
is the way Newton showed that there was no force on a particle inside
a shell.)

A2
P

r2

r1
A1

Figure 9.11. A spherical mass shell and two cones with

common vertex at P.

Problem 9.17. Use Gauss’s law to determine the field inside and
outside of: (a) A sphere of uniform mass density, (b) A homogeneous
hollow spherical shell. Let the mass and radius be M and R in both
cases.
Problem 9.18. Use Gauss’s law to determine the field inside and
outside an infinite cylinder of radius R and uniform mass density ρ.
Express your answer in terms of r, the distance from the axis.
Problem 9.19. Use Gauss’s law to determine the field above and
below an infinite plane of surface mass density σ.
Problem 9.20. An infinite mass plane gives rise to a constant
gravitational field, and a sphere of mass M gives rise to the gravita-
tional field given by Newton’s law of universal gravitation. Consider
an object that looks like Saturn, but instead of rings, the equatorial
plane of the sphere is a uniform mass plane of infinite extent and mass
density σ (per unit area). Let the radius of the sphere be R. Deter-
mine the work done as a particle of mass m is moved from the “north
pole” of the sphere to a distance h above it (h > R). (The principle
of superposition tells us the field of two mass distributions is just the
vector sum of the field due to each.)
Problem 9.21. The gravitational field in some region of space is
given by g = −kr3 r̂ where k is a constant. What is the mass density
ρ? What is ρ if the field is given by g = −(k/r2 )r̂?
 9.10. PROBLEMS 311

Problem 9.22. A collision between Earth and an asteroid of diam-

eter 200 meters would cause widespread damage and loss of life. Space
scientists are interested in devising ways to deflect an approaching as-
teroid. Not too long ago, two NASA engineers14 suggested placing a
spacecraft near such an asteroid and use the gravitational attraction
between spacecraft and asteroid to tow it away. This “gravitational
tractor” would hover a distance d from the asteroid. Its engines would
be directed as shown in Figure 9.12 so that they would not blast the
surface of the asteroid. The angle φ is the half angle of the rocket
jet; you may assume a value of 20◦ for this angle. (a) Show that the
minimum thrust required to just balance the gravitational attraction
of the asteroid is
GM m/d2
T =
cos[sin−1 (r/d) + φ]
where r is the radius of the asteroid. (b) Evaluate T for a 200 meter
diameter asteroid (r = 100 m), having a density of 2X103 kg/m3 , if a
20 tonne spacecraft hovers at d = 2r. (c) Show that the velocity change
imparted to the asteroid per second is
∆v = Gm/d2 .

Figure 9.12. A “gravitational tractor.” The shaded ar-

eas are the exhaust plumes from the rocket motors which
are tilted away from the asteroid so as not to push on it.
The angle φ is the half angle of the plume.

COMPUTATIONAL PROJECTS
14E. T. Lu and S. G. Love, Gravitational Tractor for Towing Asteroids, Nature,
438, 177-178 (2005).
312 9. THE GRAVITATIONAL FIELD

Computational Technique: Numerical Integration

Two simple numerical integration schemes are called the “Trape-
zoidal Rule” and “Simpson’s Rule.” These are two applications of
a more general method called the “Newton-Cotes Technique.” Both
schemes assume you know the value of the integrand at the end points
and at a number of intermediate points.
The trapezoidal rule assumes the integrand f (x) varies linearly be-
tween the “data” points. Thus, if we know f (x) at xi and at xi+1 ,
then Z xi+1
1
f (x)dx = (xi+1 − xi )(fi+1 + fi )
xi 2
The total integral from initial to final values of x will be the sum of
such terms. (Note that the area under the f (x) curve is broken up into
trapezoids. The results are often not very accurate unless the x values
are very closely spaced.)
Simpson’s rule uses a quadratic fit to a series of three equally spaced
points, separated by equal distances d. Then
Z xi+1
d
f (x)dx = (fi + 4fi+1 + fi+2 )
xi 3
Note that the number of values of x must be odd.

Computational Project 9.1. A sphere of radius 1 meter has a

density that varies as ρ(r) = 5r3 for 0 ≤ r ≤ 1. Obtain the mass of
the sphere analytically and numerically with the trapezoidal rule and
Simpson’s rule for nine equally spaced values of r. Compare results of
the three calculations.
Computational Project 9.2. Write a computer program to de-
termine the position and velocity of a projectile fired vertically from
the Earth’s surface with an initial speed of 3000 m/s. Plot the position
and velocity as a function of time. The Earth is assumed to be a sphere.
You may neglect air resistance, the rotation of the Earth and the effect
of any other astronomical bodies, but note that the acceleration due
to gravity varies with distance from the center of the Earth.
Computational Project 9.3. Plot the potential for two equal
point masses. (To make things easier, assume you are using a system of
units in which the masses are unity, they are separated by unit distance
and the gravitational constant is unity.) You can let the two particles
lie on the x axis.
 9.10. PROBLEMS 313

Computational Project 9.4. Imagine a planet in the shape of a

cube with side 105 m. Draw a Cartesian coordinate system with origin
at the center of the cube and axes passing through the centers of the
faces. Obtain an integral expression for the potential at a point 106 m
from the origin along one of the axes. Write a computer program to
carry out the integration and obtain a numerical answer.
 Part 3

The Mechanics of Particles

Chapter 10
Central Force Motion: The Kepler
Problem

The orbital motion of a planet around the Sun was one of the first
important problems to be analyzed in terms of Newton’s three laws.
The gravitational force attracting a planet to the Sun is a central force.
The motion of a planet is a prime example of the more general problem
of the behavior of a particle acted upon by a central force.
Although we shall be primarily concerned with the motion of plan-
ets and satellites, the techniques you will learn in this chapter are ap-
plicable to any kind of central force. In this chapter, as well as learning
the laws governing the motion of celestial bodies, you will be exposed
to the concept of effective potential energy, and you will appreciate
how constants of the motion are used in solving physics problems.
Historically, the quantitative analysis of orbital motion began with
Kepler’s realization that the motion of planets can be described by
three empirical laws. An important point made in this chapter is that
Kepler’s laws of planetary motion can be derived theoretically from
Newton’s laws of motion. Additionally, Newton’s laws gives us a much
deeper understanding of Kepler’s laws. This application of Newton’s
ideas amazed and fascinated the “natural philosophers” of his era and
was one of the most important events in the history of science.1

1Much of the material in this chapter is an introduction to celestial mechanics.

Many excellent celestial mechanics books are available, including A. E. Roy, Orbital
Motion, Adam Hilger Press, Bristol, 1988, and J. M. A. Danby, Fundamentals of
Celestial Mechanics, William-Bell Inc, Richmond, VA, 1992. A more recent book
that is particularly well written is by C. D. Murray and S. F. Dermott, Solar System
Dynamics, Cambridge University Press, Cambridge,1999.
317
318 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

10.1. Johannes Kepler (Optional Historical Note)

Johannes Kepler lived from 1571 to 1630. He was born into a poor
family in a small town in Germany. His father was a professional sol-
dier who spent much of his time away from home and his mother was
a quarrelsome woman who was accused of witchcraft in her old age. It
would be fair to say that Kepler had an unhappy childhood. Neverthe-
less, he was very intelligent and an excellent student. The prince who
ruled the region sent him to study at a Lutheran seminary. Eventu-
ally, Kepler graduated from the University of Tubingen. Originally he
planned to study divinity, but before he was ordained, the authorities
in the seminary convinced him that he was not cut out for the clergy.
With the help of his advisors, he was appointed to a position teaching
mathematics in Graz, Austria. He was a quiet, introspective person
and not greatly interested in teaching; in fact, he was probably a ter-
rible teacher for he had the reputation of interrupting his own lectures
to silently mull over some idea that had just occurred to him.
In Graz, Kepler had an idea that changed the course of his life.
This idea, which became an obsession with him, was (he thought) a
glimpse into the mind of God: A vision of the basic structure of the
universe. It was, so to speak, revealed to him why there are only five
planets, and why they are in their particular orbits around the sun.
Kepler’s mind-boggling inspiration was this: there are only five planets
(besides Earth) because there are only five “perfect” solids, and the
orbits of the planets correspond to spheres circumscribed about the
perfect solids when they are nested, one inside the other.
The “perfect” solids (or “simple polyhedrons”) are the geometrical
figures formed from regular polygons. A regular polygon is one with
equal sides. Thus, for example, a tetrahedron (or equilateral pyra-
mid) is made up of four equilateral triangles. A cube is made up of
six squares. Similarly, an octahedron is an eight sided solid made of
equilateral triangles, an icosahedron is made of twenty equilateral tri-
angles and the dodecahedron is composed of twelve pentagons. Many
other solids can be constructed from polygons; for example a “soccer
ball” shape can be made up of pentagons and hexagons. This is the
structure of Carbon-60, the so-called “Bucky balls,” named in honor
of Buckminster Fuller who studied the properties of such structures.
There are, however, only five “perfect” solids whose faces are a single
type of regular polygon. There is a very neat proof that there can only
be five such solids; this proof can be traced back to the ancient Greeks.
 10.1. JOHANNES KEPLER (OPTIONAL HISTORICAL NOTE) 319

If you are interested, it is reproduced in the book Cosmos, by Carl

Sagan.2
Kepler’s idea was that if the perfect solids were placed one inside
the other, and each was circumscribed with a sphere, then the sun
would be at the center of the system and each of the five planets would
orbit the sun in a circular orbit whose radius would be equal to the
radius of the corresponding circumscribed sphere. Kepler built models
of these nested solids and their circumscribed spheres, but he could not
prove his theory because he did not have enough information on the
distances of the five (known) planets to the Sun. Figure 10.1 illustrates
the idea behind Kepler’s model.

Figure 10.1. A nested tetrahedron and cube with in-

scribed and circumscribed spheres. In Kepler’s model the
nested perfect solids (from innermost outwards) were an
octahedron, an icosahedron, a dodecahedron, a tetrahe-
dron and a cube.

At that time, the best astronomical data in Europe were in the

observatory of an eccentric Danish astronomer named Tycho Brahe.3
2Carl Sagan, Cosmos, Random House, New York, 1980 (Appendix 2).
3Tycho Brahe (1546-1601) set up an observatory at Uraniborg on the Danish
island of Hven. A stream of young assistants from all over Europe came to Urani-
borg where they carried out experiments in chemistry during the day and observed
the heavens at night. Uraniborg has been described as the first research institution
involved in “big science.” For the full story read the book by John Robert Christian-
son, On Tycho’s Island: Tycho Brahe and his Assistants, 1570-1601, Cambridge
University Press, Cambridge, 2000. At the time Kepler went to visit him, Brahe
had moved to Prague.
320 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

Kepler went to visit Brahe to get his data. He was certain that Brahe
would be overwhelmed by his wonderful new theory. Brahe was not
overwhelmed. In fact, at first Brahe would not even let Kepler see
the data! However, Brahe soon realized that Kepler was an excellent
mathematician and he offered to let him work on a small portion of
the data to calculate the orbit of Mars. This was certainly not what
Kepler had in mind, but he grudgingly agreed.
This was, perhaps, the original “graduate student - professor” re-
lationship in science. To this day the same pattern exists. A young,
aspiring scientist makes the pilgrimage to the laboratory of the estab-
lished professor with the hope of being allowed to share in the profes-
sor’s knowledge and data. You yourself may be doing this a few years
from now. I hope your relationship with your graduate advisor will less
tempestuous than that of Kepler and Tycho. They did not get along
at all. Tycho was a man who loved a party, who spent his evenings
eating, drinking, and carousing, whereas Kepler was somber and rather
puritanical and did not at all approve of Tycho’s lifestyle.
Some years after the collaboration began, Tycho died. (Rumor has
it that he died of a burst bladder while on a drinking spree.) Kepler
inherited all of Tycho’s data as well as Tycho’s position as court as-
tronomer. After a great deal of analysis, much to his dismay, Kepler
found that Tycho’s data did not support his grandiose theory. In fact,
the data showed that the orbits of planets were not even circles, but
rather ellipses! Kepler tried to modify his theory by slipping elliptical
orbits between the inscribed perfect solids, but it did not quite work.
Kepler never did know that there are more than five planets, as the
discoveries of the planets Uranus and Neptune came many years after
his death. By the time these were discovered, Kepler’s theory on the
inscribed orbits of the five planets was no more than a historical oddity.
Kepler’s life is full of instructive incidents for physicists. His most
valuable contribution to science was his analysis and synthesis of the
observations of Tycho Brahe. It is interesting to consider that he was
obsessed by a beautiful theory that did not agree with experiment.
No matter how beautiful a theory may be, if it does not agree with
experimental measurements, it must be discarded! As a physicist you
must never let your theories carry you away. Physics is the study of
the physical universe and it is Nature that determines the way things
behave. It was to Kepler’s great credit that he respected and believed
the data, even though the data did not agree with his theory.
 10.3. CENTRAL FORCES 321

10.2. Kepler’s Laws

Kepler’s synthesis of Tycho’s planetary data can be expressed as
three statements that are now called Kepler’s laws of planetary motion.
They are:
1. The orbit of a planet is an ellipse with the sun at one focus.
2. The radius vector of a planet (the sun-planet line) sweeps out
equal areas in equal times.
3. The square of the period of a planet is proportional to the cube
of the semi-major axis of its elliptical orbit.
These laws express three facts about the motion of a planet. They
are “empirical laws,” that is, they were obtained from the data but they
had no theoretical basis. Kepler’s laws describe the motion of planets
but neither Kepler nor anyone else living at that time could give a
reasonable explanation why planets behave in this manner. About
70 years later, Newton applied the law of universal gravitation and his
laws of motion to the problem and succeeded brilliantly in showing that
Kepler’s laws are a consequence of some very basic physical relations.
It is no wonder that other scientists of his era were in such awe of
him. In this chapter, I will show you what Newton did, but of course I
will use modern methods. Near the end of the chapter I will return to
Kepler’s laws: by that time you should have a deeper appreciation for
them.

10.3. Central Forces

In an attempt to understand Kepler’s laws, we will begin by study-
ing the motion of a particle subjected to the gravitational force of a
second body (which we also treat as if it were a particle). For the sake
of simplicity let us assume the second body is at rest. Strictly speaking,
this cannot be true; the two objects actually orbit about their center of
mass. But if one particle is much more massive than the other, then the
more massive particle has a much smaller acceleration and in the limit,
as the ratio of the masses goes to infinity, the more massive particle
can be considered to be at rest.
To appreciate this, imagine a binary star system in which both stars
have the same mass. The two stars will orbit around their center of
mass, which is at a point halfway between them. But suppose one of
the stars is more massive than the other. The center of mass will be
closer to the more massive star. If one star is infinitely more massive
than the other, the center of mass will be at the center of the larger
star. Similarly, the space shuttle and the Earth are orbiting around
their center of mass, but for all intents and purposes, the center of
322 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

mass of this system lies at the center of the Earth and we are perfectly
justified in considering the Earth to be at rest.4
Placing the origin of the coordinate system at the center of mass
(essentially the center of the Sun) the force on a planet is an attrac-
tive force directed toward the origin. The magnitude of this force is
inversely proportional to the square of the distance from the planet to
the origin. Any force that is directed toward or away from a fixed point
(usually taken as the origin of coordinates) and whose magnitude is a
function only of the distance to the origin is called a central force. In
general, a central force will have the form
F = f (r)r̂,
where f (r) is the magnitude of the force (a function only of r) and r̂
is the radial unit vector. The direction of the force is along the line
joining the particle and the origin. Two important central forces are the
gravitational force between masses and the electrostatic force between
charges. For the gravitational force,
Gm1 m2
F=− r̂,
r2
and for the electrostatic force
Q1 Q2
F= r̂.
4πo r2
We can also imagine other central forces such as, for example,
k
F = 5 r̂.
r
Such forces may or may not exist in nature, but we can analyze them
mathematically anyway. Although this seems like a useless theoretical
exercise, such studies may have a practical outcome. For example, to
a first approximation, the force between molecules can be expressed
as the sum of two central forces. This “Lennard-Jones” force can be
expressed in terms of the potential energy as
a b
V = − 6 + 12 .
r r
Here a and b are constants that depend on the properties of the partic-
ular molecules involved in the interaction. The force itself is obtained
from F = −∇V.
4Recallthe discussion of reduced mass, defined as µ = m1 m2 /(m1 + m2 ). See
Equation (6.20). If the central force problem is treated in terms of the reduced
mass, then no approximation is made. The relations obtained are the same as
those derived here except that m is replaced by µ and r is the position of one body
relative to the other rather than the distance from the center of mass.
 10.3. CENTRAL FORCES 323

We now analyze the motion of a particle of mass m in a central force

field. We find that a particle that is acted upon by a central force will
move in a plane. This is obvious if you consider the following argument.
At any instant in time the particle is in the plane defined by the position
vector and the velocity vector. The only way the particle can move out
of this plane is if its acceleration has a component perpendicular to the
plane. But for a central force, the force (and hence the acceleration)
lies along r, and therefore there is no component of the acceleration
perpendicular to the plane.
An important property of a central force is the fact that a parti-
cle moving under the action of a central force has a constant angu-
lar momentum. By definition, the angular momentum of a particle
is l = r × p where r is the position of the particle and p is its linear
momentum. Recall from Section 7.2.2 that the time derivative of the
angular momentum is
dl d
= (r × p) = r × F.
dt dt
In the case of a central force, F = f (r)r̂. Hence,
dl
= r × f (r)r̂ = rr̂ × f (r)r̂
dt
= rf (r)(r̂ × r̂) = 0.
Consequently, for a central force, the time derivative of the angular
momentum is zero. This means that the angular momentum is con-
stant. Another way of looking at this is to recall that the time rate of
change of angular momentum is equal to the torque. A central force
cannot exert a torque on a particle. You should convince yourself that
this statement is true.
These simple physical arguments lead us to an important conserva-
tion law concerning the motion of any particle under the action of any
central force:
The angular momentum is constant.
This means that both the magnitude and the direction of the angular
momentum are constant.
For a more sophisticated proof of the constancy of angular momen-
tum, you can write the Lagrangian for a particle in a central force field.
In polar coordinates the kinetic
Rr energy is T = 12 m(ṙ2 + r2 θ̇2 ) and the
potential energy is V = − r0 f (r)dr. The Lagrangian is
Z r
1 2 1 2 2
L = T − V = mṙ + mr θ̇ + f (r)dr.
2 2 r0
324 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

Since θ is ignorable, ∂L/∂ θ̇ = constant. That is,

∂L
= mr2 θ̇ = l = constant.
∂ θ̇
We can use the constancy of angular momentum to generate another
proof that a particle in a central force field moves in a plane. Assume
the particle is at position r and has velocity v. These two vectors
define a plane and, of course, the particle lies in this plane. By the
definition of cross product, the vector r × v is perpendicular to the
plane containing the vectors r and v. But l = mr × v, so the vector l is
perpendicular to the plane containing r and v. Since l = constant, the
perpendicular to the plane containing r and v is constant. Therefore,
the particle moves in a constant plane. See Figure 10.2.

v
r
Figure 10.2. The angular momentum vector is perpen-
dicular to the plane containing r and v. Since the angular
momentum is constant, the plane is invariant.

Since the motion of the particle lies in a plane, two coordinates are
sufficient to specify its position. These can be x and y or r and θ.
The origin of the coordinate system is usually placed at the primary
(assumed to be at rest).5 To orient the coordinates, it is necessary to
specify a fixed direction. Astronomers pick an imaginary line from the
center of the Earth towards a position called “The First Point in Aries”
which is the position of the Sun at the vernal equinox.
In Figure 10.3 the symbol Υ indicates the fixed line in space. In
polar coordinates the angle θ is measured from Υ. In Cartesian coor-
dinates one usually defines the x-axis along this fixed line. The y-axis
is selected in the plane of the orbit and perpendicular to the x-axis.
The z-axis is then perpendicular to the orbit and along the angular
momentum vector. For many problems it is quite safe to assume that
the coordinate system is an inertial system.
5In
the language of astronomy, if one body is much more massive than the
other, the massive body is called the “primary.”The primary is often assumed to
be at rest. Actually, of course, both bodies move about the center of mass.
 10.3. CENTRAL FORCES 325

r
θ

Figure 10.3. The position of a planet referenced to the

first point in Aries.

Five thousand years ago the position of the Sun at noon on the
vernal equinox was in the constellation Aries (“the ram”). Since the
Earth’s axis of rotation precesses with a period of about 25,000 years,
the position of the Sun at noon on the vernal equinox has changed and
it is presently in the constellation Pisces; within a few hundred years
it will enter into the constellation Aquarius. However, the name “First
Point in Aries” and the symbol Υ, representing ram’s horns, are still
used to represent this arbitrary fixed line in space. It is amusing to
note that due to the precession of the Earth’s axis, the positions of
the zodiacal constellations have shifted, but astrologers still use the
values of 5000 years ago. Thus, people born when the Sun was in the
constellation Pisces think they are Aries, those who were born when
the Sun was in Aries think they are Taurus, and so on. If astrology
had any validity, this horrible mix-up in the zodiacal signs would be
serious indeed!

Worked Example 10.1. A particle is in a circular orbit under

the action of an attractive central force given by f (r) = −k/r3 .
Obtain an expression for the angular momentum and show that it
is constant.
Solution: By Newton’s second law, mr̈ = −f (r)r̂. = − rk3 Re-
call from our discussion of plane polar coordinates in Chapter 2
that
r̈ = (r̈ − rθ̇2 )r̂ + (rθ̈ + 2ṙθ̇)θ̂.
For a circular orbit, the magnitude r is constant, so
r̈ = −rθ̇2 r̂ + rθ̈θ̂ = −f (r)r̂.
326 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

Therefore, for an inverse cube force law, mr̈ = −f (r)r̂ yields two
relations, namely
k
−mrθ̇2 = −f (r) = − 3 ,
r
and
mrθ̈ = 0.
The second equation boils down to θ̈ = 0, telling us that the angular
velocity, θ̇ is constant. The first equation yields
mr4 θ˙2 = k.
But since l = mr2 θ̇ we see that the left hand side is l2 /m and
consequently
l2 = mk,
and hence the angular momentum is constant. Actually, this result
also follows from r = constant and θ̇ = constant.

Worked Example 10.2. Consider a particle in an attractive

force field whose potential energy is of the form V (r) = krn+1 .
Show that for a periodic orbit the average kinetic energy is related
to the average potential energy by
n+1
hT i = hV i .
2
Apply to the gravitational force. (This is a special case of the virial
theorem.)
Solution: This problem asks us to determine the relationship
between the average values of the kinetic and potential energy. The
time average of any quantity (such as the kinetic energy) is defined
thus:
1 τ
Z
hT i = T (t)dt.
τ 0
For this problem, it is reasonable to let τ be the orbit period.
If a particle is in orbit, after one period it will have returned to
its original location and have the same velocity as it did initially.
Therefore, the quantity p · r repeats periodically. Let us define the
periodic function G(t) = p · r. Consider the time average of the
derivative of G with respect to time:
1 τ dG 1 τ
  Z Z
dG 1
= dt = dG = [G (τ ) − G (0)] = 0.
dt τ 0 dt τ 0 τ
 10.3. CENTRAL FORCES 327

But
dG d
= (p · r) = p · ṙ + ṗ · r
dt dt
= mv · v + F · r =2T − ∇V ·r
dV
= 2T − r,
dr
because V = V (r). Therefore,
1 τ
  Z  
dG dV
= 2T − r dt = 0,
dt τ 0 dr
and
 
dV
2 hT i − r = 0,
dr
 
1 dV
hT i = r .
2 dr
Since V = krn+1 we have
dV
r = (n + 1)krn r = (n + 1)krn+1 = (n + 1)V,
dr
and  
dV
r = (n + 1) hV i ,
dr
so
n+1
hT i = hV i .
2
The gravitational potential has the form V = − kr so n = −2
and
1
hT i = − hV i .
2
This relationship between kinetic and potential energy is useful
when solving orbital mechanics problems. It might be mentioned
that the virial theorem has important applications in thermody-
namics and statistical mechanics.

Exercise 10.1. Use F = −∇V to obtain an expression for the

Lennard-Jones force. Determine the value of r where the
force changes
from attractive to repulsive. Answer: F = − r̂ 6ar7
− 12b
r13
.
Rr
Exercise 10.2. Use V = − r0 F (r)dr to obtain the potential en-
ergy for the gravitational force, the electrostatic force, and the force
exerted by a spring. Select r0 appropriately.
328 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

Exercise 10.3. Assume that at some initial moment the radius

vector r and the velocity vector v are perpendicular to one another,
however the angle between them is changing with time. Explain why
the two vectors cannot ever be parallel to one another. Answer: Be-
cause angular momentum is constant.

10.4. The Equation of Motion

Our next task is to determine the equation of motion for a planet.
We will do this in two ways: first by writing down Newton’s second
law, and second by applying the Lagrangian technique.
The force acting on a particle (planet) of mass m is given by New-
ton’s law of universal gravitation
GM m
F = − 2 r̂,
r
where M is the mass of the larger attracting body, assumed fixed at
the origin of coordinates. The equation of motion of the satellite (mass
m) is
Mm
mr̈ = −G 2 r̂,
r
or
GM
r̈ = − 2 r̂. (10.1)
r
We evaluated the acceleration r̈ in polar coordinates in Chapter 2 (see
Equation 2.14), obtaining
r̈ = (r̈ − rθ̇2 )r̂ + (rθ̈ + 2ṙθ̇)θ̂.
Substituting this expression into Equation (10.1), yields
GM
(r̈ − rθ̇2 )r̂ + (rθ̈ + 2ṙθ̇)θ̂ = −
r̂.
r2
Separately equating the radial and angular components, this equation
yields the following two scalar equations
GM
r̈ − rθ̇2 = − 2 , (10.2)
r
rθ̈ + 2ṙθ̇ = 0.
These equations are a pair of coupled second order ordinary differential
equations. The second equation, the “θ-equation,” is easy to analyze
by going back to the definition of angular momentum and recalling that
l = r × p = r×mv.
 10.4. THE EQUATION OF MOTION 329

The velocity in plane polar coordinates is v = ṙr̂+rθ̇θ̂ (see Equation

2.13), so h i
l = m rr̂ × (ṙr̂+rθ̇θ̂) .
Now carry out the cross products, using the facts that r̂ × r̂ = 0 and
r̂ × θ̂ = k̂, where k̂ is perpendicular to the plane of motion. You obtain

l = mr(rθ̇)k̂ = mr2 θ̇k̂. (10.3)

This, in itself, is a useful equation. But it becomes even more useful if
you take its time derivative,
dl d  2 
= mr θ̇ k̂.
dt dt
For central forces, the angular momentum is constant, so dl/dt = 0
and the left hand side of this equation is zero. Therefore,

d  2   
0 = mr θ̇ = m 2rṙθ̇ + r2 θ̈ ,
dt  
= mr 2ṙθ̇ + rθ̈ .
Since neither m nor r is equal to zero, this implies that
rθ̈ + 2ṙθ̇ = 0.
But this is just the θ equation! Therefore, we see that the θ component
of the equation of motion (the second of equations 10.2) is a statement
that angular momentum is constant. Consequently, that equation can
be replaced by the equivalent equation
l = mr2 θ̇ = constant.
This equation, in turn, gives a nice expression for θ̇, namely,
l
θ̇ = . (10.4)
mr2
Replacing θ̇ in the r equation (the first of Equations 10.2) by ex-
pression (10.4) yields the following equation for the radial motion of
mass m:
l2 GM
r̈ − 2 3 = − 2 . (10.5)
mr r
This equation involves only r. Thus, conservation of angular momentum
de-couples the equations of motion. The equation has only one variable
(r), so it is often called a “one-dimensional equation.” But you should
always keep in mind that the motion takes place in two dimensions.
330 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

Once Equation (10.5) has been solved for r = r(t), you can use it in
Equation (10.4) to determine θ = θ(t).
We shall now use the Lagrangian technique to determine the equa-
tions of motion. (We had better obtain the same result!)
As you know, the Lagrangian is L = T − V. In this problem V is
the potential energy for a particle of mass m attracted gravitationally
to a body of mass M . According to Equation (9.6) this is
GM m
V (r) = − .
r
The kinetic energy for this two-dimensional problem is T = 12 m(ẋ2 +
ẏ 2 ), or in polar coordinates,
1 1
T = mṙ2 + mr2 θ̇2 .
2 2
Therefore the Lagrangian is
1 1 GM m
L = mṙ2 + mr2 θ̇2 + .
2 2 r
Recall that the Lagrange equations of motion have the form
d ∂L ∂L
− = 0,
dt ∂ q̇i ∂qi
where the qi ’s are now r and θ. So we have two equations, namely,
d ∂L ∂L
− = 0,
dt ∂ ṙ ∂r
and
d ∂L ∂L
− = 0.
dt ∂ θ̇ ∂θ
The partial derivatives are easily evaluated. You should prove for your-
self that the two equations of motion are
d GM m
(mṙ) − mrθ̇2 + = 0, (10.6)
dt r2
and
d  2  dl
mr θ̇ = 0, or = 0. (10.7)
dt dt
The second of these equations gives
l
θ̇ = .
mr2
Using this expression, Equation (10.6) leads to
l2 GM
r̈ − 2 3
=− 2 . (10.8)
mr r
 10.5. ENERGY AND THE EFFECTIVE POTENTIAL ENERGY 331

These are, of course, the same as the equations of motion obtained

using Newton’s second law, namely Equations (10.4) and (10.5). Note,
however, how much easier it is to use the Lagrangian than Newton’s
second law.

1
Exercise 10.4. Carry out the steps to show that T = 2
mṙ2 +
1
2
mr2 θ̇2 .

Exercise 10.5. Obtain Equations (10.6) and (10.7).

Exercise 10.6. Suppose the force between a particle of mass m

and a fixed point is given by F = −krr̂ where k is a constant. Obtain
the Lagrangian and the equations of motion. Is angular momentum
conserved for this system? Answer: mr̈ −mrθ̇2 +kr = 0; dtd (mr2 θ̇) = 0.

10.5. Energy and the Effective Potential Energy

We have (twice!) obtained the equation of motion for the radial
coordinate (Equation 10.5 and Equation 10.8). You probably expect
to proceed by integrating it to get the value of r as a function of time
and initial conditions. Indeed, we shall do that in Section 10.6, but first
let us consider what the conservation of energy principle tells us about
our problem. Since gravity is a conservative force, the total mechanical
energy is constant. Thus,

E = constant = T + V
1 Mm
= m(ṙ2 + r2 θ̇2 ) − G .
2 r

But l = mr2 θ̇ so we can replace θ̇ by l/mr2 and write

 2 
1 2 m 2 l GM m
E = mṙ + r 2 4
− ,
2 2 mr r
or
1 l2 GM m
E = mṙ2 + 2
− . (10.9)
2 2mr r
This looks like the relation E = T + V if we associate a “radial kinetic
energy” with the term 12 mṙ2 and an “effective potential energy” Vef f
332 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

with the remaining two terms. The effective potential Vef f is given by6
l2 GM m
Vef f = 2
− . (10.10)
2mr r
Although Vef f looks and acts like a potential energy and is a function
only of position, it is definitely not a potential energy since it actually
contains a kinetic energy term, namely, l2 / (2mr2 ) = 21 mr2 θ̇2 .
It is instructive to draw an energy diagram in terms of the effective
potential energy.7 Note that Vef f is the sum of two terms, one positive
and the other negative. For r → ∞, the negative term in Vef f is the
dominant term because  
1  1 
> 2 .
r r→∞ r r→∞
As r → 0, the positive term dominates because
 
1  1 
< 2 .
r r→0 r r→0
Therefore the plot of Vef f vs r must have the general shape shown
in Figure 10.4. Study this figure carefully and convince yourself it is
qualitatively correct, specifically that Vef f is positive as r → 0 and
negative as r → ∞. Notice particularly that the l2 /2mr2 term bends
much more sharply than −GM m/r. Also, keep in mind that r can
only take on positive values.
Figure 10.5 is also a plot of Vef f (r) vs r. In this plot the effective
potential energy does not have quite the right shape because I drew it
to make it easy for you to appreciate various aspects of the effective
potential that are hard to see on a more accurate plot, such as Figure
10.4. In Figure 10.5 you see four possible values for the total energy,
denoted E0, E1 , E2 , and E3 . Consider first a particle with energy E1 .
From Equations (10.9) and (10.10), we can write
1 2
mṙ = E − Vef f . (10.11)
2
6Theterms “effective potential energy” and “effective potential” are used inter-
changeably, even though a “potential” is actually potential energy per unit mass.
Some books use the term “fictitious potential.” An older term is “centrifugal po-
tential.” In general, the effective potential is defined as the sum of l2 /2mr2 and
the potential energy. Thus for an electron orbiting about a proton, the effective
potential would be
l2 e2
Vef f = − .
2mr2 4π0 r

7Atthis time you may wish to review the material on energy diagrams in
Section 5.6.
 10.5. ENERGY AND THE EFFECTIVE POTENTIAL ENERGY 333

100

80

Effe ctive P ote ntial (a rbitra ry units )

60

40

Veff l2 /2mr2
20

-20

-40

-60
-GMm/r

-80

-100
0 1 2 3 4 5 6 7 8 9 10

r (arbitrary units)

Figure 10.4. The effective potential energy Vef f (r) is

the sum of two terms, one positive and one negative.

Since 12 mṙ2 can never be negative, the particle cannot be located at

values of r for which E ≤ Vef f . From Figure (10.5) this implies that if
the energy is E1 , then r ≥ r1 where r1 is the point of closest approach.
You can imagine the particle starting at r = ∞, coming closer and
closer to the primary until it reaches r1 , (the turning point), and then
moving back out to infinity.

Figure 10.5. Energy diagram for the effective poten-

tial. The turning points for various values of total energy
are indicated.

You should keep in mind that this not a complete description of the
motion; it is only a description of the radial motion. Meanwhile the
particle is also moving in θ with a velocity given by θ̇ = l/mr2 . The
334 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

angular velocity increases as the radial distance decreases, in agree-

ment with Kepler’s second law. As we shall see, for positive values
of the energy, the path of the particle is a hyperbola as illustrated in
Figure 10.6. This could be the path of a comet coming in from infinity,
speeding up as it approaches the Sun, swooping around the Sun, and
then moving out to infinity. (Here “infinity” is a place far from the
Sun, such as the Oort cloud from which many comets are believed to
originate.)

Figure 10.6. A comet or some other celestial object in

a hyperbolic path.

Going back to the energy diagram, Figure 10.5, note that according
to Equation 10.11 the radial component of the velocity ṙ is given by
r
2
ṙ = [E − Vef f (r)]. (10.12)
m
Thus, the square of the radial speed is proportional to E − Vef f . (Note
that E − Vef f is not the kinetic energy. You may think of it, if you
like, as the “radial kinetic energy.”) On the energy diagram (Figure
10.5) the distance between the horizontal line at E1 and the heavy line
representing Vef f is proportional to ṙ2 . At r1 the value of Vef f is E1 ,
so ṙ = 0. That is, at the turning point the particle has zero radial
velocity, which makes perfect sense. At the turning point, the angular
component of the velocity is a maximum because rθ̇ = (r)(l/mr2 ) =
l/mr is greatest when r is smallest.
Next consider a particle with zero total energy (E = E0 = 0; see
Figure 10.5). This means that the positive “radial kinetic energy” 21 mṙ2
is equal in magnitude to the negative effective potential energy Vef f .
The motion of the particle as it comes from r = ∞ to the turning point
at r0 and then goes back out to r = ∞ is similar to the motion of the
particle with energy E1 . As we shall see, the main difference is that
the trajectory for energy E1 > 0 is a hyperbola and the trajectory for
energy E0 = 0 is a parabola. For the parabolic orbit, the radial speed
of the particle ṙ is zero at infinity as well as at the turning point. As
the particle comes in from infinity, ṙ increases, reaching a maximum
at r4 where Vef f reaches is greatest negative value, then slows down to
 10.5. ENERGY AND THE EFFECTIVE POTENTIAL ENERGY 335

zero at r0 . The angular velocity is given by θ̇ = l/mr2 . (You should

be able to describe the angular velocity as the particle comes in from
infinity to the point of closest approach and then moves back out to
infinity.)
If the particle has negative total energy, as indicated by E2 in Figure
10.5, the motion is quite different; there are now two turning points and
the particle can neither reach r = 0 nor move out to r = ∞. That is,
the motion is bounded. The particle is trapped in a potential well. As
it moves back and forth radially between the two points denoted by r2
and r3 , it is also moving azimuthally with a varying angular velocity θ̇.
As we shall see shortly, this combination of radial and angular motion
represents a trajectory which is an ellipse.
Finally, if the particle has energy E3 , the minimum possible total
energy, the value of ṙ is zero at all times and the particle is at a constant
radial position r = r4 . The path is a circle. The angular velocity is
θ̇ = l/mr42 = constant. The particle is therefore moving with constant
angular velocity in a circular path.
Depending on the value of the energy, the trajectory or orbit of the
particle is a hyperbola, a parabola, an ellipse, or a circle. These are
called conic sections because they can be generated by cutting a cone
in various different ways, as shown in Figure 10.7.
We have been considering the motion of massive bodies interacting
gravitationally, but the ideas and methods developed here are quite
general and can be easily adapted to any central force problem, such
as the motion of an electron in orbit around a proton in the Bohr model
of the hydrogen atom.

Exercise 10.7. A particle is in a parabolic orbit. Where is its

turning point? Answer: r0 = l2 /GM m2 .

Exercise 10.8. In the Bohr model of the hydrogen atom, an elec-

tron in the lowest orbit has angular momentum l = h, where h is
Planck’s constant divided by 2π. The electrostatic potential energy of
the electron and proton is V = −e2 /4π0 r. Show that the total energy
(mechanical plus electrostatic) of the lowest orbit is −(1/2)me4 /(4π0 h)2 .
Assume a circular orbit.
336 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

Circle

Ellipse

Hyperbola Parabola
Figure 10.7. The conic sections. If the cone is cut as
shown, the cross sections of the cuts will be a circle, an
ellipse, a parabola, and one branch of a hyperbola.

10.6. Solving the Radial Equation of Motion

The radial equation of motion (10.5) is a second order ordinary
differential equation of the form
d2 r l2 GM
= − .
dt2 m2 r 3 r2
In previous chapters we learned two different ways of solving such an
equation. In Section 3.6 in considering forces as a function of position,
2
we integrated directly by setting ddt2r = v dv
dr
. In Section 5.7 we solved
for the motion by using the energy integral in the form of Equation
(5.15), namely,
Z x(t) r
dx 2
p = t.
x0 E − V (x) m
Applying the energy approach to the present problem, we see that
Equation (10.9) can be written as
 1
l2
 
dr 2 GM m 2
= E− + . (10.13)
dt m 2mr2 r
This equation yields the following definite integral
Z r(t) Z t
dr
q
2 l2 GM m

= dt = t.
r0
m
E − 2mr2
+ r
0
 10.7. THE EQUATION OF THE ORBIT 337

Integrating gives
t = t(r, r0 , E, l),
which can, in principle, be inverted to yield
r = r(t, r0 , E, l).
In this solution, the total energy E and angular momentum l are ar-
bitrary constants. Note the use of the term arbitrary constant. Such
constants are arbitrary in the sense that the differential equation is
satisfied by the solution no matter what values the constants happen
to have. But for a particular problem, these constants are anything but
arbitrary! In simple kinematics problems the “arbitrary constants” are
usually the initial position and initial velocity. In more complex prob-
lems such as the one we just solved, they tend to be other constants of
the motion such as the energy and angular momentum.

Exercise 10.9. Obtain t = t(r, r0 , E, l) for a parabolic orbit (for

which E = 0). (You will have to look up the integral.) Note that
it would be very tedious to invert the expression to obtain r = r(t).
Answer:
√ r
l2

2m l2
t= GM mr + GM mr −
3(GM m)2 m 2m
√ r
l2
 
2m l2
− GM mr 0 + GM mr 0 −
3(GM m)2 m 2m

10.7. The Equation of the Orbit

Many physicists work in the space program. In a few years you
may be involved in planning a Moon shot or a planetary probe to
photograph the rings of Saturn. If that happens, you will need to
know the position of the spacecraft as a function of time. You will
have to calculate r = r(t) and you will use the technique described in
the previous section.
On the other hand, if you wish to describe the orbit of a satellite,
planet or comet, you are not interested in its position at a particular
time. Rather, you want a description of the path followed by the celes-
tial object. Mathematically, this is given by the equation of the orbit,
r = r(θ). Once you have this equation, you can determine r for every
value of θ and thus trace out the path.
338 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

I will first present a “brute force” method and then a “sophisticated

technique” for obtaining the orbit.

Brute Force Method

To determine the equation of the orbit, r = r(θ), start with the
expression for dr
dt
obtained using the energy equation (Equation 10.13).
Using the chain rule,
dr dr dθ dr
= = θ̇.
dt dθ dt dθ
But recall that the angular momentum is l = mr2 θ̇ so
dr l dr
= . (10.14)
dt mr2 dθ
Substituting into Equation (10.13) yields
 1
mr2 2 l2
 
dr GM m 2
= E− +
dθ l m 2mr2 r
1
2GM m2 3 2

2mE 4 2
= r −r + r
l2 l2
1
= αr2 + βr3 + γr4 2 ,


where I used the Greek letters α, β, and γ to illustrate the form of this
equation. Rewriting and integrating the last equation gives
Z r Z θ
dr
2 1/2
= dθ. (10.15)
r0 r(α + βr + γr ) θ0

This integral can be found in tables of integrals.

Thus, in principle, the problem of determining the equation of the
orbit is solved. The only things left to do are: (1) Integrate Equation
(10.15), and (2) Invert the result to obtain r = r(θ).

Sophisticated Technique
There is a different way to obtain the equation of the orbit that
cleverly avoids evaluating the complicated integral in (10.15). Bear
with me for a little while because the math is a bit involved (but not
difficult).
We will begin with Equation (10.2), the radial equation of motion:

GM
r̈ − rθ̇2 = − .
r2
 10.7. THE EQUATION OF THE ORBIT 339

Using l = mr2 θ̇ to eliminate θ̇ gives Equation (10.8) which is repeated

here for convenience:
l2 GM
r̈ − 2 3 = − 2 .
mr r
Now let us introduce a new variable, u, defined as the inverse of r.
That is,
1
u≡ .
r
Then r = u−1 and dr = −(1/u2 )du. Therefore,

dr 1 du 1 du dθ
=− 2 =− 2 ,
dt u dt u dθ dt
where the last step uses the chain rule and the fact that u = u(θ).
So,
dr 1 du du
= − 2 θ̇ = −r2 θ̇ .
dt u dθ dθ
2
But r θ̇ = l/m. Replacing,
dr l du
=− .
dt m dθ
Taking the derivative with respect to time again,
   
d dr d l du
= − ,
dt dt dt m dθ
   
l d du l d du dθ
r̈ = − =− ,
m dt dθ m dθ dθ dt
l d2 u l 2 d2 u
 
l
= − = − .
m dθ2 mr2 m2 r2 dθ2
But 1/r2 = u2 so
l 2 u 2 d2 u
r̈ = − .
m2 dθ2
Substituting this into Equation (10.8) and using 1/r3 = u3 we have
l2 u2 d2 u l2 2
− − u = −GM u2 ,
m dθ2 m2
or
d2 u m2 l2 u3
 
2
=− 2 2 − GM u .
dθ2 l u m2
So,
d2 u GM m2
+ u = . (10.16)
dθ2 l2
340 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

This is a very interesting equation. It looks like the equation for simple
harmonic motion except for the additional constant term on the right
hand side. It can be made to look exactly like the simple harmonic
motion equation by defining a new variable
GM m2
w =u− .
l2
Then, since the last term is a constant,
dw du
= ,
dθ dθ
and
d2 u d2 w
= .
d2 θ dθ2
Equation (10.16) can now be written in the form of the SHM equation:
d2 w
+ w = 0.
dθ2
As noted in Worked Example (3.5), the solution is sinusoidal and we
can write
w = A cos(θ − θ0 ).
Consequently,
GM m2
u− = A cos(θ − θ0 ).
l2
Now u is just the inverse of r, so the derivation finally gives an equation
for r in terms of θ :
1 1
r = = GM m2 .
u l 2 + A cos(θ − θ 0 )
Dividing top and bottom by GM m2 /l2 puts this in a nicer form:

l2 /GM m2
r= Al2
. (10.17)
1 + GM m2
cos(θ − θ0 )
From a mathematical point of view, the problem is now solved
because r has been expressed in terms of θ and other known quantities
(such as the angular momentum) and two constants of integration, A
and θ0 .
Equation (10.17) describes all the possible orbits for the two-body
problem, i.e., circles, ellipses, parabolas, and hyperbolas. Since ellipti-
cal motion is of particular interest, let us assume that the total energy
E is negative. Then the motion is bounded and the particle (planet)
oscillates radially between turning points r2 and r3 as illustrated in
Figure 10.5 for energy E2 .
 10.7. THE EQUATION OF THE ORBIT 341

At the turning points, the effective potential energy is equal to the

total energy. VefT Pf = E. This fact allows us to evaluate r2 and r3 which
we write, in general, as rtp . From Equation (10.10) we have

l2 GM m
VefT Pf = 2
− = E.
2mrtp rtp
Therefore, in terms of u ,
l2 2
u − GM mutp − E = 0.
2m tp
The two solutions of this quadratic equation for utp are:
r !
m 4l 2E
u± = 2 GM m ± (GM m)2 + . (10.18)
l 2m
(Note that 1/u± are the r2 and r3 of Figure 10.5.)
But we have seen that
GM m2
u = A cos(θ − θ0 ) + .
l2
The maximum and minimum values of this expression occur when
cos(θ − θ0 ) = ±1. That is,
GM m2
u+ = A + ,
l2
and
GM m2
u− = −A + .
l2
Using u+ and equating the expression above to the relation given by
Equation (10.18) leads to
r
GM m2 GM m2 m 2+
2l2 E
A+ = + (GM m) ,
l2 l2 l2 m
or
1
(GM m)2 m2 2Em 2

A= + 2 . (10.19)
l4 l
Thus A is related to the total energy and the angular momentum in a
rather complicated way.
Let us now turn our attention to θ0 , the other constant of in-
tegration. Since the maximum and minimum values of r occur at
cos(θ − θ0 ) = ±1, it is common to set θ0 = 0 and measure angles
from the x-axis. Then the initial time, t = 0, is the time the planet
passes through perihelion - the point of closest approach to the Sun.
342 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

See Figure 10.8. More generally, the angle θ is measured from some
fixed line in space. This would be appropriate if the ellipse is precessing
and the x-axis is rotating relative to the inertial frame. In that case
the angle θ0 gives the angle between the fixed line and the major axis
of the elliptical orbit at the initial time.

x
θ θ0
θ x

θ0=0

Figure 10.8. The position of a satellite relative to the

major axis is given by the angle θ. The major axis can
be defined to be the x axis. The angle θ0 gives the ori-
entation of the elliptical orbit relative to a fixed line in
space.

We have, then, determined A in terms of E and l and given the

physical interpretation of θ0 as a description of the orientation of the
ellipse with respect to inertial axes. Now the problem is indeed fully
solved. Unfortunately, as you have no doubt noticed, our expressions
are rather unwieldy. In a moment I will write them in a neater form.
But first I need to discuss some properties of conic sections, particularly
ellipses.

10.8. The Equation of an Ellipse

You probably remember the grade school method for drawing an
ellipse. You place a sheet of paper on a cork board and stick two tacks
in it. Then you take a piece of string, tie the ends together, and loop it
around the tacks. Finally you place a pencil in the loop of string and
move it around the tacks, keeping the string taut, and if you are very
careful you will draw an ellipse on the paper. See Figure 10.9.
As the pencil traces out the ellipse, the distance from the pencil to
either tack varies. But the sum of the distances to the two tacks remains
the same because the string is of constant length and the distance
between the two tacks does not change. Therefore the distance from the
first tack to the pencil plus the distance from the pencil to the second
 10.8. THE EQUATION OF AN ELLIPSE 343

Figure 10.9. How to draw an ellipse.

tack is constant. This fact is used for the mathematical definition of

an ellipse.
An ellipse is the locus of points whose distances
from two fixed points sum to a constant.
That is a bit complicated to say, but it is expressed quite simply as
an equation. First choose the two fixed points, call them focal points
and denote them F and F 0 . The distance from a point on the ellipse to
F will be called r and the distance from that same point to F 0 will be
r0 . See Figure 10.10. Then an ellipse is defined as the locus of points
such that r + r0 = constant.

r'
r
P' θ P
F' f f F

a
Figure 10.10. An ellipse. The total distance r + r0 +
F 0 F is a constant, so r + r0 must be constant. Points P 0
and P are called the apsides.

It is easy to obtain an equation for the ellipse in polar coordinates

by introducing the angle θ between r and
the major axis of the ellipse.
The length of the semimajor axis 21 P P 0 will be denoted a, and the
semiminor axis is denoted b. The distance from the center of the ellipse
to either focal point will be f = ea, where e is a number less than one,
called the eccentricity. The eccentricity e and the semimajor axis a
determine the size and shape of the ellipse.
344 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

It turns out that

r + r0 = 2a.
This relation may surprise you, but it is easily proved. Consider point
P . There r = F P and r0 = F 0 P . By symmetry, F 0 P = F P 0 and
therefore r + r0 = F P + F P 0 = 2a. Since r + r0 is a constant, if it is
equal to 2a at one point, it is equal to 2a at any point.
Now consider the triangle formed by r, r0 , and 2f , where f = ea.
See Figure 10.10. According to the law of cosines, if θ is the “external”
angle,
r02 = r2 + (2ae)2 + 2r(2ae) cos θ. (10.20)
0
Substituting r = 2a − r into Equation (10.20) gives
(2a − r)2 = r2 + 4a2 e2 + 4aer cos θ,
and a little bit of algebra leads to
a(1 − e2 )
r= . (10.21)
1 + e cos θ
This is the equation for an ellipse in plane polar coordinates. If we
were to go through the same process for hyperbolas and parabolas, we
would obtain similar equations for r = r(θ). In fact, the equation for
any conic section can be expressed in the form
p
r= , (10.22)
1 + e cos θ
where p = a(1 − e2 ) for an ellipse, p = a(e2 − 1) for a hyperbola8 and
p = a for a parabola (a is a characteristic constant for each type of
curve and, in general, p = l2 /GM m2 .) The various curves correspond
to different choices of the eccentricity e. Specifically,
e = 0 gives a circle,
e < 1 gives an ellipse,
e = 1 gives a parabola,
e > 1 gives a hyperbola.
The semimajor axis a determines the size of an ellipse and the
eccentricity determines its shape. At zero eccentricity, the ellipse de-
generates into a circle. As the eccentricity gets closer and closer to one,
the ellipse gets flatter and flatter. That is, as e approaches unity, the
8A hyperbola has two branches and is described by the equation
a(e2 − 1)
r= ,
±1 + e cos θ
where the “plus branch” corresponds to an attractive force and the “minus branch”
corresponds to a repulsive force.
 10.8. THE EQUATION OF AN ELLIPSE 345

ratio of the semiminor axis to the semimajor axis approaches zero. As

you will prove in Exercise 10.11,
b
1/2
= 1 − e2 . (10.23)
a
You probably know from geometry that the area of an ellipse is
given by S = πab. It is sometimes convenient to write this in terms of
a only, thus, using Equation 10.23,

1/2
S = πa2 1 − e2 . (10.24)
Having considered the properties of conics, let’s go back to the
motion of planets. Comparing the equation of an ellipse (10.21) with
the equation of the orbit (10.17), you can see that the orbit is an ellipse
(or, more generally, a conic section). A term by term comparison of
these two equations shows that
l2
= a(1 − e2 ), (10.25)
GM m2
and that
Al2
= e.
GM m2
But we showed in Equation (10.19) that
1
(GM m)2 m2 2Em 2

A= + 2 ,
l4 l
so
(GM m)2 m2 l4 2Em l4
e2 = + .
l4 (GM m)2 m2 l2 (GM m2 )2
Therefore,
2El2
e2 = 1 + . (10.26)
m(GM m)2
This gives us an expression for the eccentricity in terms of constants of
the motion. Note that for E > 0 we get e > 1, a hyperbola. For E = 0
we get e = 1, a parabola. For E < 0, we get e < 1, an ellipse.
Plugging Equation (10.26) into (10.25) gives the following expres-
sion for the semimajor axis:
GM m
a=− . (10.27)
2E
The minus sign is necessary because the total energy is negative for an
ellipse. Equation 10.27 shows that the semimajor axis depends only on
the energy.
346 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

Worked Example 10.3. A comet of mass m starts from in-

finity with velocity v0 and impact parameter b. (If undeflected, the
path of the comet would be a straight line passing the Sun at a
distance b.) (a) Show that the distance of closest approach to the
Sun is approximately b2 v02 /GM. (b) Write the equation of the orbit
in polar coordinates in terms of M, v0 , and b, where M is the mass
of the Sun.
Solution: (a) Let the Sun-comet distance at perihelion be de-
noted d. At that point, the speed of the comet is vd and its velocity
is perpendicular to the Sun-comet line. By conservation of angular
momentum l∞ = ld , and
mv0 b = mvd d.
By conservation of energy
1 2 1 2 GM m
mv = mv − ,
2 0 2 d d
so  2
b 2GM
1= − .
d dv02
Multiply by d2 and obtain the quadratic
2GM
d2 + 2
d − b2 = 0,
v0
s 2
GM GM
d=− 2 ± + b2 .
v0 v02
Since d > 0, we use the positive sign and write
1/2 !
b2 v04

GM
d= 2 −1 + 1 + 2 2 .
v0 GM
Apply the binomial expansion to get
1 b2 v04
 
GM
d = −1 + [1 + + ···] ,
v02 2 G2 M 2
GM 1 b2 v04 b2 v02
d ' = .
v02 2 G2 M 2 2GM
 10.8. THE EQUATION OF AN ELLIPSE 347

(b) In polar coordinates the hyperbolic orbit under an attractive

force is
a(e2 − 1)
r= .
1 + e cos θ
The problem asks us to express a and e in terms of the given
parameters, namely v0 , b and constants. The speed of the comet
squared is
v 2 = ṙ2 + r2 θ̇2 .
Taking the derivative of r we have
[a(e2 − 1)]
ṙ = e sin θθ̇,
(1 + e cos θ)2
Now l = mr2 θ̇ so
l l (1 + e cos θ)2
θ̇ = =
mr2 m a2 (e2 − 1)2
and we can write
e2 l2 /m2
ṙ2 = sin2 θ.
[a(e2 − 1)]2
The second term in v 2 is r2 θ̇2 . But,
1 l l 1 + e cos θ
rθ̇ = = ,
rm m a(e2 − 1)
l2 1 + 2e cos θ + e2 cos2 θ
∴ r2 θ̇2 = ,
m2 [a(e2 − 1)]2
so
2 l2 (1 + 2e cos θ + e2 (sin2 θ + cos2 θ))
v =
m2 [a(e2 − 1)]2
l2 2 + 2e cos θ + e2 − 1
 
=
m2 a(e2 − 1) a(e2 − 1)
l2 e2 − 1
 
1 + e cos θ
= 2 +
m2 a(e2 − 1) a(e2 − 1) a(e2 − 1)
l2
 
2 1
= 2 2
+ .
m a(e − 1) r a
a(1−e2 )
For an elliptical orbit, r = 1+e cos θ
and l2 = GM m2 a(1 − e2 ).
a(e2 −1)
Similarly, for a parabolic orbit r = 1+e cos θ
, so l2 = GM m2 a(e2 −1).
348 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

Consequently
GM m2 a(e2 − 1) 2 1
   
2 2 1
v = + = GM + .
m2 a(e2 − 1) r a r a
At r = ∞, the velocity is v0 and we see that
GM
v02 = ,
a
so
GM
a= 2 .
v0
Also
l2
a(e2 − 1) = .
GM m2
Solving for e we obtain
 2 !1/2
v0 b
e= 1+ ,
(GM )
and finally
v02 b2 /GM
r= r  2 2 .
bv0
1+ 1 + GM cos θ

Exercise 10.10. Starting with Equation (10.20), obtain Equa-

tion(10.21).
1/2
Exercise 10.11. Show that for an ellipse b/a = (1 − e2 ) . (Hint:
Apply the Pythagorean theorem to the triangle FOP in Figure 10.11.)

P
b
ea ea
F' 0 F

Figure 10.11. An ellipse. Note that the distance from

the center to the focal point is ea and that the sum of
the distances F P and F 0 P is 2a.

Exercise 10.12. A planet is in an elliptical orbit with semimajor

axis a. By averaging the largest and smallest values of r, show that it
 10.9. KEPLER’S LAWS REVISITED 349

leads to an “average” value of the potential energy equal to −GM m/a.

(Compare with the virial theorem discussed in Worked Example 10.2.)
Exercise 10.13. Plot Equation (10.22) for e = 2, e = 1.0, e = 0.5,
and e = 0.

10.9. Kepler’s Laws Revisited

Having applied Newton’s second law and Newton’s law of univer-
sal gravitation to the problem of planetary motion, let us now return
to Kepler’s laws that were given in Section 10.2. You will find that
Kepler’s laws are simply descriptions of some aspects of the theory we
have developed. Recall that the first law states:
The orbit of a planet is an ellipse with
the Sun at one of the focal points.
You have seen that the trajectory of a particle under the action of an
inverse square central force is a conic section. Therefore, Kepler’s first
law is proved. Actually the proof is even more general than Kepler’s
first law, for we have shown that objects in the solar system acting
under the inverse square gravitational force will move in a trajectory
which may be an ellipse, but it could also be one of the other conic
sections.
Kepler studied the orbit of Mars all his life, and he deduced other
important facts about its motion. For one thing, he found that the
motion of Mars as a function of time was not uniform. Mars speeds up
as it approaches the Sun, moves fastest at its point of nearest approach,
then slows down as it moves away from the Sun. He expressed this fact
in a quantitative way in his “Law of Areas” or, as we call it, Kepler’s
second law:
The radius vector of a planet sweeps out
equal areas in equal times.
This is illustrated in Figure 10.12. If the planet moves from a to
b in the same time as it goes from c to d, then according to Kepler’s
second law, the two shaded areas are equal. So another way to state
Kepler’s second law is, “The areal velocity of a planet is constant.”
In point of fact, we already proved Kepler’s second law, but you
might not have noticed it because it was stated in quite a different way.
We said, “For a central force, the angular momentum is constant.”
Are the two statements equivalent? Does the fact that the angular
momentum is constant imply that the areal velocity is constant? The
350 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

a
d

c
b

Figure 10.12. Kepler’s second law states that equal

areas are swept out in equal times.

answer is yes. To prove that this is true, consider a planet located

at vector position r. After a time interval dt the planet will be at
r+dr = r + vdt. See Figure 10.13. The shaded region is the area swept
out. Recall that |a × b| is equal to the area of a parallelogram whose
sides are the vectors a and b. The shaded region in Figure 10.13 is one
half of the parallelogram formed from r and dr. Therefore, the area
swept out by the planet’s radius vector in time dt is
1 dt
dS = |r × vdt| = |r × v| .
2 2
But by the definition of angular momentum, |r × v| = l/m. Therefore,
1 l
dS = dt,
2m
or
dS l
= areal velocity = .
dt 2m

dr
r
r +d
r

Figure 10.13. The area of the triangle is one half the

area of the parallelogram. Therefore,
∆S = (1/2) |r × dr| = (1/2) |r × v∆t| .

The mass of a planet is constant, so the areal velocity is proportional

to the angular momentum and dS dt
= constant is completely equivalent
to the relation l = constant. Kepler’s second law is simply a conse-
quence of the fact that in central force motion, the angular momentum
is conserved.
Kepler’s third law states:
 10.9. KEPLER’S LAWS REVISITED 351

The period of a planet squared is

proportional to its semimajor axis cubed.
In mathematical terms, if τ is the time required for a planet to orbit
the Sun and if a is the semimajor axis of the orbit, the third law can
be written
τ 2 = Ka3 ,
where K is some constant. Kepler discovered this law after years of
studying planetary data. On the other hand, our analytical study of
elliptical motion leads to it in a fairly simple way. We just note that
I I
dS l l
= ⇒ dS = dt.
dt 2m 2m
orbit orbit

Let S equal the area of the ellipse. Then if τ is the period,

l
S= τ.
2m
But we have seen that S = πa2 (1 − e2 )1/2 so
l
τ = πa2 (1 − e2 )1/2 .
2m
According to Equation(10.25), (1 − e2 ) = l2 /GM m2 a. Squaring we
obtain
l2 2 2 4 l2
τ = π a
4m2 GM m2 a
or
4π 2 3
τ2 = a.
GM
This relation states that the period squared is proportional to the
semimajor axis cubed. Thus, Kepler’s third law is proved. Note that
the constant of proportionality depends only on the mass of the Sun,
so the ratio τ 2 /a3 should be the same for all the planets. Therefore,
determining the period of a planet or asteroid allows us to evaluate
its semimajor axis. Actually, this result is not completely accurate.
There is a very small correction that is due to the fact that the Sun is
not really at rest at the origin. The Sun and the planet move around
their common center of mass. This point is almost, but not quite, at
the center of the Sun. This introduces a small correction into Kepler’s
third law so that the quantity M in the denominator should be re-
placed by M + m. Once again we appreciate the technique in physics
of first solving an easier idealized problem and introducing the com-
plications later. (Problem 10.7 asks you to carry out the appropriate
calculations.)
352 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

Worked Example 10.4. A particle moves in an elliptical or-

bit with semimajor axis a and eccentricity e. The velocity of the
particle is observed to vary from a minimum value v1 to a maximum
value v2 . Determine the period.
Solution: The angular momentum of the particle is constant:
l = mvr = constant.
Therefore, the maximum (and minimum) speeds occur when r is
at the smallest (and greatest) distance from the the force center.
Call these distances r1 and r2 . Note that
mv1 r1 = mv2 r2 .
Furthermore, it is easy to appreciate that the points at r1 and r2 as
well as the force center all lie on the semimajor axis. Consequently,
r1 + r2 = 2a.
Hence
l = mv1 (2a − r2 ) = mv2 r2 ,
2av1
∴ r2 = ,
v1 + v2
so the angular momentum can be written
2av1 v2
l=m .
v1 + v2
The angular momentum is related to the areal velocity by
l dS
= = areal velocity.
2m dt
The period is equal to the area divided by the areal velocity
area πab πab(2m) πb(v1 + v2 )
τ= = = 2av1 v2 = .
areal velocity (l/2m) m v1 +v2 v1 v2
But b is the semiminor axis and equal to a(1 − e2 )1/2 , so
v1 + v2
τ = πa(1 − e2 )1/2 .
v1 v2

Worked Example 10.5. Suppose you were familiar with Ke-

pler’s work, so you knew that the orbits of planets are ellipses and
 10.9. KEPLER’S LAWS REVISITED 353

that the angular momentum is constant. This suggests to you that

the force on the planet is a central force, so F = f (r), but you
don’t know any other properties of the force. Show that the force
obeys the inverse square law, that is, show that f (r) ∝ 1/r2 .
Solution. The radial equation of motion Equation (10.2) can
be written
r̈ − rθ̇2 = −f (r)/m,
and the angular equation of motion leads to
θ̇ = l/mr2 ,
so
l2 f (r)
r̈ − 2 3
=− .
mr m
2 2 2
In terms of u = 1/r, recalling that r̈ = − lmu2 ddθu2 , we obtain
l2 u2 d2 u l2 f (r)
− 2 2
− 2 3
=− ,
m dθ mu m
d2 u m
2
+ u = 2 2 f (r). (10.28)
dθ l u
Recall that the equation of an ellipse has the form
p
r= ,
1 + e cos θ
1 e
∴ u = + cos θ.
p p
Therefore
d2 u e
2
= − cos θ,
dθ p
and Equation (10.28) becomes
 
e 1 e m
− cos θ + + cos θ = 2 2 f (r)
p p p l u
1 m m
= 2 2 f (r) = 2 r2 f (r).
p l u l
That is,
l2 1 1
f (r) = 2
∝ 2
mp r r
Q.E.D.
354 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

Worked Example 10.6. A uniform spherical planet of radius

a revolves about the Sun in a circular orbit of radius r0 and rotates
about its axis with angular velocity ω0 , normal to the plane of
the orbit. As noted in the discussion accompanying Figure 7.3,
tides raised on the planet by the Sun cause its angular velocity of
rotation to decrease. Find a formula expressing the orbit radius r
as a function of angular velocity ω of rotation at any later or earlier
time. Apply your formula to the Earth, neglecting the effect of the
Moon, and estimate how much further the Earth will be from the
Sun when the day has become equal to the present year.
Solution: By Kepler’s third law τ 2 ∝ r3 . If the period of the
planet is denoted Y (for “year”), Kepler’s law is
Y = kr3/2 ,
with k a proportionality constant. Let Y0 = length of present year,
3/2
so Y0 = kr0 .
The total angular momentum, orbital plus rotational, is con-
stant, so
mr2 Ω + Iω = constant,
where Ω = 2π/Y = orbital angular velocity, and ω = rotational
velocity of the planet = 2π/d where d = 1 day. The moment of
inertia of a sphere about a diameter is I = (2/5)ma2 . So
mr2 2π/Y + (2/5)ma2 ω = mr02 2π/Y0 + (2/5)ma2 ω0 .
Therefore,
r2 r2 2 1 2
= 0 + a (ω0 − ω).
Y Y0 5 2π
3/2
But 1/Y = (1/k)r−3/2 and k = Y0 r0 , so
−3/2 k 2
r2 r−3/2 /k = r02 r0 /k + a (ω0 − ω),
5π
Y 0 a2
 
1/2 1/2
r = r0 1+ (ω0 − ω) .
5π r02
Square both sides to get
2
Y 0 a2

r = r0 1 + (ω0 − ω) .
5π r02
Applying to Earth, the numerical values are: a = 6.37 × 106 m,
r0 = 1.50 × 1011 m. Note that
2π 2π
(ω0 − ω) = − ,
d0 d
 10.9. KEPLER’S LAWS REVISITED 355

where d0 = 1 day = Y0 /365.25 years and d = 1 year. In terms of

Y0 ,
2π(365.25) 2π 364.25
(ω0 − ω) = − = 2π .
Y0 Y0 Y0
So
2
Y0 a2 2π

r = r0 1 + (364.25) = r0 (1 + 2.63 × 10−7 )2 .
5π r02 Y0
Using the binomial expansion, (1 + x)n ≈ 1 + nx (for small x),
r = r0 (1 + 2(2.63X10−7 )) = r0 (1 + (2)(2.63X10−7 )
= r0 + 78951 (meters).
Therefore the increased distance between Earth and Sun is only
about 79 kilometers.
If you include the effect of the Moon which is causing the day
to grow longer, the time for the day to equal the year would be
even less and the distance the Earth would move away from the
Sun would also be less.

Exercise 10.14. A certain asteroid in the solar system has a period

of four years. What is its semimajor axis in AU? (Assume you do not
know the gravitational constant G or the mass of the Sun, but you do
know that the Earth is at a distance of 1 AU from the Sun.) Answer:
2.52 AU.

Exercise 10.15. (a)Using the form τ 2 = (4π 2 /GM )a3 and assum-
ing you do not know the mass of the Sun, determine the semimajor
axis of Saturn in AU, given its period is 29.5 Earth years. (b)Using
the correction (M + m) and looking up the appropriate values, obtain
a corrected value. Answers: 9.55 AU, 9.52 AU

Exercise 10.16. For a planet in a circular orbit, F = ma can be

written as GM m/r2 = mv 2 /r. Use this to derive Kepler’s third law for
circular orbits.

Exercise 10.17. Halley’s comet has an eccentricity of 0.967. Its

perihelion distance is 8.81×1010 m. What is its period? Answer: 75.4
years.
356 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

10.10. A Perturbed Circular Orbit

In this section you will study the stability of a circular orbit. When
a physical system in an equilibrium state is perturbed by a small force,
it will often oscillate about equilibrium. This is considered a stable
equilibrium. On the other hand, if the perturbation causes the system
to undergo a large change, the equilibrium is unstable. (The usual
example of stable equilibrium is a marble at the bottom of a depression,
and unstable equilibrium would be a marble balanced on the peak of a
hill.) This section is both an application of the concepts studied in the
previous sections and an introduction to the methods for dealing with
small perturbations.
Consider a particle moving in a circular orbit under the action of
a central force. For reasons that will eventually become clear, I will
not require the force to obey the inverse square law. That is, although
F = f (r)r̂, the functional form of f (r) will be left undetermined for
now.
The equations of motion are given by generalizations of Equations
(10.2):

m(r̈ − rθ̇2 ) = f (r), (10.29)

m(rθ̈ + 2ṙθ̇) = 0.
As before,
l
θ̇ = , (10.30)
mr2
with l = constant. Inserting this expression for θ̇ into the radial equa-
tion of motion (Equation 10.29), leads to
l2
mr̈ = f (r) + . (10.31)
mr3
If the particle is moving in a circular orbit of radius a, then

r = a = constant.
Consequently,
r̈ = 0.
The radial equation of motion (Equation 10.31) then reduces to
l2
f (a) = − . (10.32)
ma3
I will come back to this equation shortly.
Now consider the question of the stability of a perturbed circular
orbit. For example, we might be considering the motion of a planet in a
 10.10. A PERTURBED CIRCULAR ORBIT 357

perfectly circular orbit which is hit by a comet (as happened some years
ago when comet Shoemaker-Levy collided with Jupiter). We want to
know if the planet will continue moving in a stable orbit after a slight
perturbation, or if it will behave in an erratic manner. In other words,
we would like to know if a collision with a relatively small body could
cause Jupiter to go flying out of the solar system.
After a collision or some other sort of perturbation, the radial po-
sition is no longer exactly equal to a, but it is still nearly equal to a,
so we can write
r = a + η, (10.33)
where η << a, i.e., η is a very small quantity.
Inserting expression (10.33) for r into the radial equation of motion
in the form (10.31) we obtain
d2 l2
m (a + η) = f (a + η) + ,
dt2 m(a + η)3
or
l2
mη̈ = f (a + η) + . (10.34)
ma3 (1 + ηa )3
The force at the position r = a + η is very nearly equal to the force
at r = a so we are justified in expanding f (a + η) in a Taylor’s series
expansion and keeping only the leading terms. Thus:
1 2 d2 f 
 
df 
f (a + η) = f (a) + η  + η + · · ·.
dr r=a 2 dr2 r=a
The term (1 + η/a)3 in the denominator of the last term of Equation
(10.34) can also be expanded. Using the binomial expansion:9
1  η −3 η  η 2
= 1 + = 1 − 3 + 6 + · · ·.
(1 + η/a)3 a a a
Consequently, Equation (10.34) can be written (to first order in η) as
l2

df  η
mη̈ = f (a) + η  + · · · + 3
(1 − 3 + · · ·),
dr a ma a
9If
you have not already done so, you should immediately memorize the follow-
ing extremely important series expansions:
Taylor’s Series:
1 2 d2 F  1 3 d3 F 
  
dF 
F (a + δx) = F (a) + δx + δx + δx + ··· .
dx 
a 2! dx2 
a 3! dx3 
a
Binomial Expansion:
n(n − 1) 2 n(n − 1)(n − 2) 3
(1 + x)n = 1 + nx + x + x + ··· .
2! 3!
358 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

or
l2 η l2

. df 
mη̈ = f (a) + η  + − 3 .
dr a ma3 a ma3
Now recall that Equation (10.32) states that f (a) = −l2 /ma3 , so the
first and third terms on the right hand side cancel. This leaves
η l2

. df 
mη̈ = η  − 3
dr a ma3
 a
3l2

df 
= η − ,
dr a ma4
or  2  
3l 1 df 
η̈ + η − = 0. (10.35)
m2 a4 m dr a
This equation has the form
η̈ + Kη = 0.
We will be considering equations of this form in detail later, but for
now it is sufficient to note that the general form of the solution depends
on whether K is positive or negative. For positive K the solution has
the form √ √
η = A sin Kt + B cos Kt, (10.36)
and for negative K the general form of the solution is
√ √
η = Ce+ |K|t + De− |K|t . (10.37)
Thus, if K > 0, the solution is simple harmonic motion (see Section
3.6). On the other hand, if K is negative, the solution is the sum of an
exponential decrease plus an exponential increase. The exponentially
decreasing term quickly dies out and η(t) increases with time exponen-
tially. We can interpret this result physically to mean that for K < 0
the orbit is unstable because the “distance” η from the radius a of a
circular orbit grows without bounds. On the other hand, if K > 0,
then η oscillates back and forth around zero, meaning that the parti-
cle oscillates about r = a in simple harmonic motion. Therefore, the
condition for a stable orbit is that K > 0. That is,
3l2

1 df 
− > 0.
m2 a4 m dr  r=a

Once again we can use the fact that l /ma3 = −f (a), and rewrite this
2

equation as 
3 df 
− f (a) −  > 0. (10.38)
a dr r=a
 10.10. A PERTURBED CIRCULAR ORBIT 359

df
The gravitational force is f = −GM m/r2 and dr = +2GM m/r3 . So,
for gravity, Equation (10.38) becomes
 
3 GM m 2GM m GM m
− 2
− 3
= + 3 > 0,
a a a a
and consequently, the gravitational force leads to stable orbits.
But what if the force is not inverse square? A more general central
force might have the form
c
f =− .
rn
Here n is an integer that can be greater or less than zero. (For the
gravitational force, n = 2; notice that I explicitly incorporated the
negative sign on the right hand side because only attractive force laws
lead to orbital motion.)
If f = − rcn , then

df  cn
= n+1 .
dr a a


Consequently, the stability condition (10.38) is

3  c  cn
− − n  − n+1 > 0
a r a a
3 −c cn
− − > 0
a an an+1
c
n+1
(3 − n) > 0
a
n−3 < 0

or
n < 3.
The inverse square force law (n = 2) leads to orbits that are stable
under small perturbations. If the gravitational force had the form
F ∝ 1/r4 , a planetary system would be impossible because any tiny
perturbation would cause the planets to spiral away from their orbits.
Notice that in determining the stability of circular orbits, we also
solved the problem of determining the frequency of small oscillations
in a perturbed orbit because Equation (10.35), the equation of motion
for η, has the form
..
η + Kη = 0,
360 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

and, as mentioned previously, this is the equation for√simple harmonic

motion. The frequency of harmonic motion is ω0 = K, so
3l2

2 1 df 
ω0 = −
m2 a4 m dr r=a

3 1 df 
= − f (a) − .
ma m dr  r=a
For a particle subjected to an inverse square force law, such as the
gravitational force,
f = −cr−2 ,

df  2c
= + 2cr−3 r=a = 3 .

dr r=a
 a
Also
c
f (a) = − .
a2
Consequently,
 
3  c 1 2c c
ω02=− − 2 − 3
= . (10.39)
ma a m a ma3
This equation gives the frequency of the radial oscillations, Note that
2π/ω0 is the time for the planet to go from perihelion through aphelion
and back to perihelion. While carrying out this radial motion, the
planet has an angular velocity given by

l  l
θ̈ = = .
mr r=a ma2
2 
Therefore,
l2 −f (a)ma3 −(c/a2 )ma3 c
θ̈2 = 2 4 = 2 4
= − 2 4
= . (10.40)
ma ma ma ma3
Comparing Equation (10.39) with Equation (10.40) we see that ω0 = θ̇,
so for a 1/r2 force law, the radial oscillations have exactly the same
frequency as the orbital motion. Therefore, in the time required for
the particle to go completely around in its orbit, it will go through
one complete radial oscillation. The combination of these two motions
converts a circular orbit into an elliptical orbit. You might think that
the eccentric orbits of planets are a result of collisions with asteroids
and comets. However, astronomers do not believe that this accounts
for the eccentricities of planetary orbits.
It can happen that an orbiting particle is perturbed in such a way
that the orbital period and the period of radial oscillation are not
equal. For example, the Earth is not a perfect sphere; it is an oblate
 10.11. RESONANCES 361

spheroid and bulges slightly at the equator. This bulge exerts a force
on an orbiting artificial satellite, and gives rise to a radial oscillation of
frequency ωr that is not equal to θ̇. The satellite moves in an elliptical
orbit, but since the radial motion is not exactly synchronized with
the angular motion, the ellipse slowly precesses. In other words, the
major axis of the ellipse slowly rotates at a rate which is the difference
between the two oscillations. The angular frequency of the precession
of the ellipse, ωp is given by

ωp = ωr − θ̇
where ωr is the frequency of the radial oscillations.

Exercise 10.18. Draw the orbit of a planet assuming the radial

oscillation has a period half that of the angular motion.
Exercise 10.19. By simple substitution, show that Equations 10.36
and 10.37 are solutions to the simple harmonic motion differential equa-
tion.

Figure 10.14. Commensurable and incommensurable

orbits. The two orbits on the left are commensurable
and sooner or later the planet traces out the same path.
The orbit on the right is incommensurable and the planet
never repeats exactly the same path.

10.11. Resonances
An orbit is usually thought of as the path of a particle that retraces
its motion over and over again, as in the case of a single planet around
a star. If the particle always passes through the same points, the orbit
is said to be closed. In a simple closed orbit, the radial period and the
362 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

angular period are equal. That is, the planet goes from one turning
point to the other and back again in the same time as the angular
displacement goes from 0 to 2π. If the radial period is slightly greater
than the angular period, the planet will reach perihelion a short time
after its angular position has advanced 2π. The orbit is not closed
and the perihelion precesses. The orbit can actually be closed even
if the two periods are not equal as long as they are in the ratio of
integers, as in the middle picture of Figure 10.14. When two periods
or (equivalently) two frequencies are in the ratio of small integers, they
are said to be commensurable or in resonance. If the ratio of the radial
frequency
√ to the angular frequency is an irrational number such as π
or 2, the orbit is not closed; it will never repeat itself. Figure 10.14
illustrates two commensurable orbits and one incommensurable orbit.
Resonances are very important in celestial mechanics. One often
hears of a system being “locked” into a particular resonance. For ex-
ample, the rotation of the Moon is locked into a 1:1 resonance with
its orbital motion. Therefore, the same side of the Moon always faces
Earth. Similarly, the orbit of a 24 hour satellite is in a 1:1 resonance
with the rotation of the Earth. An interesting case is the 2:5 resonance
between the periods of Jupiter and Saturn. In old fashioned terminol-
ogy used in celestial mechanics, this resonance is called “The Great
Inequality.”

10.12. Summary
A central force is directed towards or away from the origin and has a
magnitude that depends only on the distance to the origin, F = f (r)r̂.
An example is the gravitational force acting on a planet as it orbits the
Sun.
Kepler studied the motion of the planets and determined that this
motion obeys three relations which we call Kepler’s Laws.
A particle moving in a central force field has constant angular mo-
mentum. The equations of motion for a particle of mass m in the field
of a body of mass M are
l2 GM
r̈ − 2 3
= − 2
mr r
d dl
(mr2 θ̇) = = 0.
dt dt
The effective potential energy for a particle in a gravitational field
is
l2 GM m
Vef f = 2
− .
2mr r
 10.13. PROBLEMS 363

The radial motion of such a particle can be visualized on a plot of Vef f

vs r, such as Figure 10.5. Such plots indicate that the orbit of the
particle (planet) is a conic section.
The equations of motion can be used to obtain the position of the
planet as a function of time or to determine the equation of the orbit,
r = r(θ). If the total energy is negative, the equation of the orbit is
l2 /GM m2
r= Al2
,
1 + GM m2
cos(θ − θ0 )
which has the form of the equation for an ellipse,
a(1 − e2 )
r= .
1 + e cos θ
Using these relations one can relate Kepler’s laws to physical concepts
as follows.
First Law: Planets move in elliptical orbits. This is a consequence
of the inverse square nature of the gravitational force.
Second Law: Planets sweep out equal areas in equal times. This is
a consequence of the conservation of angular momentum in a central
force field.
Third Law: The period squared is proportional to the semimajor
axis cubed. This is a consequence of the conservation of energy and the
fact that the magnitude of the total energy is inversely proportional to
the semimajor axis.
When a planet in a circular orbit is perturbed, it will oscillate radi-
ally around the circular path, converting the orbit into an ellipse. The
orbit is stable for any force law having the form f = −crn as long as n
is greater than -3.

10.13. Problems
Problem 10.1. (Bohr model of the atom.) Consider an electron in
a circular orbit around a proton. Assume the angular momentum can
only take on values equal to nh where n is an integer (n = 1, 2, 3, ...)
and h is a constant. Determine the possible values of the radius for the
orbit and the possible values for the total energy. Make up a table of
the energy for the first four energy states (that is, for the electron in
the four smallest orbits).
Problem 10.2. Integrate Equation (10.15) directly and show that
it leads to the equation of an ellipse. Note that α = −1, β is positive
and γ is negative.
364 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

Problem 10.3. A satellite is in an elliptical orbit around Earth.

Somehow you determine that the maximum and minimum speeds of the
satellite are v1 and v2 . Find the values of a, e, the angular momentum
per unit mass, and the period in terms of v1 , v2 , G, and ME , the mass
of Earth.
Problem 10.4. An interplanetary spacecraft is parked in an el-
liptical orbit about the Sun. The period is τ. The rocket motors
are fired in a short burst, and the speed of the spacecraft increases
from V to V + ∆V. What is the change in the period? (Answer:
∆τ = 3(2πGM )−2/3 τ 5/3 V ∆V where M is the mass of the Sun.)
Problem 10.5. (a) By some magical process, the Sun suddenly
loses half of its mass. Show that the Earth’s orbit will be a parabola
(and so the Earth will escape to infinity). (b) By some other magical
process, the Sun’s mass suddenly doubles. What will the Earth’s orbital
period be in this case?
Problem 10.6. Explain why the Moon raises two tides, on oppo-
site sides of the Earth (one centered on the sub-lunar point and the
other on the opposite side). It is conceptually somewhat simpler to
consider the tides raised by the Sun, because the distance to the center
of mass of the Sun-Earth system is essentially the same as the Sun-
Earth distance whereas the center of mass of the Moon-Earth system
is inside the Earth. You may ignore the daily rotation of the Earth,
but not its rotation about the Sun. (Hint: Draw the vector forces act-
ing on particles on the surface of Earth and resolve into a component
parallel to the Sun-Earth line to yield the centripetal force and note
the directions of the unbalanced force components.)
Problem 10.7. A certain binary star system is composed of two
stars of comparable masses m1 and m2 . (For this system we cannot
assume one mass is infinitely greater than the other.) Let r1 and r2 be
the positions of m1 and m2 relative to an inertial origin and let r01 and
r02 be their positions relative to the center of mass. (a) Show that the
primed coordinates are related to the relative coordinate r by
m2
r01 = − r,
m1 + m2
m1
r02 = + r,
m1 + m2
where the relative coordinate r points from m1 to m2 .(b) Obtain the
Lagrangian in terms of the relative coordinate r and the position of the
center of mass R. (c) Obtain the equations of motion in terms of the
relative coordinate. (You may assume the center of mass remains at
 10.13. PROBLEMS 365

rest.) (d) Express the radial equation (for the magnitude of r) in terms
of the angular momentum and the reduced mass. (e) By comparing the
radial equation with Equation (10.8), obtain an expression for Kepler’s
third law for this situation.
Problem 10.8. Show that for a planet in an elliptical orbit about
the Sun, the radial velocity at any time is given by
 
2 2 GM
r ṙ = (a[1 + e] − r) (r − a[1 − e]) .
a
where M is the mass of the Sun. (Hint: Use Equations (10.27) and
(10.25).)
Problem 10.9. A particle of mass m is acted upon by an attractive
central force given by K/r4 . The particle ispplaced a distance a from
the force center and given an initial velocity 2K/3ma3 at right angles
to the radius vector. (a) Show that the particle spirals into the force
center by deriving the equation for the orbit. (b) Determine the time
for the particle to collide with the force center. (Hint: For this force,
3
p energy is V = −K/3r .) Answers: (a) r = (a/2)(1+cos θ),
the potential
(b) 3π/8 3ma5 /2K.
Problem 10.10. Consider a central force given by F (r) = −K/r3
with K > 0. Plot the effective potential and discuss possible types of
motion.
Problem 10.11. A particle is subjected to a central force
K K0
F (r) = − 2 + 3 .
r r
Assume K > 0 and consider both signs for K 0 . (a) Draw the effective
potential and discuss possible types of motion. (b) Solve the orbital
equation and show that the bounded orbits have the form
a(1 − e2 )
r=
1 + e cos αθ
2 0
as long as l > −mK .
Problem 10.12. The first artificial satellite of the Earth was the
Russian Sputnik I. Its perigee was 227 km. above the Earth’s surface.
At this point its speed was 28,710 km/hr. Determine its period of
revolution. What is the maximum distance from the satellite to the
surface of the Earth? Given: the Earth has a radius of 6.37X103 km.
Problem 10.13. A certain satellite has a perigee of 360 km and an
apogee of 2549 km above the Earth’s surface. Find its distance above
366 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

the surface when it is 90◦ from perigee, as measured from the center of
Earth.
Problem 10.14. For the two body problem there is a conserved
quantity called the Laplace vector. (It is also called the Runge-Lenz
vector.) The Laplace vector is a vector pointing towards periapsis. Its
magnitude is proportional to the eccentricity. It can be expressed as:
A = p × l−GM m2 r̂
where M is the mass of the primary, p is the linear momentum and l
is the angular momentum.
a) Show that A lies in the orbit plane.
b) Show that A is a constant of the motion.
c) Show that the magnitude of A is GM m2 e. (Hint: Evaluate r · A.)
Problem 10.15. Imagine a circular orbit passing through the ori-
gin. Then r = A cos θ. Show that a central force that gives rise to such
an orbit has a 1/r5 dependence.
Problem 10.16. A particle moves under the action of a central
force in a spiral described by r = Aebθ . Show that the force acting on
it is inversely proportional to the cube of r.
Problem 10.17. A satellite is in a circular orbit, a distance r from
the center of Earth. It has a known velocity
√ v. Show that this satellite
has an escape velocity (from r to ∞) of 2v.
Problem 10.18. Two stars of equal mass orbit around their com-
mon center of mass with period τ. Show that if they were suddenly
stopped, dead still, √
and allowed to fall toward each other, they would
collide in a time τ / 32.
Problem 10.19. Consider a binary star system. (a) Show that
Kepler’s third law can be written as
a3 = τ 2 (M1 + M2 )
where a is in AU (astronomical units) τ is in Earth years and M1 + M2
is the sum of the masses of the stars in solar masses. (Note: All the
equations we developed in this chapter assumed M >> m. Hint: Red-
erive Equation (10.1) and show that if the two masses are comparable,
that GM should be replaced by G(M1 + M2 ) everywhere.) (b) Star
A and star B are members of a binary star system. An astronomer
determines that the period of this system is 32 years and that the stars
are separated by 16 AU. Furthermore, star A is found to be 12 AU
from the center of mass of the system. Determine the masses of these
 10.13. PROBLEMS 367

stars. (Note: This is one of the ways in which astronomers evaluate

the masses of stars) Answer: Mass of A = 1 solar mass, mass of B = 3
solar masses.
Problem 10.20. The Rutherford problem consists in the motion
of an alpha particle interacting with a gold nucleus. The force is a
repulsive force of magnitude F = K/r2 .(a) Show that the eccentricity
can be expressed as
2El2
e2 = 1 + .
mK 2
(b) Show that the distance of closest approach is
r1 = a(e + 1).
(Hint: Recall that for a repulsive force the orbit is the “minus branch”
of the hyperbola.)
Problem 10.21. A body is moving in an elliptical orbit. Its max-
imum velocity is vmax and its minimum velocity is vmin . Express the
eccentricity of the orbit in terms of vmax and vmin .
Problem 10.22. Assume the space shuttle is in a circular orbit
about Earth with its nose pointing forward. The rocket motors are
fired for a few moments. You would expect the shuttle to speed up.
Show that, instead, it actually rises to a higher orbit and slows down.
(This behavior is called the “satellite paradox.”)
Problem 10.23. Assume the force between two particles is a cen-
tral attractive force proportional to the separation between the parti-
cles. That is, F = −krr̂. You can assume one of the particles is much
more massive than the other one. Show that the motion of the lighter
mass is an ellipse with the more massive particle at the center of the
ellipse (rather than at a focal point).
Problem 10.24. One of the strongest arguments made for the
general theory of relativity is that it predicts the correct value for
the precession of the orbit of the planet Mercury. Before Einstein
formulated that theory, it was suggested that the precession could be
explained if the solar system were filled with dust having a low density
ρ. The spherical distribution of dust would exert an additional central
force on a planet given by F 0 = −mKr where m is the mass of the
planet and K is a constant equal to (4π/3)ρG. Show that for F 0  F
where F is the gravitational force of the Sun, a planet will move in an
elliptical orbit whose major axis precesses at an angular velocity
3/2
p
ωp = 2πρr0 G/M,
368 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

where M is the mass of the Sun, and r0 is the average radius of the
orbit.
Problem 10.25. Consider a particle that is acted upon by the
central force:
A B
F = 2 + 4,
r r
where A and B are constants. The orbit is a circle of radius a. Deter-
mine the condition for the orbit to be stable. (Answer: a2 A > B.)
Problem 10.26. A particle is moving in a circular path under the
action of an attractive central force given by
1 −r/a
F = e .
r2
Show that if the radius of the circle is greater than a the motion is
stable, and if the radius of the circle is less than a, the motion is
unstable.
Problem 10.27. An ideal massless string passes through a small
hole in a perfectly smooth table. Two equal masses are attached to the
ends of the string so that one mass is on the table a distance a from
the hole, and the other mass is hanging freely. See the sketch (Figure
√
10.15). The mass on the table is set in motion with a velocity ga
perpendicular to the direction of the string. Show that the mass on
the table moves in a circular path. Now the hanging mass is perturbed
slightly so that it
poscillates up and down. Show that the period of this
oscillation is 2π 2a/3g.

Figure 10.15. Mass on frictionless table (see Problem 10.27)

Problem 10.28. Determine whether motion under a central force

inversely proportional to the third power is stable or unstable.
 10.13. PROBLEMS 369

Problem 10.29. Given, f (r) ∝ 1/rn . Show that closed orbits

exist for n = −2 and n = 1, but not for n = 2 or n = 3. (Hint:
For a closed orbit, the apsidal angle must be a rational fraction of 2π.
The apsidal angle is the angle through which the radius vector rotates
between apsides.)
COMPUTATIONAL PROJECTS
Computational Techniques: The Runge-Kutta Method
In Section 3.7 I described the Euler Cromer algorithm for solving
ordinary differential equations (ODE’s). Here I will describe the Runge-
Kutta method which is more accurate and perhaps more elegant, but
less transparent. It is based on the truncated Taylor Series expansion
for a function g(t):

dg 
g(t + τ ) = g(t) + τ
dt 
ξ

The second term is usually evaluated at t, but to make the solu-

tion more accurate, the Runge-Kutta technique evaluates it half way
through the time step. That is ξ = t + τ /2.
An economical way of expressing the ODE is to define two vectors,
x and f in the following way. Assume a two dimensional system so the
position is given by x and y and the velocity by vx and vy . Then
x(t) = [x(t) y(t) vx (t) vy (t)]
f (x,t) = [vx (t) vy (t) ax (t) ay (t)]
and the ODE can be written
dx
= f (x(t), t).
dt
The second-order Runge-Kutta algorithm is obtained by first defining
a new vector x∗ by:
1
x∗ = x∗ (t + τ /2) = x(t) + τ f (x(t), t).
2
Then
x(t + τ ) = τ f (x∗ , t + τ /2).
It is not too difficult to appreciate that this is equivalent to the trun-
cated Taylor Series. However, the most common ODE solver is not the
second-order Runge-Kutta, but the fourth-order Runge-Kutta, which
is also based on the Taylor Series but in a much less transparent way.
In vector form, the value of x at time t + τ is given by
1
x(t + τ ) = x(t) + τ (F1 + 2F2 + 2F3 + F4 ) ,
6
370 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

where

F1 = f (x, t),
1 1
F2 = f (x + τ F1 , t + τ ),
2 2
1 1
F3 = f (x + τ F2 , t + τ ),
2 2
1
F4 = f (x + τ F3 , t + τ ).
2
Computational Project 10.1. A comet is jostled loose from the
Oort Cloud and heads toward the Sun. Suppose that when it passes
the orbit of Pluto it has velocity components vx = −0.01, vy = 0.05
AU/year. Plot the trajectory of the comet using three algorithms:
Euler-Cromer, Second Order Runge Kutta and Fourth Order Runge
Kutta. Show that the angular momentum is constant. (The only bodies
involved are the comet and the Sun. Hint: Use GM = 4π 2 .)

Computational Project 10.2. This problem is the same as the

previous problem, but now include the effect of Jupiter (which you can
assume has a circular orbit). The trajectory of the comet lies in the
plane of the orbit of Jupiter. Use RK4.

Computational Project 10.3. Write a program to trace out the

positions of the planets as a function of time, using the information in
the following table. You may assume the planets have zero inclination
(so they all move in the same plane). Start the planets so that they are
all lined up in a straight line from the Sun. (This is called a “conjunc-
tion”.) (a) Determine the number of years between the conjunctions
of Jupiter and Saturn. (b) What fraction of the time is Pluto nearer
the Sun than Neptune? (P.S. I know Pluto is no longer a planet.)

Planet a (AU) e Mass (1024 kg)

Mercury 0.387 0.206 0.33
Venus 0.723 0.007 4.87
Earth 1.000 0.017 5.97
Mars 1.524 0.093 0.64
Jupiter 5.203 0.048 1898.6
Saturn 9.537 0.054 568.46
Uranus 19.191 0.047 86.83
Neptune 30.069 0.009 102.43
Pluto 39.482 0.249 0.013
 10.13. PROBLEMS 371

Computational Project 10.4. Compute and plot the motion of

a particle under the action of the central force
K α
F = − 3 (1 − )r
r r
where K and α are constants. Show that this orbit precesses. Show
how your choice of K and α affect the motion.
Computational Project 10.5. We consider the effect of atmo-
spheric drag on an artificial Earth satellite. The drag force acts tan-
gentially to the orbit. This force can be expressed as
1
F =− CD Aρv 2 ,
2m
where m = mass of the satellite, CD = the drag coefficient, A = the
cross sectional area of the satellite, ρ = the air density, and v = the
velocity of the satellite. The density of air at altitude η is given ap-
proximately by
ρ = ρ0 exp [−(η − η0 )/H]
−12
where ρ0 = 6.5 × 10 kg/m3 , η0 = 384 km and H is a “scale height”
for which we can use 40 km at altitudes above about 100 km.
It is not difficult to show that the effect of drag on the semi-major
axis and the eccentricity of the orbit are given by the expressions
Z 2π
A 2 (1 + e cos E)3/2
∆a = − CD a ρ(η) dE
m 0 (1 − e cos E)1/2
Z 2π
A 2 (1 + e cos E)1/2
∆e = − CD a(1 − e ) ρ(η) cos EdE
m 0 (1 − e cos E)1/2
where E is a parameter called the eccentric anomaly (see Computa-
tional Project 10.6). Evaluate ∆a and ∆e by numerically integrat-
ing the given expressions. You may assume CD = 1.05, A = 6 m2 ,
and m = 1000 kg. Note: The distance from the center of Earth to
the satellite can be expressed in terms of the eccentric anomaly by
r = a(1 − e cos E).
Computational Project 10.6. In Exercise (10.12) you evalu-
ated the average potential energy of a planet in orbit to be −GM m/a
by simply averaging the maximum and minimum values of r. You are
now asked to show that the time average of potential energy is given by
that expression. You should determine the potential energy at equal
time steps as the planet goes around in orbit. For simplicity let the
planet be at 1 AU, have a period of one year and an eccentricity e = 0.2.
The problem is complicated by the fact that the planet does not orbit
372 10. CENTRAL FORCE MOTION: THE KEPLER PROBLEM

at a constant angular speed. However, you can get the position of the
planet at equal intervals of time by determining the eccentric anom-
aly E which is related to the average angular velocity n by Kepler’s
equation10
E − e sin E = nt.
This is a transcendental equation so you will have to solve for E nu-
merically. Write a program that obtains E iteratively, using the fact
that e is a small quantity. That is, write Kepler’s equation as
E = nt + e sin E.
Since e is small, the first approximation to E is E = nt. Use this for E
on the right hand side and get the second approximation. Repeat over
and over. Thus:
E0 = nt
E1 = nt + e sin E0
E2 = nt + e sin E1
etc.
To solve for the average potential energy, let the planet orbit the Sun
10 times, determine the time average potential energy and compare it
with −GM m/a.

10The eccentric anomaly is a parameter that is well known by celestial me-

chanicians and astronomers. Kepler’s equation is a famous equation whose ana-
lytical solution has not been found. You can read about it in Chapter 2 of C. D.
Murray and S. F. Dermott, Solar System Dynamics, Cambridge University Press,
Cambridge, 1999.
Chapter 11
Harmonic Motion

This chapter is an in-depth study of harmonic motion. You are

familiar with two examples of simple harmonic oscillators, namely, a
pendulum and a mass on a spring. Here you will study several different
kinds of harmonic motion, including underdamped, overdamped, and
critically damped oscillatory motion, as well as forced harmonic motion
and the motion of coupled harmonic oscillators. The techniques you
will learn here are used in nearly every branch of physics.

11.1. Springs and Pendulums

A very simple oscillatory system consists of a mass m connected to
a spring of constant k. To avoid any complications due to forces such
as gravity or friction, assume the mass is on a perfectly frictionless
horizontal surface and can slide back and forth, as shown in Figure
11.1.

x
k
m

Figure 11.1. A mass on a frictionless surface connected

to a spring. The distance x is measured from the equi-
librium position.
373
374 11. HARMONIC MOTION

The kinetic energy of the mass is T = 12 mẋ2 and the potential

energy of the spring is V = 12 kx2 . Therefore, as described in Section
4.3 the Lagrangian for the system is
1 1
L = T − V = mẋ2 − kx2 .
2 2
The equation of motion is obtained from Lagrange’s equation:
d ∂L ∂L
− = 0,
dt ∂ ẋ ∂x
which yields
k
ẍ + x = 0. (11.1)
m
Another simple oscillatory system is a pendulum consisting of a
particle of mass m (the bob) on an inextensible, massless string of
length l. See Figure 11.2. The kinetic energy is T = 21 m(ẋ2 + ẏ 2 ) where
x = l sin θ and y = −l cos θ (assuming the origin is located at the point
where the string is attached to the support).
1 1
T = m(l2 θ̇2 cos2 θ + l2 θ̇2 sin2 θ) = ml2 θ̇2 .
2 2
The potential energy is
V = −mgl cos θ.
Consequently, (see Equation 4.4),
1
L = T − V = ml2 θ̇2 + mgl cos θ.
2
The Lagrange equation of motion is:
d ∂L ∂L
− = 0.
dt ∂ θ̇ ∂θ
That is,
d
(ml2 θ̇) + mgl sin θ = 0,
dt
or
ml2 θ̈ + mgl sin θ = 0. (11.2)
◦
By restricting the motion to small angles (θ . 20 ) you can replace
sinθ by θ (in radians). Then, dividing by ml2 you get
g
θ̈ + θ = 0. (11.3)
l
Note that Equations (11.1) and (11.3) have exactly the same form;
they both tell us that the second derivative of the position variable is
negatively proportional to the displacement from equilibrium, meaning
the system is subjected to a restoring force that is always directed
 11.1. SPRINGS AND PENDULUMS 375

l
θ

Figure 11.2. A simple pendulum.

toward the equilibrium point. This form for the equation of motion is
characteristic of oscillatory systems. Whenever you see it, you can be
sure that you are faced with a case of simple harmonic motion. You can
easily show by substitution that possible solutions to the two simple
harmonic motion equations above are:
r !
k
x(t) = A cos t+β , (11.4)
m
and
r 
g
θ(t) = A cos t+β . (11.5)
l
These solutions describe
p a sinusoidal oscillation of amplitude Apand
angular frequency k/m for the spring and angular frequency g/l
for the pendulum. The quantity β is related to the initial displacement.
Consider now the somewhat more complicated case of an oscillating
system subjected to a retarding force such as air resistance or some sort
of friction. For the sake of a specific example, let the system be a mass
on a spring acted upon by a resistive force that is proportional to the
velocity. Then the total force acting on the mass is not just −kx, but
rather,
F = −kx − bẋ,
where b is a constant of proportionality. The minus sign guarantees
that the retarding force opposes the motion. The −bẋ term is called a
damping force. The equation of motion will now be
mẍ + bẋ + kx = 0. (11.6)
This is the equation of motion for a damped harmonic oscillator (as-
suming the damping force is proportional to the first power of the
velocity).
376 11. HARMONIC MOTION

Finally, a harmonic oscillator may be subjected to some external

force, Fe . Then, we simply add the external force to the other forces
(but put it on the other side of the equation) and write
mẍ + bẋ + kx = Fe . (11.7)
This is the equation for a forced, damped, harmonic oscillator.
Note that these equations of motion are all linear, ordinary, second-
order differential equations with constant coefficients. You learned how
to solve such equations in your course on differential equations. I will
give a short review of the technique, but if you are uncertain as to any
of the details please review your differential equations textbook.

Exercise 11.1. (a) Prove by substitution that Equation (11.4) and

Equation (11.5) are solutions to their respective equations of motion.
(b) Prove by substitution that
p p
x = A cos k/mt + B sin k/mt
is a solution to Equation(11.1). We shall show that this is one way to
express the most general solution to (11.1).
Exercise 11.2. A certain spring stretches 6 cm when a force of
6 N is applied to it. Assume a 5 kg mass is attached to the spring,
then it is stretched 3 cm and released with a push so that its initial
speed is -6 cm/s. Assume the system lies on a frictionless horizontal
surface. Find an equation for the position of the mass as a function
of time. (Hint: Use the general form given in Exercise 11.1) Answer:
x = 0.030 cos(4.47t) − 0.013 sin(4.47t).
Exercise 11.3. At time t = 0 a simple harmonic oscillator with
angular frequency ω is at x = 3 cm and has a speed of 1 cm/sec. Obtain
an expression for x = x(t). Answer: x = 3 cos ωt + (1/ω) sin ωt.

11.2. Solving the Differential Equation (Optional)

An nth order linear differential equation with constant coefficients
has the form
dn x dn−1 x d2 x dx
an n + an−1 n−1 + · · · + a2 2 + a1 + ao x = f (t).
dt dt dt dt
If the quantity f (t) on the right hand side is zero, the equation is
called homogeneous, otherwise, it is inhomogeneous. By “linear” we
 11.2. SOLVING THE DIFFERENTIAL EQUATION (OPTIONAL) 377

mean that the dependent variable (x) and its derivatives appear only
to the first power. However, the independent variable (t) can be raised
dx 2
to any power. Thus dx dx
2 2


dt
= 3t is linear, but dt
= 3x and dt
= 3t
are not linear.1
11.2.1. Homogeneous Linear Differential Equations. A ho-
mogeneous differential equation of the form given above has a solution

x = ept .
The technique for obtaining the general solution is to plug x = ept into
the equation and carry out the obvious mathematical operations. This
yields an equation for p in terms of the constants. I will illustrate this
procedure in a moment. But first I want to mention that a first-order
differential equation yields a single value for p, a second-order differen-
tial equation yields two values for p and so on. Thus, depending on the
order of the differential equation, you get several different solutions.
Which one of these various solutions should you use? The answer is:
all of them!
Rule 1: If x1 (t) and x2 (t) are both solutions of a linear
homogeneous differential equation then x1 (t) + x2 (t) is also a
solution.
For simple harmonic motion we have a second order differential
equation so we will obtain two values for p. Let us call these two values
p1 and p2 . The two solutions are:
x1 (t) = ep1 t and x2 (t) = ep2 t .
By Rule 1, the sum of these two solutions is also a solution. That is,
x(t) = ep1 t + ep2 t
is a solution.
We can generalize even further using another simple fact.
Rule 2: If x(t) is a solution of a linear homogeneous differ-
ential equation, then Cx(t), where C is an arbitrary constant,
is also a solution.
Consequently,
x = C1 ep1 t + C2 ep2 t
is a solution to a second-order differential equation. In fact, it is the
most general form for the solution to a homogeneous linear second-order
1The “degree” of a differential equation is the power of its highest order deriv-
ative. Thus a linear differential equation is of the first degree. The “order” of a
differential equation is the order of the highest derivative appearing in the equation.
378 11. HARMONIC MOTION

differential equation with constant coefficients. Any other solution can

be expressed as a linear combination of ep1 t and ep2 t .
Similarly, the general solution for a third-order differential equation
is
x = C1 ep1 t + C2 ep2 t + C3 ep3 t ,
and so on.
Note that the values of p (p1 , p2 , · · ·) come from the parameters
of the problem (such as the masses and spring constants) whereas the
values of the constants (C1 , C2 , · · ·) come from the initial conditions.
Let me apply these concepts to my favorite differential equation,
..
mx + kx = 0.
If x = ept then ẍ = p2 ept . Plug in to obtain
mp2 ept + kept = 0.
Dividing by ept we obtain an equation for p,
mp2 + k = 0.
This is called the auxiliary equation. Solving for p gives
r r
k k
p = ± − = ±i = ±iω
m m
√ p
where i ≡ −1 and ω = k/m.
Thus we have two (linearly independent) solutions, namely,
x = e+iωt and x = e−iωt .
Since the sum of solutions is also a solution and since we can multiply
each solution by an arbitrary constant, the general solution to the
second order differential equation is
x = C1 e+iωt + C2 e−iωt . (11.8)
This equation is in a perfectly acceptable form. However, it can also
be expressed in terms of sines and cosines, which may be preferable in
some cases.
Using “Euler’s Formula”
e±iθ = cos θ ± i sin θ,
Equation (11.8) can be written
x = C1 (cos ωt + i sin ωt) + C2 (cos ωt − i sin ωt)
or
x = (C1 + C2 ) cos ωt + i(C1 − C2 ) sin ωt.
 11.2. SOLVING THE DIFFERENTIAL EQUATION (OPTIONAL) 379

You might think that since x is a real quantity, you should discard the
last term because it is imaginary; however, this would be unjustified
because C1 and C2 are not yet determined and they could very well be
complex numbers.
You now know basically everything you need for solving simple
harmonic motion problems, except for one small detail: If the auxiliary
equation for a second order differential equation has two equal roots,
then p1 = p2 = p and the general solution is
x = C1 ept + C2 tept . (11.9)
This is easily verified by substitution. (Note the factor of t in the
second term.)
Worked Example 11.1. Solve
d2 x
− 4x = 0,
dt2 
given the initial conditions x|t=0 = 0, dx  = 3.
dt t=0
Solution: Note that this is not the SHM equation because
the coefficient of x is negative. To solve, substitute ept into the
equation and obtain
p2 − 4 = 0.
Therefore, p = ±2 and the solution is
x = C1 e2t + C2 e−2t .
The initial condition x(t = 0) = 0 leads immediately to C2 = −C1
so
x = C1 (e2t − e−2t ) = 2C1 sinh(2t).
The initial condition ẋ = 3 leads to
ẋ = 2C1 (2) cosh(2t) = 4C1 cosh(0) = 3.
∴ C1 = 3/4.
and
3
x= sinh(2t).
2

Exercise 11.4. (a) Prove Rule 1 for a linear homogeneous second

order differential equation. (b) Prove Rule 2 for a linear homogeneous
second order differential equation.
Exercise 11.5. Verify that the relation y = Aeax cos bx+Beax sin bx
d
satisfies the equation [(D − a)2 + b2 ]y = 0, where D ≡ dx .
380 11. HARMONIC MOTION

Exercise 11.6. Solve the equation

d3 x d2 x dx
− 4 + + 6x = 0.
dt3 dt2 dt
Answer: x = C1 e−t + C2 e2t + C3 e3t .
d
Exercise 11.7. Solve (D2 +2D-3)y=0 where D≡ dx
.
Answer: y = C1 ex + C2 e−3x .
d
Exercise 11.8. Solve (D2 − 6D + 9)y = 0, where D ≡ dt
.
Exercise 11.9. Obtain formulas for sin2θ

2 and cos2θ in terms of
sinθ and cosθ. [Hint: Note that e2iθ = eiθ .]

11.2.2. Example: A Mass on a Spring. Consider once again a

block of mass m attached to a spring of constant k. Suppose you pull
the mass until the spring is stretched to x0 . Now release it from rest.
You are asked to determine the motion, x = x(t) and to evaluate the
amplitude and period of the oscillations.
As given by Equation (11.8), the solution has the form
x(t) = C1 eiωt + C2 e−iωt ,
p
with ω = k/m. The only thing you need to do now is determine the
constants C1 and C2 . The initial conditions are
x(t = 0) = x0 ,
and
ẋ(t = 0) = 0.
Plugging these into the solution you easily obtain

x(t = 0) = x0 = C1 + C2 ,
and
ẋ(t = 0) = 0 = iωC1 − iωC2 .
The second equation requires that C2 = C1 . Then the first equation
implies that C1 = C2 = 12 x0 . Therefore, the solution is
1 1
x = x0 (e+iωt + e−iωt ) = x0 (2 cos ωt) = x0 cos ωt.
2 2
The last equation tells us that the mass oscillates back and forth be-
tween the values +x0 and −x0 . The quantity x0 is called the amplitude.
The motion takes place with a frequency
r
ω 1 k
f= = .
2π 2π m
 11.2. SOLVING THE DIFFERENTIAL EQUATION (OPTIONAL) 381

A plot of position as a function of time is presented in Figure 11.3.

Note that the mass returns to its original position after a time t given
by
2π 1
t= = = P = period.
ω f
The period (P ) is the time for one complete oscillation. The frequency
(f ) is the number of oscillations per unit time and is equal to the inverse
of the period. The quantity ω = 2πf is called the “angular frequency”
but you should be aware that often it is carelessly referred to as the
frequency.

x
x0
P/2
t
P

-x0
P=2π
Figure 11.3. A plot of position as a function of time
for a mass undergoing simple harmonic motion.

If the initial conditions are different, the form of the solution is

slightly different. For example, suppose that you had been told that at
t = 0 the block was at x = 0 and had a velocity v0 . That is, x(t = 0) = 0
.
and x(t = 0) = v0 . Then, using Equation (11.8) again you obtain
0 = C1 + C2 ,
and
v0 = iω(C1 − C2 ).
Therefore,
C1 = −C2 and C1 = v0 /2iω.
Consequently,
v0 v0 v0
eiωt − e−iωt =


x= (2i sin ωt) = sin ωt. (11.10)
2iω 2iω ω
The solution, x = x(t), also gives information on the energy of the
system. When x = 0, ẋ = v0 , and the potential energy is zero whereas
the kinetic energy is a maximum. Also, when x = x0 the potential
382 11. HARMONIC MOTION

energy is a maximum but the kinetic energy is zero. Since the total
energy is constant,2
1 2 1 2
mv = kx .
2 0 2 0
You can use this equation to relate x0 and v0 , thus:
r
m v0
x0 = ± v0 = ± .
k ω
The result (11.10) can be written in the more familiar form
x = x0 sin ωt.
The two examples you have just considered had two very simple
sets of initial conditions. In both cases C1 and C2 were replaced by
a single constant. In general, however, the solution will contain two
different constants. The general solution can also be expressed in the
form
x = x0 sin(ωt + β).
Worked Example 11.2. A mass of 0.25 kg is attached to a
spring of force constant 1.0 N/m. The mass is displaced 0.15 m
from its equilibrium point and released with zero initial velocity.
Evaluate the total energy of the oscillator (kinetic plus potential).
What is the maximum velocity? What is the period of the oscilla-
tion?
Solution: p p
ω = k/m = 1/.25 = 2,

x = 0.15 cos 2t,

ẋ = v = −0.30 sin 2t.
1 2 1 2 1 1
E = mv + kx = (0.25)(−0.30 sin 2t)2 + (1)(0.15 cos 2t)2
2 2 2 2
2 2
= 0.0113 sin 2t + 0.0113 cos 2t = 0.0113 (joules).
Maximum velocity is ẋmax = 0.30 m/s and period of oscillation
is P = 2π/ω = π seconds.

2Note that this result is in agreement with results from the virial theorem
discussed in Worked Example 10.2. In the present problem, n = 1 so
< T >= n+1 2 < V > yields < T >=< V >
 11.3. THE DAMPED HARMONIC OSCILLATOR 383

Exercise 11.10. Show that x = x0 sin(ωt + β) is equivalent to

x = C1 cos ωt + C2 sin ωt.
Exercise 11.11. A spring of constant 10 N/m is connected to a
mass of 2 kg. It is initially at rest in an unstretched position. It is given
an initial speed of 3 m/s. What is the amplitude of the oscillation?
Answer: 1.34 m.

11.3. The Damped Harmonic Oscillator

Consider now an oscillator with a retarding force proportional to
the velocity. This is a damped harmonic oscillator. See Figure 11.4.
Common sense tells us that such an oscillator will lose energy and
eventually stop.
x

f.s = -kx
k x
m
.
ff = -bx

Figure 11.4. A damped harmonic oscillator. At the

moment illustrated, the block is moving to the left, to-
ward its equilibrium position. The spring force (fs ) is
directed toward the left and the frictional force (ff ) is
directed toward the right, opposite to the velocity.

The equation of motion of a damped harmonic oscillator is given

by Equation (11.6) which I repeat here:
mẍ + bẋ + kx = 0.
This is a second order linear ordinary homogeneous differential equation
with constant coefficients. As previously described, to solve it you make
the substitution x = ept ,
mp2 ept + bpept + kept = 0.
Dividing by ept gives the auxiliary equation:
b k
p2 + p + = 0.
m m
384 11. HARMONIC MOTION

This is a quadratic equation with two solutions:

s 2
b b k
p=− ± − . (11.11)
2m 2m m

Let us call the two solutions p1 and p2 . Since ep1 t and ep2 t are both
solutions to the differential equation, the general solution is the sum of
these;
x = C1 ep1 t + C2 ep2 t , (11.12)
where C1 and C2 are constants to be determined from the initial condi-
tions.
If p1 and p2 are complex, having an imaginary component such as
iω, then the solution contains terms of the form
eiωt = cos ωt + i sin ωt.
From your knowledge of the behavior of sines and cosines, you know
that such a solution represents oscillatory motion. On the other hand,
if p1 and p2 are real, then the solution is made up of terms having the
form
e−at and e+at ,
where a is a real quantity. Such a solution will either represent an
exponential decrease or an exponential increase (depending on whether
the sign is negative or positive). From physical considerations we can
rule out exponentially increasing solutions for the problem of a mass
on a spring.
From the equation for p (Equation 11.11) you note that p is complex
if and only if the quantity under the square root is negative, that is, iff
 2
b k
< .
2m m
In this case the system oscillates. On the other hand, if
 2
b k
≥ ,
2m m
then p is real and the motion dies out exponentially.
The relation of k/m to (b/2m)2 characterizes the three types of
motion for the damped oscillator. These are called “underdamped,”
“overdamped,” and “critically damped.” They are described in the fol-
lowing table:
 11.3. THE DAMPED HARMONIC OSCILLATOR 385

Name Condition Description

underdamped (b/2m)2 < k/m oscillatory motion
critically damped (b/2m)2 = k/m exponential decrease
overdamped (b/2m)2 > k/m exponential decrease
The motion of the damped oscillator depends on the relative values
of the parameters b, k, m. If b/m is large, the system is subjected to
a considerable retarding force and the oscillations will die out quickly.
On the other hand, if b/m is small, there is little damping and the
oscillator will remain in motion for a long time. Furthermore, if k/m
is large, the frequency of oscillation will be large, but if k/m is small,
the system will oscillate slowly.
The Underdamped Oscillator
Let us begin by considering the underdamped oscillator. Then,
(b/2m)2 < k/m and Equation (11.11) shows that p will be a complex p
quantity. It is convenient to change notation at this point. Since k/m
is the “natural” frequency of the undamped oscillator, we shall denote
k/m by ω02 . Furthermore, b/2m can be denoted by γ the “damping
coefficient.” The new notation is :
k
= ω02 = frequency of undamped oscillator,
m
b
= γ = damping coefficient.
2m
Using this notation, Equation (11.11) becomes
q
p = −γ ± γ 2 − ω02 . (11.13)

An underdamped oscillator has γ 2 < ω02 so the last term is the square
root of a negative number, and p is complex. In that case it is conve-
nient to write p in the form
p = −γ ± iω1
p
where ω1 = + ω02 − γ 2 . Note that ω12 = ω02 − γ 2 is positive, so ω1 is a
real number.
Inserting the two values of p into the solution (11.12) gives
x(t) = C1 e(−γ+iω1 )t + C2 e(−γ−iω1 )t
= e−γt (C1 eiω1 t + C2 e−iω1 t ).
C1 and C2 are complex constants related to the initial conditions. These
constants can be written in a variety of different ways. You obtain a
386 11. HARMONIC MOTION

very nice looking expression if you write them in the form

1 1
C1 = Aeiθ and C2 = Ae−iθ . (11.14)
2 2
This basically replaces the two complex constants C1 and C2 by the
real constants A and θ. Then
1
x(t) = e−γt A ei(ω1 t+θ) + e−i(ω1 t+θ) ,


2
or
x(t) = Ae−γt cos(ω1 t + θ). (11.15)
Plotting x vs t generates an oscillatory solution (due to the cosine
term), but the amplitude of the oscillations is Ae−γt , so the ampli-
tude decreases exponentially with time. This behavior is illustrated in
Figure 11.5.
x
+A
Ae-ϒt

-A
Figure 11.5. The motion of an underdamped oscillator
consists of rapid oscillations [described by cos(ω1 t + θ)]
modulated by an exponentially decreasing amplitude
(described by Ae−γt ).

Note that the oscillations have an angular frequency ω1 , which is

smaller than the natural frequency ω0 . That is, the damped system
oscillates more slowly than the undamped system. You probably ex-
pected this type of behavior.

Worked Example 11.3. An underdamped harmonic oscilla-

tor has k = 2 N/m, m = 1 kg, and b = 0.1 kg/s. How many
oscillations does the system make before the amplitude decreases
to 1/e of its initial value?
 11.3. THE DAMPED HARMONIC OSCILLATOR 387

Solution: The time when the amplitude is 1/e of its initial

value is obtained from
1 2m
Ae−γt = A ∴ t = 1/γ = = 20 sec.
e b
Then,
q 
p 
ω1 t = ω02
− γ2
t= 2
k/m − γ t
√ 
= 2 − 0.052 (20) = 28.26rad,
and the number of oscillations is 28.26/2π = 4.5.

Exercise 11.12. The complex constants C1 and C2 can be written

in the form C1 = a1 + ib1 and C2 = a2 + ib2 . But there can only be
two independent constants, so a1 , b1 and a2 , b2 are not all independent.
Using Equation (11.14) show that a2 = a1 and b2 = −b1 .
Exercise 11.13. Obtain the general solution of the following dif-
ferential equations. (a) mẍ + bẋ − kx = 0, (b) mẍ − bẋ + kx = 0.
Assume m, b, and k are positive real quantities.

The Overdamped Oscillator

Now consider the overdamped oscillator, defined to be an oscillator
for which  2
b k
> .
2m m
According to Equation (11.11) or (11.13), p will now have two real
values denoted γ1 and γ2 . Using the notation of Equation (11.13), the
solutions of the auxiliary equation are
q
p1 = −γ1 = −γ − γ 2 − ω02 ,
and q
p2 = −γ2 = −γ + γ 2 − ω02 .
Note that both γ1 and γ2 are positive and that γ1 is greater than γ2 .
Plugging these expressions into the general solution of the differen-
tial equation (11.12) leads to
x(t) = C1 e−γ1 t + C2 e−γ2 t , (11.16)
where γ1 and γ2 are real. Both terms are exponentially decreasing with
time, as indicated in Figure 11.6.
388 11. HARMONIC MOTION

x
C2e- 2t

x(t)
t
C1e- 1t

Figure 11.6. An overdamped harmonic oscillator. The

motion is the sum of the two curves. The constants C1
and C2 depend on the initial conditions.

The motion of the oscillator (that is, the value of x as a function

of time) is, of course, the sum of two curves, as illustrated in Figure
11.6. The constants C1 and C2 depend on the initial conditions. For
example, if you pull the mass to some position x0 and release it from
rest (v0 = 0), at t = 0, Equation (11.16) reduces to
x0 = C 1 + C 2 ⇒ C 1 = x0 − C 2 .
Since ẋ = −γ1 C1 e−γ1 t − γ2 C2 e−γ2 t ,
v0 = −γ1 C1 − γ2 C2 .
Now v0 = 0, so C2 = −(γ1 C1 /γ2 ), and hence,
C1 = x0 − C2 = x0 + (γ1 /γ2 )C1 .
Therefore,
γ2 x0
C1 = − ,
γ1 − γ2
and
γ1 xo
C2 = .
γ1 − γ2
Consequently,
γ2 x0 −γ1 t γ1 x0 −γ2 t
x(t) = − e − e .
γ1 − γ2 γ1 − γ2
Depending on the initial conditions the oscillator will behave in one
of three ways. The displacement x(t) may rapidly decay to zero (as
illustrated in Figure 11.6). Or x(t) may increase for a time and then
decay to zero. Finally, x(t) could change sign, reach a local maximum
or minimum, then approach zero.
 11.4. THE FORCED HARMONIC OSCILLATOR 389

Exercise 11.14. Consider the motion of an overdamped oscillator.

Assume that initially x(t = 0) = 1 and ẋ(t = 0) = 0. Let γ1 = 3.414
and γ2 = 0.586. Plot x(t).
Exercise 11.15. Obtain an expression for x(t) for an overdamped
oscillator if the initial conditions are x(t = 0) = 0 and ẋ(t = 0) = v0 .

The Critically Damped Oscillator

Finally, consider the case of a critically damped harmonic oscillator
for which  2
b k
= .
2m m
Now the term under the root in Equation (11.11) is zero.
For convenience, define the quantity γc as follows:
r
b k
γc = = = ω0 .
2m m
Then Equation (11.13) yields a single value for p namely p = −γ.
Recall that if p has only one value, the solution to the differential
equation has the form of Equation (11.9), that is
x(t) = C1 e−γc t + C2 te−γc t ,
or
x(t) = (C1 + C2 t) e−γc t .
For critical damping x decays quickly to zero.

Exercise 11.16. A critically damped harmonic oscillator is initially

at its equilibrium position with velocity v0 . Obtain expressions for C1
and C2 . Answer: C1 = 0, C2 = v0 .
Exercise 11.17. A critically damped harmonic oscillator has b =
0.3 kg/s and k = 0.4 N/m. (You are not given the mass.) It is released
from rest at x = 0.04 m. Evaluate the constants C1 and C2 . Answer:
C1 = 0.04, C2 = 0.107.

11.4. The Forced Harmonic Oscillator

11.4.1. Statement of the Problem. A forced harmonic oscilla-
tor is an oscillator which is subjected to some sort of driving force. The
external driving force is often periodic with the same frequency as the
oscillator. When you push your little brother on a swing, you give the
390 11. HARMONIC MOTION

seat a shove each time it comes back to you at the top of its swing.
The periodicity of your shoves is the same as the natural period of
the swing. Similarly, the pendulum in a “grandfather” clock receives
a small push on each swing. An old-fashioned pocket watch has an
oscillating spring wheel that receives small periodic impulses from a
spring and ratchet system.
The oscillator will, in general, be damped, so the equation of motion
for such a system is

mẍ + bẋ + kx = F (t), (11.17)

where F (t) is the driving force. The purpose of the driving force is
to overcome the damping. If F (t) is too small, the oscillations will
gradually die out. On the other hand, if F (t) is too large, the amplitude
may grow unacceptably large. (If you push a swing too hard it will “go
over the top” leading to unpleasant consequences.) As you shall see
shortly, the amplitude of the oscillations also depends on the frequency
of the driving force. This is related to the phenomenon of resonance.
The equation of motion of the driven harmonic oscillator (Equation
11.17) is an inhomogeneous second order linear differential equation. I
will now make a short aside to describe how to solve such an equation.

11.4.2. Inhomogeneous Linear Ordinary Differential Equa-

tions. The title of this section describes the kind of equation we want
to solve. You should understand the implication of each of the many
longish words in the description. The equation is, of course, a differ-
ential equation because there are derivatives in it. It is linear because
the dependent variable (x) and its derivatives are not raised to a power
larger than the first. We are dealing with an ordinary differential equa-
tion because (11.17) has no partial derivatives in it.
You probably recall that differential equations are often expressed
2
in terms of the differential operator D ≡ dtd . Then D2 = dtd 2 and so on.
A linear homogeneous differential equation is an equation which can
be put in the form

an Dn x + an−1 Dn−1 x + · · · + a1 Dx + ao x = 0.

Although the ai ’s are usually constants, they can be functions of t.

Note the zero on the right hand side.
Of course, the equation need not have a zero on the right. There
could be a function of x and t on the right hand side. Thus,

an Dn x + an−1 Dn−1 x + · · · + a1 Dx + ao x = f (x, t).

 11.4. THE FORCED HARMONIC OSCILLATOR 391

Such an equation is said to be inhomogeneous. The solution of an inho-

mogeneous differential equation can be determined using the following
rule.
Rule 3: The solution of an inhomogeneous differential
equation is the sum of the general solution of the correspond-
ing homogeneous equation and a particular solution of the
inhomogeneous equation.
Therefore, the first step in solving an inhomogeneous differential
equation is to set the right hand side equal to zero. This generates the
corresponding homogeneous differential equation which we can solve by
the method of Section 11.2. The general solution will, of course, contain
as many arbitrary constants as the order of the equation. A second
order equation has a general solution with two arbitrary constants.
We denote this general solution of the homogeneous equation by xg .
It is usually called the “complementary function.” Next we consider
the inhomogeneous equation. We can often figure out a solution by
inspection. Any solution will do. We denote this particular solution
by xp . The general solution of the inhomogeneous equation is xp + xg .
It is easy to prove this assertion, and it is left as an exercise for you.
We now return to the problem of the driven harmonic oscillator.

Exercise 11.18. Show that the general solution of an inhomoge-

neous differential equation is the sum of a particular solution and the
general solution of the homogeneous differential equation.

11.4.3. Obtaining the Particular Solution. The equation of

motion for the driven, damped harmonic oscillator is

mẍ + bẋ + kx = F (t).

The forcing term, F (t), is usually periodic because that will keep the
oscillator in motion, overcoming the effect of damping. Another com-
mon situation is an oscillator that has been subjected to an impulsive
force. In general, the forcing term can be any function of time. It
could even be a constant. In fact, for the sake of making the analysis
simple, I will begin with a constant force. For example, a mass hanging
from a spring in an external gravitational field (see Figure 11.7) has
the equation of motion

mẍ + bẋ + kx = mg, (11.18)

392 11. HARMONIC MOTION

where the damping could be due to air resistance, for example. The
mass is only allowed to oscillate vertically; it does not simultaneously
swing like a pendulum.

Figure 11.7. A mass hanging from a spring in a con-

stant gravitational field.

The corresponding homogeneous equation was solved in the previ-

ous section. If the oscillation is underdamped, for example, the general
solution of the homogeneous equation is given by Equation (11.15) as:
xg = x(t) = Ae−γt cos(ω1 t + θ).
Now you need to find a particular solution to Equation (11.18). That is,
you need to find any function of x which satisfies (11.18). After a little
creative staring at that equation you will realize that if x is constant,
the first two terms are zero and you are left with kx = mg. Therefore,
x = mg/k is a possible solution to the equation. It is the particular
solution, xp . The general solution of the inhomogeneous equation is,
therefore,
mg
x(t) = xg + xp = Ae−γt cos(ω1 t + θ) + .
k
This tells you that the effect of the gravitational force is just to stretch
the spring, giving a different (lower) value for the equilibrium point.
I will now go on to a more interesting problem, namely a driving
force that varies sinusoidally. Let the driving frequency be denoted ωd ,
so F (t) = F0 sin ωd t. For simplicity, we shall begin with an undamped
oscillator. Then the equation of motion is

mẍ + kx = F0 sin ωd t,
or
F0
ẍ + ω02 x = sin ωd t. (11.19)
m
 11.4. THE FORCED HARMONIC OSCILLATOR 393

The general solution of the homogeneous equation is

xg (t) = A cos(ω0 t + θ0 ),
where A and θ0 depend on the initial conditions. For example, θ0 = 0
corresponds to the oscillator at position x = A at time t = 0. (For a
pendulum this means that at time zero the bob was at the top of the
swing, and for a mass on a spring, it means that at time zero the spring
was fully extended.)
Having a general solution to the homogeneous equation, you still
need a particular solution to the inhomogeneous equation. Previously,
we obtained a particular solution by inspection. Often this will be
sufficient. However, there are techniques available for determining a
particular solution.3 One technique that always works for a second
order differential equation is to integrate the equation twice. To apply
the method, first write the equation in the form
(a2 D2 + a1 D + a0 )x = f (t). (11.20)
This equation can be expressed as the product
(D − b1 )(D − b2 )x = f (t). (11.21)
Let u = (D − b2 )x. Then the differential equation is
(D − b1 )u = f (t).
You now have a first order linear differential equation that you can
solve for u(t). Next, going back to the definition of u, you solve (D −
b2 )x(t) = u(t) for x(t). This technique produces the general solution for
the inhomogeneous equation, that is, it generates the complementary
function as well as the particular solution. However, it may involve
a considerable amount of mathematics, so I want to show you two
other techniques which might be classified as “tricks” for finding the
particular solution. (When a mathematical trick becomes very familiar
we quit calling it a trick and start calling it a technique!)
Trick #1
The first trick (or technique) is used when the function on the right-
hand side is of the form f (t) ∝ eat . Then, writing the second order
differential equation in the form
(D − b1 )(D − b2 )x = Keat (11.22)

3See, for example, Mary L. Boas, Mathematical Methods in the Physical Sci-
ences, 3rd ed. John Wiley & Sons, Inc., Hoboken, NJ, 2006. pp 417 ff.
394 11. HARMONIC MOTION

the particular solution is

xp = Ceat if a 6= b1 and a 6= b2 , (11.23)
at
xp = Cte if a = b1 or a = b2 but b1 6= b2 ,
2 at
xp = Ct e if a = b1 = b2 ,
where C is determined by plugging xp back into the differential equa-
tion.
Trick #2
The second trick is based on the first trick and we use it when f (t)
(the right-hand side) is either a sine or a cosine. In that case, write the
equation in one of the two following forms:
 
K sin at
(D − b1 )(D − b2 )x = (11.24)
K cos at
noting that K sin at is the imaginary part of Keiat and K cos at is the
real part of Keiat . We then solve (D − b1 )(D − b2 )x = Keiat using
formulas (11.23) and take either the real or the imaginary part of the
answer. This method is based on the fact that
eiθ = cos θ + i sin θ.
Finally, we can generalize the first trick if the right-hand side is an
exponential, eat , times a polynomial of degree n, Pn (t). Then the par-
ticular solution of
(D − b1 )(D − b2 )x = eat Pn (t)
is  
 eat Qn (t) if a 6= b1 and a 6= b2 
xp = teat Qn (t) if a = b1 or a = b2 and b1 6= b2
 t2 eat Q (t) if a = b = b 
n 1 2
where Qn (t) is a polynomial of degree n :
Qn (t) = A + Bt + Ct2 + · · ·.
To determine A, B, C, · · · you plug xp into the differential equation
and equate the coefficients of the different powers of t.

Worked Example 11.4. Find the general solution for (D2 −

4)x = 2 − 8t when t = 0, x = 0, and ẋ = 5. Identify the particular
solution.
Solution: The differential equation can be written as
(D − 2)(D + 2)x = 2 − 8t.
 11.4. THE FORCED HARMONIC OSCILLATOR 395

Let u = (D + 2)x and write

(D − 2)u = 2 − 8t.
Recall that for a first order differential equation such as
(D + f (t))x = g(t),
R
the integrating factor is e f (t)dt . Multiplying by the integrating
factor e−2t we obtain
du
e−2t − 2ue−2t = 2e−2t − 8te−2t ,
dt
d
(ue−2t ) = 2e−2t − 8te−2t ,
Z dt Z Z
−2t
d(ue ) = 2e dt − 8te−2t dt,
−2t

and hence,
u = C1 e2t + 4t + 1.
But since u = (D + 2)x we have
dx
+ 2x = C1 e2t + 4t + 1.
dt
Now the integrating factor is e2t so we multiply through by this
factor to get
d
(xe2t ) = C1 e4t + 4te2t + e2t ,
dt
and consequently,
1 1
x(t) = C2 e−2t + C1 e2t + 2t − .
4 2
Inserting initial conditions leads to C1 = 4 and C2 = −1/2, and
finally
1 1
x(t) = e2t − e−2t + 2t − .
2 2

Exercise 11.19. Obtain the general solution to (D2 −D −6)x = 8,

where D ≡ dtd . Answer: x(t) = C1 e3t + C2 e−2t − 43 .
Exercise 11.20. Obtain the general solution to (D2 − 9)x = 5e−2t
assuming that x(t = 0) = 0 and ẋ(t = 0) = 1 (in the appropriate set
of units). Answer: x(t) = 13 e3t + 23 e−3t − e−2t .
Exercise 11.21. Obtain the general solution of (D2 − 4)x = sin t.
Answer: x(t) = C1 e2t + C2 e−2t − 15 sin t.
396 11. HARMONIC MOTION

11.4.4. The Forced Undamped Oscillator. Now that you have

seen some methods for solving inhomogeneous second order differential
equations, let us return to the problem of the forced harmonic oscillator.
For the sake of simplicity we begin with the undamped oscillator.
The differential equation we set out to solve was Equation (11.19):
F0
ẍ + ω02 x = sin ωd t.
m
The right-hand side of this equation is proportional to sin ωd t so we use
trick #2 and write
F0 iωd t
(D − iω0 )(D + iω0 )ξp (t) = e . (11.25)
m
According to trick #2, the particular solution is the imaginary part of
the solution of Equation (11.25), namely:

ξp (t) = Ceiωd t .
(I wrote ξp rather than xp because this is not yet the particular solu-
tion.) To obtain C, plug ξp back into the differential equation, thus:
 2 
d F0 iωd t
2
2
+ ω0 Ceiωd t = e .
dt m
Hence,
F0
−ωd2 C + ω02 C = ,
m
and
F0 /m
C= .
ω02 − ωd2
Finally, the particular solution is
 
F0 /m iωd t F0 /m
xp (t) = Im (ξp (t)) = Im 2 2
e = 2 sin ωd t,
ω0 − ωd ω0 − ωd2
and the general solution of Equation (11.19) is
F0 /m
x(t) = A cos(ω0 t + θ0 ) + sin ωd t.
ω02 − ωd2
The problem is now solved, except for determining θ0 from the
initial conditions. However, before leaving this problem, I want you
to notice that there is something quite interesting about the particular
solution,

F0 /m
xp = sin ωd t.
ω02 − ωd2
 11.4. THE FORCED HARMONIC OSCILLATOR 397

This expression indicates that as the driving frequency ωd gets closer

and closer to ω0 , the natural frequency of oscillation, the amplitude
of the oscillations get larger and larger. If ωd = ω0 , the amplitude
is infinite. This phenomenon is known as resonance. We will consider
resonance further in the next section, refining our analysis and showing
that the amplitude does not actually become infinite for a real physical
system.

Exercise 11.22. A simple harmonic oscillator consists of a mass of

3 kg on a spring of force constant 0.15 N/m. Determine the frequency
of the driving force which will cause the amplitude of oscillations to
grow extremely large. Answer 0.022 hz.

11.4.5. The Forced Damped Oscillator. Consider now a damped,

driven, harmonic oscillator and assume the driving force is, once again,
F0 sin ωd t. The equation of motion is
b k F0
ẍ + ẋ + x = sin ωd t.
m m m
p
Note that k/m = ω0 is the “natural” frequency of the undamped os-
cillator. (For the sake of a specific example, let us assume the oscillator
is underdamped.)
The technique for finding the particular solution is the same as
before, so set
ξp = Ceiωd t ,
and plug it into the differential equation. This gives
b F0 iωd t
C(−ωd2 )eiωd t + C(iωd )eiωd t + ω02 Ceiωd t = e ,
m m
where sin ωd t was replaced by eiωd t . (This means that at the end of the
procedure we must only keep the imaginary part of the solution.)
Solving for C yields
F0 /m
C= 2 ,
(ω0 − ωd2 ) + iωd b/m
which is complex. To make this quantity easier to manipulate, the
imaginary term is usually placed in the numerator. To achieve this,
multiply top and bottom by the complex conjugate of the denominator
and obtain
(F0 /m) [(ω02 − ωd2 ) − iωd b/m]
C= h i .
2
(ω02 − ωd2 ) + (ωd b/m)2
398 11. HARMONIC MOTION

You have already seen that the general solution of the homogeneous
equation for the underdamped harmonic oscillator is given in Equation
(11.15) as x = Ae−γt cos(ω1 t + θ). The general solution of the forced,
underdamped harmonic oscillator is, then,
 
2 2
F0 /m [(ω0 − ωd ) − iωd b/m] iωd t 
x(t) = Ae−γt cos(ω1 t + θ) + Im  h i e .
2
(ω02 − ωd2 ) + (ωd b/m)2

Eventually, the first term will die out. (This is, of course, also true
for overdamped and critically damped oscillators.) Dropping the first
term, you are left with
 
2 2
F0 /m [(ω0 − ωd ) − iωd b/m]
x(t) = Im  h i (cos ωd t + i sin ωd t) .
2 2 2 2
(ω0 − ωd ) + (ωd b/m)

After a bit of algebra, this expression can be written

F0 (ω02 − ωd2 ) sin ωd t − (ωd b/m) cos ωd t
x(t) = 2 . (11.26)
m (ω02 − ωd2 ) + (ωd b/m)2
At resonance, when ω0 = ωd , the denominator is minimized, but the
damping term prevents it from going to zero. Therefore, at resonance
you get a large amplitude, but it is not infinite. In real physical systems,
there is always a damping term preventing the system from having
infinite amplitude at resonance.
Equation (11.26) can be expressed in a more convenient way by
noting that
√
A sin ωt + B cos ωt = A2 + B 2 cos(ωt − φ),
where φ = tan−1 (A/B). So,
" #1/2
(F0 /m)2 [(ω02 − ωd2 )2 + ωd2 b2 /m2 ]
x(t) = 2 cos(ωd t − φ)
[(ω02 − ωd2 )2 + ωd2 b2 /m2 ]
F0
= 1/2
cos(ωd t − φ), (11.27)
[m2 (ω0 − ωd2 )2 + ωd2 b2 ]
2

where
ω02 − ωd2
 
−1
φ = tan − .
ωd b/m
It is interesting to note that the amplitude of x(t) is not maximized
at ωd = ω0 but rather at a nearby frequency called the resonant fre-
quency which you can determine by evaluating the derivative of the
 11.4. THE FORCED HARMONIC OSCILLATOR 399

denominator and setting it equal to zero, thus:

d  2 2 1/2
m (ω0 − ωd2 )2 + ωd2 b2 = 0.
dωd
This leads to the following condition for a minimum
b2
ωd2 = ω02 − . (11.28)
2m2
The greatest amplitude occurs when the driving frequency is equal to
the resonant frequency which we can denote by ω 0 .
r
b2
ω 0 = ω02 − .
2m2
This is neither the frequency of the undamped oscillator (ω0 ), nor the
frequency of the damped oscillator (ω1 ) .
When ωd = ω 0 the amplitude of the oscillation is given by
F0
A= 1/2
.
[m2 (ω02 − ω 02 )2 + ω 02 b2 ]
The resonance phenomenon is often associated with the transfer of
energy. The average energy of a harmonic oscillator is proportional
to the amplitude squared, so an important and useful parameter in
studying resonance is A2 , given by
F02
A2 = 2 2 .
m (ω0 − ω 02 )2 + ω 02 b2

Worked Example 11.5. Show that h 2the2 iphase angle φ in

ω −ω
Equation (11.27) is given by φ = tan−1 − ω0d b/md .
Solution: We have shown (Equation 11.26) that

F0 (ω02 − ωd2 ) sin ωd t − (ωd b/m) cos ωd t

x = x(t) = 2 .
m (ω02 − ωd2 ) + (ωd b/m)2
Writing this in the form A sin ωt + B cos ωt we have
F0 (ω02 − ωd2 )
A= ,
m (ω02 − ωd2 )2 + (ωd b/m)2
and
F0 −(ωd b/m)
B= .
m (ω02 − ωd2 )2 + (ωd b/m)2
400 11. HARMONIC MOTION

So φ = tan−1 (A/B) yields

(ω02 −ωd2 )
2
( ω0 −ωd +(ωd b/m)2
2 2
) (ω02 − ωd2 )
φ = tan−1 −(ωd b/m)
= tan−1
2
−(ωd b/m)
(ω02 −ωd2 ) +(ωd b/m)2
− (ωd2 − ω02 )
= tan−1 .
ωd b/m

Exercise 11.23. Fill in the missing steps in the derivation of Equa-

tion (11.28).
√
Exercise 11.24. Show that A sin ωt+B cos ωt = A2 + B 2 cos(ωt−
φ), where tan φ = A/B.
Exercise 11.25. A damped harmonic oscillator is driven by a force
−at
F0 e . Find the general solution. (Note: The particular solution can
be expected to have the same time dependence as the driving force.)

Resonance in Electrical Circuits (Optional)

Resonance is an interesting physical phenomenon of importance in
such diverse fields as the design of electrical circuits and the motion
of celestial bodies. For example, the RLC circuit you studied in your
introductory electricity course is illustrated in Figure 11.8.

R
C L

Figure 11.8. An RLC circuit.

The charge on the capacitor plates is q. If the plates are initially

charged, there will be a flow of charge (a current) through the inductor,
di
setting up a back emf or voltage difference given by L dt . This can also
d2 q
be written as L dt2 . The voltage drop across the resistor is iR and this
can be written as R dqdt
. Finally, the voltage drop across the capacitor is
given by Cq . The sum of voltage drops around the circuit is zero, so
d2 q dq 1
L 2
+R + q = 0.
dt dt C
But this is exactly the equation for a damped harmonic oscillator! If
you add a “driving force” by placing some source of emf in the circuit
 11.4. THE FORCED HARMONIC OSCILLATOR 401

(such as an oscillator or power supply or even a common AC generator)

you willp have a driven oscillator. The natural frequency of the circuit
is ω0 = 1/LC, as you can appreciate from the form of the equation if
the damping term is removed. If the driving frequency is near or equal
to the natural frequency, the amplitude of the current increases greatly.
This is the principle behind tuning a radio or television receiver. The
“driving” voltage is supplied by the antenna picking up electromagnetic
radiation. Suppose a station you want to tune to is broadcasting at a
frequency ωd . The receiver contains a variable capacitor whose capac-
itance you change by turning the knob on your receiver. Changing C
will, of course, change the value of ω0 and hence the resonant frequency
ω 0 . When ω 0 is equal to ωd , the current through the resistor reaches
a much higher value than for other frequencies. This is illustrated in
Figure 11.9.
As mentioned in Chapter 10, celestial bodies are affected by reso-
nances. For example, the orbital period and the rotational period of
the Moon are “locked” into a 1:1 resonance. In celestial mechanics, a
resonance is defined as a system in which the frequency of the “driv-
ing” force (usually the effect of a perturbing body) and the “natural”
frequency of the system are in the ratio of small numbers. Resonances
explain why there are gaps in the rings of Saturn and in the asteroid
belts between Mars and Jupiter, why certain satellites (including the
Moon) have rotation rates equal to their orbital rates, and many other
effects. (The resonance phenomenon in celestial mechanics is often
called a “commensurability.”)

ω
ω

Figure 11.9. A plot of the current (I) as a function of

the angular frequency ω, illustrating the resonance phe-
nomenon in an LRC circuit. When the frequency ω is
nearly equal to the resonant frequency ω 0 , the current in
the circuit increases greatly. If there were no resistance
in the circuit, the current would become infinite at ω =
ω0 .
402 11. HARMONIC MOTION

Exercise 11.26. An RLC series circuit consists of a 10 ohm resis-

tor, a 6 microfarad capacitor, and an 0.2 henry inductor. Determine
the resonant frequency for the circuit. (Answer: 913 rad/s.)

11.5. Coupled Oscillators

In the real physical world, oscillators are seldom isolated from their
environment and an oscillating body will set up oscillations in nearby
bodies. The very good example of coupled oscillators is the motion of
atoms in a simple solid. A vibrating atom will cause neighboring atoms
to start oscillating.
The analysis of coupled oscillators can be quite complicated. This
section is a study of the simplest of such systems, namely two un-
damped oscillators coupled by springs.

k3 k2
k1

m1 m2

X1 X2

Figure 11.10. Coupled oscillators. Note that the posi-

tions of the two masses are measured from their equilib-
rium points.

Figure 11.10 illustrates two masses attached to rigid walls by springs

of constants k1 , k2 , and k3 . Let x1 and x2 be the displacements of masses
m1 and m2 measured from their equilibrium positions. The equations
of motion for the masses are easily determined from the Lagrangian
or from Newton’s second law. (See Exercise 11.27.) We obtain the
following coupled differential equations:4
m1 ẍ1 = −k1 x1 − k3 (x1 − x2 ), (11.29)
m2 ẍ2 = −k2 x2 − k3 (x2 − x1 ).
4When applied to the atoms in a solid, we assume all the masses are equal and
all the “spring constants” are equal, yielding n equations of the form
mẍi = k(xi − xi−1 ) + k(xi+1 − xi ) i = 1, · · · , n
These relations reduce to Equations 11.29 for the situation pictured in Figure 11.10.
 11.5. COUPLED OSCILLATORS 403

For simplicity, assume that m1 = m2 = m and that k1 = k2 = k,

but allow k3 to be different. Then the equations of motion reduce to

mẍ1 + (k + k3 )x1 − k3 x2 = 0,
mẍ2 + (k + k3 )x2 − k3 x1 = 0.
Denote k + k3 by k 0 . Dividing by m yields:
ẍ1 + (k 0 /m)x1 − (k3 /m)x2 = 0,
ẍ2 + (k 0 /m)x2 − (k3 /m)x1 = 0.
Now replace k 0 /m by ω02 . The reason for this notation is that ω0 is the
frequency of oscillation of either one of the masses if the other mass is
held at rest. Furthermore, let us denote k3 /m by ∆ω 2 , for reasons that
will soon become clear.
Using this new notation the equations become
ẍ1 + ω02 x1 − ∆ω 2 x2 = 0, (11.30)
ẍ2 + ω02 x2 − ∆ω 2 x1 = 0.
This system of coupled linear differential equations can be solved by
assuming the solutions
x1 = Aeiωt ,
x2 = Beiωt .
We assume oscillatory solutions (that is why the exponent is imagi-
nary) with both of the variables x1 and x2 having the same angular
frequency ω. (Except for a slight change in the notation, this is the
same technique previously used in assuming solutions of the form ept .)
Plugging the expressions for x1 and x2 into the coupled equations of
motion yields
A(iω)2 eiωt + ω02 Aeiωt − ∆ω 2 Beiωt = 0,
B(iω)2 eiωt + ω02 Beiωt − ∆ω 2 Aeiωt = 0,
or
(−ω 2 + ω02 )A − ∆ω 2 B = 0, (11.31)
−∆ω 2 A + (−ω 2 − ω02 )B = 0.
These two equations are a system of homogeneous coupled linear alge-
braic equations in the unknowns A and B. The technique for solving
systems of coupled algebraic equations is to apply Cramer’s rule. I be-
lieve you are familiar with this rule, but I will review it briefly anyway.
404 11. HARMONIC MOTION

Consider the pair of equations

ax + by = e,
cx + dy = f.
To solve, first evaluate the determinant of the coefficients, thus
 
 a b 
D =  .
c d 
The first column of this determinant is formed from the coefficients of
x and the second column contains the coefficients of y. Then, according
to Cramer’s rule, the solutions are:
   
 e b   a e 
   
 f d   c f 
x= and y = .
D D
Note that the determinant in the numerator of the equation for x is ob-
tained from D by replacing the column of the coefficients of x with the
constants on the right-hand side of the equations. Similarly y is given
by the same equation but with the column of coefficients of y replaced
by the right-hand constants. (You may recall using Cramer’s rule to
determine the currents in an electric circuit when applying Kirchhoff’s
laws.)
Unfortunately, when Cramer’s rule is applied to homogenous linear
equations such as (11.31), the zeros on the right-hand side mean that
the determinants in the numerators have a column of zeros. This leads
to the trivial solutions x = 0 and y = 0. However, if the denominator
is also zero (if D = 0), then x = 00 and y = 00 , both of which are
undefined. You may think this is not much of an advantage, but it
actually allows you to find the solution! In fact it is a general rule that
coupled homogeneous algebraic equations have nontrivial solutions if
and only if the determinant of the coefficients is zero.
Going back to Equations (11.31), the condition for a nontrivial
solution is that the determinant of the coefficients is equal to zero:
 
 −ω 2 + ω02 −∆ω 2 
D=  = 0,
−∆ω 2 −ω 2 + ω02 
or
(−ω 2 + ω02 )2 − (∆ω 2 )2 = 0,

ω 2 = +ω02 ∓ ∆ω 2 .
That is,
1
ω = ±(ω02 ∓ ∆ω 2 ) 2 .
 11.5. COUPLED OSCILLATORS 405

This yields four possible values for ω. Two of them can be written
1
ω1 = ±(ω02 − ∆ω 2 ) 2 ,
and the other two are
1
ω2 = ±(ω02 + ∆ω 2 ) 2 . (11.32)
The general solutions for x1 and x2 are the sum of all possible solutions:
x1 = A1 eiω1 t + A−1 e−iω1 t + A2 eiω2 t + A−2 e−iω2 t , (11.33)
x2 = B1 eiω1 t + B−1 e−iω1 t + B2 eiω2 t + B−2 e−iω2 t .
These two expressions involve eight constants. But solving two second
order differential equations should only lead to four constants. There-
fore, not all of the eight constants are independent. We now determine
the four independent constant amplitudes.
Note that each term in our expression for x1 satisfies the differential
equation. Thus, for example, a possible solution involves oscillations
of frequency ω1 . Then,
x1 = A1 eiω1 t
x2 = B1 eiω1 t
where ω12 = ω02 −∆ω 2 . Plugging into the equations of motion (Equation
11.30) yields
(ω02 − ω12 )A1 − ∆ω 2 B1 = 0
and
−∆ω 2 A1 + (ω02 − ω12 )B1 = 0
Either one of these yields the same result, namely,
−ω02 − (ω02 − ∆ω 2 ) A1 − ∆ω 2 B1 = 0,

or
+∆ω 2 A1 − ∆ω 2 B1 = 0.
Therefore,
B1 = A1 .
Similary by considering the solution in −ω1 we obtain B−1 = A−1 , and
±ω2 lead to B2 = −A2 and B−2 = −A−2 .
So our equations reduce to
x1 = A1 eiω1 t + A−1 e−iω1 t + A2 eiω2 t + A−2 e−iω2 t ,
x2 = A1 eiω1 t + A−1 e−iω1 t − A2 eiω2 t − A−2 e−iω2 t .
These solutions are not in a convenient format because x1 and x2 must
be real, but e±iω1 t and e±iω2 t are complex, so the coefficients Ai must
406 11. HARMONIC MOTION

also be complex. These complex coefficients can be expressed in terms

of real quantities, thus:
1
A1 = C1 eiθ1 ,
2
1
A−1 = C1 e−iθ1 ,
2
1
A2 = C2 eiθ2 ,
2
1
A−2 = C2 e−iθ2 .
2
Here C1 , C2 , θ1 and θ2 are all real. Then,
1
1
x1 = C1 eiθ1 e+iω1 t + e−iθ1 e−iω1 t + C2 eiθ2 e+iω2 t + e−iθ2 e−iω2 t ,


2 2
or
x1 = C1 cos(ω1 t + θ1 ) + C2 cos(ω2 t + θ2 ), (11.34)
and similarly,
x2 = C1 cos(ω1 t + θ1 ) − C2 cos(ω2 t + θ2 ). (11.35)

For the more general case in which k1 6= k2 and m1 6= m2 the

expressions are more complicated and lead to
∆ω 2 m2
 r 
x1 = C1 cos(ω1 t + θ1 ) + C2 cos(ω2 t + θ2 ),
κ2 m1
∆ω 2 m1
 r 
x2 = −C1 cos(ω1 t + θ1 ) + C2 cos(ω2 t + θ2 ),
κ2 m2
√
where κ2 = k3 / m1 m2 .

Exercise 11.27. (a) Derive the equations of motion (11.29) using

the Lagrangian technique. (b) Derive these equations using Newton’s
second law.
Exercise 11.28. Use Cramer’s rule to evaluate x,y,z if
2x + 3y + z = 1,
−5x − 2y + z = 5,
x + 2y + 2z = 13.
Answer: x = 3, y = −5, and z = 10.
 11.5. COUPLED OSCILLATORS 407

11.5.1. Normal Modes. Suppose that all of the masses in a sys-

tem of coupled oscillators are oscillating at the same frequency. This
type of motion is called a “normal mode.” You might suppose that
a normal mode is a very special situation that would not occur very
often. You would be correct. Nevertheless, normal modes are quite im-
portant because it turns out that the motion of a system of oscillators
can always be expressed as a linear combination of normal modes.
As a simple example, assume the system described by Equations
(11.34) and (11.35) was set in motion in such a way that C2 = 0. The
positions of the two masses would then be described by the equations
x1 = C1 cos(ω1 t + θ1 ), (11.36)
x2 = C1 cos(ω1 t + θ1 ).
In this situation, both masses oscillate with the same frequency ω1 .
Furthermore, x1 = x2 at all times. This is a normal mode. What does
the motion look like? The two masses are moving in phase with each
other. The frequency of the motion is

1
ω1 = ω02 − ∆ω 2 2 .
Notice that this is a somewhat lower frequency than the natural fre-
quency of the individual masses.
How could this normal mode be set up? As your physical intuition
probably suggests, you could generate the motion by pulling the two
masses equal distances to the right or left, and releasing them from
rest.
Suppose both masses are each pulled a distance b from equilibrium
and released from rest. Then

x1 (t = 0) = b,
x2 (t = 0) = b.
But in general,
x1 (t) = C1 cos(ω1 t + θ1 ) + C2 cos(ω2 t + θ2 ),
x2 (t) = C1 cos(ω1 t + θ1 ) − C2 cos(ω2 t + θ2 ).
so,
b = C1 cos θ1 + C2 cos θ2 , (11.37)
b = C1 cos θ1 − C2 cos θ2 .
The velocities of the blocks are:
408 11. HARMONIC MOTION

ẋ1 (t) = −C1 ω1 sin(ω1 t + θ1 ) − C2 ω2 sin(ω2 t + θ2 ),

ẋ2 (t) = −C1 ω1 sin(ω1 t + θ1 ) + C2 ω2 sin(ω2 t + θ2 ).
Since the blocks are released from rest at t = 0,
0 = −C1 ω1 sin θ1 − C2 ω2 sin θ2 ,
0 = −C1 ω1 sin θ1 + C2 ω2 sin θ2 .
Adding these two equations shows that θ1 = 0 and subtracting them
yields θ2 = 0. Equations (11.37) then become
b = C1 + C2 ,
b = C1 − C2 .
Hence,

C1 = b,
C2 = 0.
Therefore the motion is described by Equations (11.36) as desired. Fur-
thermore, C1 = b and θ1 = 0, so
x1 (t) = b cos ω1 t, (11.38)
x2 (t) = b cos ω1 t,
p
where ω1 = ω02 − ∆ω 2 . This is the low frequency normal mode.
Similarly, a different normal mode can be generated by either pulling
the two masses apart or pushing them together by equal amounts, and
releasing them from rest. Assuming the initial displacements are equal
to b0 (but in opposite directions), we find C1 = 0 and C2 = b0 .
The motion is then described by

x1 = b0 cos ω2 t, (11.39)
x2 = −b0 cos ω2 t,
where 1
ω2 = (ω02 + ∆ω 2 ) 2 .
This means the system oscillates at a higher frequency than the “nat-
ural” frequency ω0 .
Note that the general solutions (Equations 11.34 and 11.35) are
simply linear combinations of the normal mode solutions, Equations
(11.38) and (11.39). As mentioned before, the motion of any system
of coupled oscillators is always a linear combination of normal modes.
The problem gets more complicated when the number of oscillators
 11.6. SUMMARY 409

increases. This leads to the subject of a continuous system of oscillators

that is treated in Chapter 18

Exercise 11.29. Using the given initial conditions show that C2 =

b and θ2 = 0 in Equations (11.39).

11.6. Summary
This chapter was an in-depth study of harmonic motion using a
mass connected to a spring as an example. There were several digres-
sions to discuss the mathematical techniques involved in solving the
equations of motion. These techniques are used frequently in many
areas of physics, so you should make a special effort to understand and
remember them.
The analysis started with an undamped simple harmonic oscillator
with equation of motion

mẍ + kx = 0.
Next we generalized to a damped harmonic oscillator in which there
was a frictional force acting to slow the system down, with equation of
motion
mẍ + bẋ + kx = 0.
There were three cases: the underdamped oscillator with (b/2m)2 <
k/m, the critically damped oscillator with (b/2m)2 = k/m, and the
overdamped oscillator with (b/2m)2 > k/m. The underdamped oscil-
lator is the only one exhibiting truly oscillatory motion with the mass
moving back and forth in a regular, repetitive way. Note, however, that
in all three cases the amplitude of the motion decreases as time goes
on. Mathematically this is described by the amplitude being given by
Ae−γt so that the motion of an underdamped oscillator is
x(t) = Ae−γt cos(ω1 t + θ),
the motion of a critically damped oscillator is
x(t) = (C1 + C2 t) e−γc t ,
and the motion of an overdamped oscillator is:
x(t) = C1 e−γ1 t + C2 e−γ2 t .
Note also that the frequency of the motion is affected; an underdamped
oscillator has a frequency smaller than the frequency of the undamped
oscillator. The overdamped oscillator has a “frequency” greater than
410 11. HARMONIC MOTION

that of the undamped oscillator, but it does not really oscillate, the
motion simply dies out before a single full oscillation can occur.
Next you studied the forced harmonic oscillator. In this case the
motion is represented by the following inhomogeneous differential equa-
tion:
mẍ + bẋ + kx = F (t).
Obtaining the motion required summing the general solution of the ho-
mogeneous differential equation and a particular solution of the inho-
mogeneous equation. You learned a number of “tricks” for determining
the particular solutions of inhomogeneous equations.
Finally, you learned about coupled undamped harmonic oscillators
whose equations of motion are
m1 ẍ1 = −k1 x1 − k3 (x1 − x2 ),
m2 ẍ2 = −k2 x2 − k3 (x2 − x1 ).
This is an important problem in physics because there are many phys-
ical systems that can be represented as coupled harmonic oscillators.
Normal modes were defined and it was noted that the general motion of
coupled oscillators can be expressed as a linear combination of normal
mode solutions.

11.7. Problems
Problem 11.1. Determine the condition such that x = Ctept is a
.. .
solution of mx + bx + kx = 0.
Problem 11.2. Determine the time average of kinetic energy and
potential energy for a simple harmonic oscillator and show that they
are equal.
Problem 11.3. A bathroom scale should not oscillate. Ideally it
would be critically damped. Show that if a scale is critically damped for
a person of weight W it will be overdamped for a person whose weight
is less than W. If it is desired that for critical damping, the platform
deflect 2 cm for a 70 kg person, determine the spring constant k and
the damping constant b.
Problem 11.4. A 0.25 kg mass is attached to a spring of force
constant 0.02 N/m. It is placed in a resistive medium and released
from rest at a distance of 0.1 m from equilibrium. After 5 seconds, it
is observed that the amplitude of the oscillations is 0.05 m. What is
the damping constant b? What is the frequency of the oscillations?
 11.7. PROBLEMS 411

Problem 11.5. Consider p the underdamped harmonic oscillator so-

lution. Show that if ω1 = ωo2 − γ 2 is very small, the underdamped
solution is approximately equal to the critically damped solution for a
short time interval. Estimate the time (in terms of the relevant con-
stants) when the two solutions can no longer be considered equivalent.
Problem 11.6. In a railroad switching yard, at the end of each
track there is a large wooden bar connected to heavy springs. The
purpose of this bumper is to bring a rolling freight car to a stop. Clearly,
it is not desirable for the freight car to bounce off the bumper and go
rolling back on the track. (Note that the freight car is not attached to
the bumper. ) (a) Using appropriate sketches of position as a function
of time explain why the car will come to a stop in contact with the
bumper if it is critically damped or overdamped. (b) Assume critical
damping and obtain an expression for the maximum compression of
the springs in terms of m, k, bc and the speed of the car when it hits
the bumper.
Problem 11.7. Show that the critically damped oscillator will,
in general, approach zero faster than the overdamped oscillator. De-
termine the special condition such that the overdamped oscillator ap-
proaches zero faster than the critically damped oscillator.
Problem 11.8. For the critically damped harmonic oscillator ob-
tain expressions for C1 and C2 for arbitrary initial conditions x0 and
v0 .
Problem 11.9. A long pendulum with a heavy bob and small
amplitude oscillations will swing back and forth for a long time before
coming to rest. This means that the damping is small, that is, γ  ω0 .
Nevertheless, the pendulum is losing energy. Show that the rate of
change of the energy can be expressed as
d
(ln E) = −2γ.
dt
(You can use the equations we obtained for a mass on a spring, since
the mathematics is the same.)
Problem 11.10. In a certain industrial process, packages fall off
of the end of a conveyer belt onto a scale that is a distance h below it.
It is found that if the package has mass M the scale ends up a distance
δ below its equilibrium position. It is desired that the scale reach
this equilibrium position as rapidly as possible without overshooting.
Obtain the optimum value of the damping coefficient of the scale.
412 11. HARMONIC MOTION

Problem 11.11. Determine the motion x(t) for a critically damped

harmonic oscillator subjected to an applied force given by F0 cos ωd t.
You may assume that initially the mass m was displaced from the
equilibrium point by a distance x0 and that it had zero velocity.
Problem 11.12. The motion of an underdamped harmonic oscilla-
tor is x(t) = Ae−γt cos(ω1 t + θ). Assuming the initial conditions x0 = 1
and v0 = 0, determine A and θ.
Problem 11.13. Determine the motion for a driven damped har-
monic oscillator which is acted upon by an exponentially increasing
force given by F0 (1 − e−at ). The oscillator consists of a block of mass m
on a horizontal surface and attached to a spring of constant k = 4ma2 .
The damping coefficient is b = ma. Assume the oscillator is at rest at
t = 0. Plot the displacement of the mass as a function of time.
Problem 11.14. Consider a driven underdamped harmonic oscilla-
tor. Let the driving frequency ωd be equal to ω1 , which is the frequency
of the undriven oscillator and which is close to the resonance frequency
ω 0 . (a) Show that the ratio of the square of the maximum amplitude
squared (the value of A2 when ωd = ω 0 ) to the amplitude at the nearby
frequency is
A2max 4m2 (ωd − ω 0 )2 + b2
= .
A2 b2
(b) Show that the resonance half width is ∆ = b/2m. The resonance
half-width is the half-width of the resonant peak (ω 0 − ωd ) at a point
where the amplitude squared has fallen to half of its maximum value.
Problem 11.15. Two blocks of mass m1 and m2 are connected
by a spring with force constant k. The masses are on a frictionless
horizontal surface. Therefore, the masses slide freely on the surface
and oscillate back and forth relative to one another. Prove that the
velocity of the center of mass of the system is constant. Determine the
frequency of oscillation of the masses in terms of m1 , m2 , and k.
Problem 11.16. Figure 11.10 illustrates a system of two masses
and three springs. The analysis in the text assumed that m1 = m2 and
k1 = k2 . Now you are asked to consider the situation when the masses
are still the same, but the spring constants are different. Assume that
k1 = 0.8k and k2 = 1.2k but k3 is weaker and given by k3 = 0.1k.
The initial conditions are that one of the masses is held at rest in
its equilibrium position and the other is displaced from equilibrium
by a distance d. Obtain expressions for x1 (t) and x2 (t). Plot these
expressions and show they are in qualitative agreement with the sketch
below.
 11.7. PROBLEMS 413

x1
t

x2
t

Figure 11.11. Problem 11.16.

Problem 11.17. Consider two masses coupled through a spring

as in Figure 11.10. Prove the assertion that the low frequency normal
mode is generated if both masses are displaced an equal amount either
to the right or the left.
Problem 11.18. For the system pictured in Figure 11.10, assume
the two masses are equal but all three springs have different force con-
stants. Find the characteristic frequencies.
Problem 11.19. A system consists of three masses connected by
two springs, as shown in Figure 11.12 (This is called “coupling through
a mass.”) Set up the equations of motion for this system. The relaxed
lengths of the two springs are l1 and l2 . Separate the problem into two
problems, one involving the motion of the center of mass and the other
involving the motion of the two end masses relative to the central mass,
as given by the two coordinates x1 and x2 . Determine the normal modes
for the system.

x1 l1 l2 x2

m1 m3 m2
k1 k2

Figure 11.12. Coupling through a mass.

Problem 11.20. Consider an underdamped harmonic oscillator

subjected to a driving force F0 sin(ωt + θ0 ). Determine the rate at
414 11. HARMONIC MOTION

which work is done on the oscillator by the applied force, and eval-
uate the average power delivered by the force per cycle. Answer:
P = 12 F0 ẋmax cos β.
A few Differential Equation Problems
Problem 11.21. Obtain the general solution to (D2 +1)x = 12 cos2 t.
Answer: x = A cos t+B sin t+6−2 cos 2t. (Hint: use the method of un-
determined coefficients. You will have to look it up in your differential
equations text.)
Problem 11.22. Obtain the general solution to (D2 +4)x = 4 sin2 t.
(Hint: use the method of undetermined coefficients.)
COMPUTATIONAL PROJECT
Computational Project 11.1. A physical pendulum is illus-
trated in figure 11.13. Assuming that it is damped and driven by a
force of frequency f, its equation of motion can be written in the (sim-
plified) form
d2 θ dθ
2
+ b + sin θ = T sin(2πf t).
dt dt
It will be easiest to solve this equation numerically if you introduce
three new variables defined as follows:
dθ
x1 = ,
dt
x2 = θ,
x3 = 2πf t.
Then the equations of motion are
dx1
= T sin(x3 ) − sin(x2 ) − bx1 ,
dt
dx2
= x1 ,
dt
dx3
= 2πf.
dt
Write a computer program and plot θ (or x2 ) as a function of time for
t ≤ 250. First assume b = 0 and T = 0 to be sure that your solution
generates the oscillatory motion of a pendulum. Then set b = 0.1 and
obtain the motion of a damped pendulum. Finally, let T = 1, and
b = 0.1 and allow f to range from 0.001 to 0.1. Note that for f = 0.1
the motion is unpredictable. This is an example of chaos in a dynamical
system.
 11.7. PROBLEMS 415

Torque

θ M

Mg

Figure 11.13. A forced physical pendulum.

Chapter 12
The Pendulum

As you will soon appreciate, the physics of a pendulum is much

more complex than you might imagine.1
The pendulum has been one of our prime examples of harmonic
motion, but so far we have only considered pendulums whose motion
is restricted to a vertical plane. Furthermore, we assumed that the
amplitude of the swing was small so the sine of the angle could be
replaced by the angle itself (sin θ ≈ θ). Finally, we assumed the pendu-
lum was “ideal” and consisted of a point bob attached to a massless,
inextensible string. In this chapter we generalize the pendulum prob-
lem in several different ways. The first generalization is to allow the
pendulum swing to have an arbitrary amplitude. Next, we assume the
pendulum is an extended body, not a suspended point mass. (In that
case it is called a “physical” or “compound” pendulum.) Finally, we
consider a pendulum whose bob is not restricted to swing in a plane but
is free to trace out any motion on the surface of a sphere whose radius
is equal to the length of the pendulum. (That is called a “spherical”
pendulum.) If the bob of a spherical pendulum is restricted to move in
a horizontal circle, it is called a “conical” pendulum (because the string
sweeps out a cone). While studying these various types of pendulums
you will be exposed to valuable mathematical and physical techniques;
for example, you will find out how to use and evaluate elliptic integrals,
and you will learn the basic method for solving physics problems by
perturbation techniques. The Foucault pendulum, however, will be left

1Entire books have been written on pendulums. For example, Baken, G.L.,
and J. A. Blackburn, The Pendulum, A Case Study in Physics,Oxford University
Press, 2005, is a 300 page book on various aspects of pendulums.
417
418 12. THE PENDULUM

until the next chapter when we consider motion in accelerated reference

frames.

12.1. A Simple Pendulum with Arbitrary Amplitude

Let us begin by considering a simple (ideal) pendulum, just to get
the necessary equations in their most basic form. Then I will remove
the restriction of small angle oscillations. (The subsequent analysis
will lead to a consideration of two mathematical techniques, namely,
the use of elliptic integrals and expansions in series.)
Figure 12.1 shows a simple pendulum made up of a point mass
m and a massless inextensible string of length l. You have obtained
the equation of motion for this system using Newton’s second law (see
Equation 4.1) and also by the Lagrange technique (see Equation 11.3).
In this chapter we shall be using the equations of rotational dynamics,
so let us derive the equation of motion for the pendulum one more time,
using the rotational form of Newton’s second law (Equation 7.12)
N = Iα.
Here N is the torque, I is the moment of inertia (see Equation 7.10)
and |α|=α = θ̈ is the angular acceleration. The torque is
N = r × F = r × (T+mg).
If r is measured from the point of support, r × T = 0, leaving us with
N = r×mg.
But |r| = l so we can write
N = |N| = −lmg sin θ,
where the minus sign is due to the convention that clockwise torques
are negative. It is also a reminder that this is a “restoring torque” that
tends to make the angle θ smaller.
mi ri2 so for a simple
P
The moment of inertia is defined as I =
pendulum
I = ml2 .
Then N = I θ̈ yields
−mgl sin θ = ml2 θ̈,
or
g
θ̈ +
sin θ = 0. (12.1)
l
This seemingly simple equation turns out to be surprisingly difficult
to solve unless we make the small angle approximation. If θ . 0.1π,
 12.1. A SIMPLE PENDULUM WITH ARBITRARY AMPLITUDE 419

l
θ T

mg

Figure 12.1. A simple pendulum. The forces on the

bob are gravity (mg) and the tension (T) in the string.

replacing sin θ by θ introduces an error of less than 2%. So for small

amplitude oscillations, Equation (12.1) can be approximated by
g
θ̈ + θ = 0.
l
You recognize this as the equation for simple harmonic motion. As
discussed in Chapter 11, the solution is
θ = A cos(ωt + β), (12.2)
p
where ω = g/l = the angular frequency. A and β are constants that
depend on the initial conditions. The period of ap simple pendulum
with small amplitude oscillations is P = 2π/ω = 2π l/g.
Now I will remove the small angle restriction and determine the
motion of a pendulum with arbitrary amplitude. Instead of integrating
the equation of motion twice, I will use the energy integral described in
Section 5.7. This is convenient because it introduces a “first integral”
or constant of the motion, namely, the total energy E.
The total energy of a pendulum is constant (as long as there is no
damping). The kinetic energy of an oscillating pendulum is T = 21 Iω 2
or
1
T = ml2 θ̇2 .
2
The potential energy is
V = −mgl cos θ.
Consequently, the total energy is
1
E = T + V = ml2 θ̇2 − mgl cos θ.
2
420 12. THE PENDULUM

Solving for θ̇ yields

s  
dθ 2g E
θ̇ = = + cos θ .
dt l mgl
This is the differential equation
p we need to solve. The first step is to
separate variables. Using g/l = ω, and rearranging
dθ √
r  = 2ωdt.
E
mgl
+ cos θ

Integrating,
Z θ(t)
dθ √ Z t
r  = 2ω dt,
θ0 =0 E 0
mgl
+ cos θ
or Z θ
1 dθ
t= √ r , (12.3)
2ω 0 E
mgl
+ cos θ

where the lower limit, θ0 = 0, means the bob was at the bottom of the
swing when the timer was started.

12.1.1. Solution in Terms of Elliptic Integrals. It may sur-

prise you to learn that the integral in Equation (12.3) is not in your
table of integrals. However, it can be put into the form of a function
called an elliptic integral. Elliptic integrals are tabulated in all good
books of mathematical tables and are included in the libraries of many
computer languages. A complete elliptic integral of the first kind has
the form
Z π/2
dφ
p (12.4)
0 1 − k 2 sin2 φ
and is a function of k. Writing the integral in Equation (12.3) in the
form of (12.4) yields an expression for the period of a pendulum of
arbitrary amplitude in terms of an elliptic integral. This may sound
easy, but the algebra gets a bit complicated.
For convenience, assume the pendulum is not “going over the top”
so it has a maximum angular displacement, θm . That is, at the top of
the swing, θ = θm . But at the top of the swing, the kinetic energy is
zero, so the total energy is
E = −mgl cos θm .
 12.1. A SIMPLE PENDULUM WITH ARBITRARY AMPLITUDE 421

Since the total energy is constant, we can replace E/mgl in Equation

(12.3) by − cos θm . Thus,
Z θ
1 dθ0
t= √ p .
2ω 0 (cos θ0 − cos θm )
The period (P ) of this pendulum is the time required for the bob to
go through four quarter cycles where a quarter cycle is a swing from
θ = 0 to θ = θm . In the expression above, t is the time for the bob to
swing from θ = 0 to an arbitrary angle θ. Therefore,
 Z θm 
1 dθ
P =4 √ √ . (12.5)
2ω 0 cos θ − cos θm
This is still not in the form of the elliptic integral of Equation (12.4).
To put (12.5) into the required form recall the trigonometric identity
cos θ = 1 − 2 sin2 2θ .
Consequently,
p
1/2
1 − 2 sin2 2θ − 1 − 2 sin2 θ2m


cos θ − cos θm = ,
1/2
= 2 sin2 θ2m − 2 sin2 2θ
 
,
" #1/2
2 θ
1/2 sin
= 2 sin2 θ2m 2
 
1− 2 θm
.
sin 2

For convenience, let k = sin θ2m and write

√
r
p 1
cos θ − cos θm = k 2 1 − sin2 2θ .
k2
Therefore, Equation (12.5) is
Z θm
4 dθ
P =√ √ q . (12.6)
2ω 0 1 2 θ
2k 1 − k2 sin 2

Even this is not in the right form, but it is getting close. Define a new
angle φ by the relation
1 θ
sin φ = sin .
k 2
When θ = θm , the value of φ is π/2. Now,
1 θ 1 θ θ 1
cos 2θ dθ,


d(sin φ) = d( sin ) = cos d 2
=
k 2 k 2 2k
422 12. THE PENDULUM

so
2k cos φdφ 2k cos φdφ 2k cos φdφ
dθ = θ
=q =p .
cos 2 1 − sin2 2θ 1 − k 2 sin2 φ
Plugging into Equation (12.6) you obtain
Z φ=π/2
4 2k cos φdφ 1
P = √ p √ p ,
2 ω φ=0 1 − k 2 sin2 φ 2k 1 − sin2 φ
Z π/2
8 cos φdφ 1
= p .
2ω 0 1 − k 2 sin2 φ cos φ
Hence,
4 π/2
Z
dφ
P = p , (12.7)
ω 0 1 − k 2 sin2 φ
which has the desired form.
This somewhat tiresome derivation yielded the expression for the
period of a pendulum of arbitrary amplitude. The period is, however,
expressed in terms of an elliptic integral. Let me remind you that an
elliptic integral is a function like the sine or cosine in the sense that
its value can be looked up in any standard set of math tables. (I do
not know if any handheld calculators have elliptical integrals stored in
them.)
Elliptic integrals come in various different “flavors,” the primary
ones being:
(1) Elliptic Integrals of the First Kind:
Z θ
dφ
F (k, θ) = p . (12.8)
0 1 − k 2 sin2 φ
(2) Elliptic Integrals of the Second Kind:
Z θq
E(k, θ) = 1 − k 2 sin2 φ dφ. (12.9)
0
(3) Elliptic Integrals of the Third Kind:
Z θ
dφ
Π(n, k, θ) = 2
p . (12.10)
0 (1 − n sin φ) 1 − k 2 sin2 φ
(4) Complete Elliptic Integrals: These are the values of F and
E when θ = π/2. Thus, for example, the complete elliptical
integral of the first kind is:
Z π/2
π dφ
F (k, ) = F (k) = p .
2 0 1 − k 2 sin2 φ
 12.1. A SIMPLE PENDULUM WITH ARBITRARY AMPLITUDE 423

Note that the period of a pendulum is expressed as a complete

elliptic integral of the first kind.
Worked Example 12.1. A certain pendulum has a small am-
plitude period of two seconds. Determine its period when it swings
to a maximum angle of 80◦ .
Solution: The angular frequency for small amplitude oscilla-
tions is ω = 2π/P = 2π/2 = π = 3.14 rad/s. The quantity k
is
θm
k = sin = sin 40◦ = 0.64
2
Looking in a table of complete elliptic integrals of the first kind we
find
sin−1 k F (k, π/2)
0◦ 1.5708
.. ..
. .
39◦ 1.7748
40◦ 1.7868
◦
41 1.7992
Therefore, the period when the amplitude is 80◦ is
4
P = (1.7868) = 2.28 sec.
π
Exercise 12.1. Use a table to determine the values of (a) the el-
liptic integral of the first kind for k=0.5 φ=30◦ , (b) the elliptic integral
of the second kind for k=0.5 φ=30◦ (c) the complete elliptic integral of
the second kind for k=0.5. (The purpose of this exercise is to get you
to look at a table of elliptic integrals.)

12.1.2. Solution Expressed as a Series Expansion. If you do

not have a book with tabulated values of the elliptic integrals and your
computer is not available, you can still solve Equation (12.7) the “old
fashioned way” by expanding the integrand. This is a particularly
useful mathematical procedure. It works in this case because both k
and sin φ are less than unity.
Let
k 2 sin2 φ = x.
Then
4 π/2 dφ 4 π/2
Z Z
P = √ = dφ[1 − x]−1/2 .
ω 0 1−x ω 0
424 12. THE PENDULUM

You can expand the term in brackets using the binomial expansion
because x < 1. Keeping the first few terms,
4 π/2
Z
1 3
P = dφ[1 + x + x2 + · · · ] (12.11)
ω 0 2 8
or
4 π/2
Z
∼ 1 3
P = dφ[1 + k 2 sin2 φ + k 4 sin4 φ]
ω 0 2 8
(Z )
π/2
1 2 π/2 2 3 4 π/2 4
Z Z
4
= dφ + k sin φdφ + k sin φdφ .
ω 0 2 0 8 0

The integrals over φ are easily evaluated, yielding

s 
2
l 1 k2
  
4 π πk
P = + + · · · = 4π + + ··· .
ω 2 4 2 g 2 8
So,  
1 θ
P ∼
m
p 2
p 2
= 2π l/g(1 + k /4) = 2π l/g 1 + sin .
4 2
Note that the first term in this expression is the period of a small
amplitude ideal pendulum.

Exercise 12.2. Use the binomial expansion to carry out Equation

(12.11) to one more term.
Exercise 12.3. A simple pendulum has a length of 1.5 m. It swings
through an arc with maximum angular displacement of 40o . Determine
its period. Answer: 2.66 sec.
Exercise 12.4. A simple pendulum swings through an angle of 45
degrees (θm =45◦ ). Show that the period is about 4% greater than that
of small angle oscillations.

12.2. The Physical Pendulum

We have considered the motion of an ideal pendulum with a point
mass bob, suspended by an inextensible, massless string. This is a
good approximation for some pendulums, but everyone is familiar with
oscillating objects that are not at all similar to an ideal pendulum.
 12.2. THE PHYSICAL PENDULUM 425

For example, a large chandelier or a hanging sign may undergo reg-

ular oscillations. Such objects are called “physical” or “compound”
pendulums.
A compound pendulum is an extended body constrained to rotate
about a fixed axis that passes through some point in the body, as
illustrated in Figure 12.2. Let the point of support (through which the
axis passes) be denoted by O and let G indicate the position of the
center of mass (or center of gravity). The angle between the line OG
and the vertical is θ.

O
Rc
θ
G
Mg

|Rc |=h

Figure 12.2. A physical pendulum. The axis of ro-

tation is perpendicular to the page and passes through
point O. The center of mass is denoted G. The distance
from the axis of rotation to the center of mass is denoted
h. That is, h = |Rc | .

A rigid body is a collection of particles. The total torque on the

body is, therefore,
X
N= (ri ×Fi )
where N is the torque about O, ri is the vector from O to the ith
particle (which has mass mi ) and Fi = mi g is the gravitational force
on the ith particle. Consequently,
X X 
N= ri × (mi g) = mi ri × g.
P
By definition, the center of mass (G) is at Rc = mi ri /M, where M
is the total mass. Therefore, the torque is
N = Rc ×M g.
This expression shows that the torque on the body is the same as if a
force M g were acting at the center of mass of the body. That is, as far
as the torque is concerned, you can assume all the mass is concentrated
at the center of mass.
426 12. THE PENDULUM

The torque is a vector pointing into the page and it produces a

clockwise rotation of the body. As mentioned previously, this is a
negative torque, so the magnitude of N is

N = −M g |Rc | sin θ = −M gh sin θ,

where the distance from the axis of rotation to the center of mass is
denoted h.
When considering the motion of rigid bodies it is convenient to
define a quantity called the “radius of gyration,” denoted by k. The
radius of gyration is defined in terms of the moment of inertia. Recall
that the dimensions of the moment of inertia are mass multiplied by
the square of a distance. If I is the moment of inertia of a body relative
to some given axis, and M is the mass of the body, then the radius of
gyration is the distance k such that

I = M k2.

Although the radius of gyration does not have a geometrical meaning,

it is sometimes helpful to think of it as the distance from the axis of
rotation to the point where all the mass of the body would have to be
concentrated to have the same moment of inertia as the actual body.
For example, the moment of inertia of a sphere with respect to an
axis through its center (a diameter) is 52 M R2 . The radius of gyration
p
(relative to that axis) is k = 2/5R. The center of mass is on the axis
so h = 0, clearly illustrating that for a given axis, the radius of gyration
(k) and the distance to the center of mass (h) are not generally equal
to one another.
The moment of inertia of the physical pendulum of Figure 12.2 with
respect to the axis through O can be written

I0 = M k02 .

Using I θ̈ = N, the equation of motion of the physical pendulum is

M k02 θ̈ = −M gh sin θ.

Consequently,
gh
θ̈ = − sin θ.
k02
Recall that the equation of motion of a simple pendulum is
g
θ̈ = − sin θ.
l
 12.2. THE PHYSICAL PENDULUM 427

Comparing the two equations you see that the compound pendulum
oscillates with the same period as a simple pendulum of length
k02
l= . (12.12)
h
That is, a simple pendulum of length k02 /h will oscillate “in time” with
the compound pendulum. Figure 12.3 illustrates such a pendulum.
The point of suspension is O and the bob is at O0 (a distance l =
ko2 /h from O). The point O0 is called the “center of oscillation” of the
compound pendulum. (If all the mass of the compound pendulum
were somehow concentrated at O0 , it would be a simple pendulum that
oscillates with the same period as the real compound pendulum.) p For
small oscillations, the period of a simple pendulum is P = 2π l/g, so
the period of a compound pendulum is
p
P = 2π k 2 /hg.

O
l=k02/h
h l=k02/h
θ
G
h'
O'

Physical Pendulum Equivalent simple pendulum

Figure 12.3. The location of the bob of an equivalent

simple pendulum would be at O0 . This point is called the
“center of oscillation.” The distance from the center of
mass (G) to the center of oscillation is denoted h0 .

If the pendulum is hung from O0 , then the center of oscillation is

at O. We now show that a pendulum suspended from O0 has the same
period as a pendulum suspended from O. From the figure note that
l = h + h0 ,
where h0 is the distance from O0 to the center of mass. Using Equation
(12.12),
k2
h + h0 = 0 ,
h
or
k02 = h2 + hh0 . (12.13)
428 12. THE PENDULUM

Now according to the parallel axis theorem, the relation between the
moment of inertia about O (I0 ) and the moment of inertia about the
center of mass (IG ) is
I0 = IG + M h2 ,
so
M k02 = M kG
2
+ M h2 ,
where kG is the radius of gyration of a pendulum suspended at its
center of mass. Dividing by M,
k02 = kG
2
+ h2 . (12.14)
Comparing Equations (12.14) and (12.13) you can see that
h0 h = kG
2
. (12.15)
0 0
The “new” center of oscillation will be a distance l from O , where
l0 = k002 /h0 . Setting
k002 = h02 + hh0
leads to
1
l0 = 0 h02 + hh0 = h0 + h = l.


h
Therefore, a pendulum suspended at point O0 has its center of oscil-
lation at O. Note the symmetry between the location of the axis of
rotation and the location of the center of oscillation.

Worked Example 12.2. A uniform hoop of radius a and mass

m is free to oscillate about an axis perpendicular to the plane of the
hoop and passing through the rim. (Given: The moment of inertia
of a hoop about an axis through its center and perpendicular to
the plane of the hoop is ma2 .) (a) Determine its period, assuming
small amplitude oscillations. (b) Locate the center of oscillation
for this compound pendulum.
Solution: For the hoop, the distance from the rim to the center
of mass is the radius, so h = a.
p
P = 2π k 2 /hg
p
Now k = I/m and by the parallel axis theorem the moment of
inertia about a point on the rim will be
I = Icm + M R2 = ma2 + ma2 = 2ma2
k 2 = 2a2 .
So p p
P = 2π 2a2 /ag = 2π 2a/g.
 12.3. THE CENTER OF PERCUSSION 429

(b) The center of oscillation is a distance k 2 /h from the point of

suspension, that is, at a distance 2a2 /a = 2a.
Exercise 12.5. Determine the radius of gyration of a disk of mass
M and radius R: (a) relative to a perpendicular axis through its center,
and (b) relative √to a parallel
p axis tangent to the edge of the disk.
Answers: (a) R/ 2 (b) R 3/2.
Exercise 12.6. A uniform rod of mass M and length L is sus-
pended from one end. Determine the distance from the axis to the
center of mass (h) and determine the√radius of gyration (k) for the
same axis. Answers: h = L/2, k = L/ 3.
Exercise 12.7. A thin uniform rod of length L and mass M os-
cillates about an axis through one end. Determine the period of small
oscillations. Answer: (8π 2 L/3g)1/2 .

12.3. The Center of Percussion

If a rigid body is struck at an arbitrary point it will move and,
in general, this motion will consist of both rotational and translational
motion. It turns out that there is a point in the body where the transla-
tional and rotational motions cancel out so this point is instantaneously
at rest. This point is called the “center of percussion.” As you will see,
the center of percussion is related to another point called the “sweet
spot” in a tennis racket or baseball bat. In everyday terms, the sweet
spot is the spot where you want to hit the ball. If you hit the ball at
any other point on the racket or bat, you will feel a stinging in your
hands due to the reaction force. In other words, an impulse delivered
at the sweet spot does not cause an impulsive reaction at the center of
percussion (which is where you hold the racket or bat).
To begin the analysis, consider a rod of mass M that receives a
blow at point A. See Figure 12.4.
First assume the rod is not constrained by an axis of rotation.
For example, it might be lying on a perfectly smooth table. What
is the motion of the rod after receiving a blow at point A? This can
be determined in a straightforward way by an application of Newton’s
laws.
If the force on the rod is an impulsive blow, the impulse J is given
by Z
J = Fdt.
430 12. THE PENDULUM

P
s

G
a
Force A

Figure 12.4. A rod on a smooth table is subjected to

an impulsive force at point A. The distance between the
center of mass (G) and A is a. The point P is a distance
s from G.

But F = dp
dt
so Z Z
dp
J= dt= dp = pf −pi .
dt
Assuming the initial momentum is zero and the final momentum is
M v, the velocity of the rod immediately after the blow is
J
v= .
M
To be a bit more explicit, this is the velocity of the center of mass of
the rod immediately after the blow.
However, it is clear from the physical situation that the rod will
also begin to rotate. Its angular momentum (taken with respect to the
center of mass) will change according to
dL
= N.
dt
2
The moment of inertia about the center of mass is I = M kG . The
magnitude of the torque on the rod (taken about the center of mass)
due to the impulsive force F is
N = |r × F| =aF,
where a is the distance from the center of mass to the line of action of
the force. Taking the derivative of L = Iω,
d
aF = (Iω).
dt
Therefore,
Z Z
aF dt = d(Iω),
aJ = Iω,
 12.3. THE CENTER OF PERCUSSION 431

and the angular velocity about the center of mass, immediately after
the impulse, will be
aJ
ω= .
I
The motion of the rod is rather complicated. The velocity of a point
on the rod is the linear velocity of the center of mass, plus the rotation
about the center of mass (given by ω ×r where r is the vector from
the center of mass). These two velocities must be added vectorially,
of course. For example, consider the motion of point P, a distance s
from the center of mass, as shown in Figure 12.4. Immediately after
the impulse, the velocity of this point is
J
vP = + ω × s. (12.16)
M
But the right hand rule and Figure 12.4 tell us that J is directed toward
the right and ω × s is directed toward the left. The magnitude of the
velocity of point P is, consequently,
J aJ
vP = − 2
s
M M kG
 
J as
= 1− 2 .
M kG
This expression leads to the interesting conclusion that immediately
after the impulse there is a point on the rod having zero velocity. That
is, vP will be zero if
as
2
= 1.
kG
Consequently, immediately after the impulse, a point located at
2
kG
s= (12.17)
a
will have zero velocity. This point is called the center of percussion.
In the notation of the previous section, the distances a and s are
0
h and h, respectively. Inserting these quantities into Equation (12.17)
gives
hh0 = kG 2
,
which is just Equation (12.15). Thus the center of percussion is located
at the same point as the center of oscillation (assuming the axis is at
A, the point where the impulse is applied).
As time goes on, the impulsive force is no longer acting and the
linear and angular velocities will be constant. It is instructive to plot
the positions of a few points in the rod as a function of time to convince
yourself that the rod will rotate about its center of mass at a constant
432 12. THE PENDULUM

angular velocity, while the center of mass itself moves at a constant

linear velocity.
Now consider the possibility that the rod in Figure 12.4 is pivoted
at point P. In this case, if the rod is given a blow at point A there
will be no impulsive reaction at the pivot. To appreciate this, assume
the rod is pivoted at a distance s from the center of mass. (We are
no longer requiring that s be the distance to the center of percussion,
even though it will turn out to be that.)
In general, the struck rod will tend to translate, but a pivot at P
will prevent this. Point P remains at rest at all times. But if point
P is to remain at rest, the net force on that point must be zero. The
forces acting on P are the force due to the blow at A and the reaction
force due to the pivot. Let the force exerted by the pivot be F1 , as
indicated in Figure 12.5.

P s
F1
G
a
Force A

Figure 12.5. A rod pivoted at point P. The applied

impulsive force is F. The reaction force of the pivot on
the rod is F1 .

The impulses on the rod are J and J1 . Generalizing the analysis

above shows that the velocity of the center of mass after the blow will
be
J − J1
v= .
M
Since the rod is constrained to rotate about the pivot located a distance
s from the center of mass, the angular velocity of the rod is
1 J − J1
ω= .
s M
After the blow, the angular momentum is L = IP ω where IP is the
moment of inertia about point P. But L is also given by
Z Z
L = N dt = (s + a)F dt = (s + a)J.
 12.3. THE CENTER OF PERCUSSION 433

Equating the two expressions for L, the angular velocity is

J(a + s)
ω= .
Ip
Now equate the two expressions for ω :
1 J − J1 J(a + s)
= .
s M Ip
Therefore, the reactive impulse is
 
M s(a + s)
J1 = J 1 − .
Ip
Finally, according to the parallel axis theorem,
2
Ip = M kG + M s2 ,
so
2
kG − as
J1 = J 2
.
kG + s2
2
J1 will be zero if s = kG /a, as before. If an extended body is struck
by an impulsive force at A, there will be zero reaction force at point P
2
located at s = kG /a. If you are holding an extended body that happens
to be a baseball bat and a ball hits it at point A, you should hope that
your hands are gripping the bat at point P - otherwise you will have to
exert a reaction force on the bat, and you will end up with red, stinging
palms.

A
a
r d

Figure 12.6. A billiard ball of radius r strikes the edge

of the table at A.

Worked Example 12.3. A billiard ball strikes the edge of a

pool table and rebounds. It is desired that the angle of incidence
be equal to the angle of reflection. To obtain this result, it is
important that the ball should not start to slide on the table. The
434 12. THE PENDULUM

question is: What is the correct height (d) for the “bumper” on
the edge of the table in Figure 12.6?
Solution: Since the ball does not slide, point C should be the
center of percussion for a blow at A. Therefore,
2
ar = kG .
For a sphere, I = (2/5)M r2 , so kG
2
= (2/5)r2 . Consequently,
ar = (2/5)r2 ,
and
d = a + r = (7/5)r.

12.4. The Spherical Pendulum

A spherical pendulum consists of a point mass m suspended by a
string (or perhaps a massless rigid rod) of length l whose other end is
attached to a fixed point. The bob is constrained to be a fixed distance
from the point of support, but otherwise it is free to move anywhere
on the surface of a sphere of radius l.
To analyze the spherical pendulum, let us determine the Lagrangian
for the system, obtain the Lagrange equations (i.e., the equations of
motion), and integrate them to determine the motion of the bob.

θ
y

φ l
x
m

Figure 12.7. A spherical pendulum. The length of the

string is l. The position of the bob is given by θ and φ.

The Cartesian and spherical coordinates for a spherical pendulum

are shown in Figure 12.7. Observe that θ is measured from the posi-
tive z-axis, so when the pendulum is at rest and the string is hanging
straight down, the value of θ is π. The relations between the Cartesian
and the spherical coordinates are
 12.4. THE SPHERICAL PENDULUM 435

x = l sin θ cos φ,
y = l sin θ sin φ,
z = l cos θ.
The kinetic energy of the bob is
1
T = m(ẋ2 + ẏ 2 + ż 2 ),
2
1 2 2
= ml (θ̇ + sin2 θφ̇2 ), (12.18)
2
and the potential energy is
V = mgl cos θ.
Consequently, the Lagrangian is
1
L = T − V = ml2 (θ̇2 + sin2 θφ̇2 ) − mgl cos θ. (12.19)
2
The Lagrange equations of motion are
d  2 
ml θ̇ − ml2 φ̇2 sin θ cos θ − mgl sin θ = 0, (12.20)
dt
d  2 2 
ml sin θφ̇ = 0. (12.21)
dt
The second of these equations can be integrated immediately to give
ml2 sin2 θφ̇ = constant = pφ . (12.22)
The constant is named pφ because it is the generalized (angular) mo-
mentum associated with the (ignorable) coordinate φ.
It is convenient to introduce the energy of the pendulum. Adding
the kinetic and potential energies we obtain
1
E = ml2 (θ̇2 + sin2 θφ̇2 ) + mgl cos θ.
2
Using the expression for pφ to eliminate φ̇ leads to
1 2 2 p2φ
E = ml θ̇ + + mgl cos θ. (12.23)
2 2ml2 sin2 θ
The form of this expression suggests introducing an “effective poten-
tial” Vef f . (Recall the use of the effective potential for the central force
problem - see Section 10.5.) For the spherical pendulum, the term
(1/2)ml2 θ̇2 is a kinetic energy term depending on the velocity, but the
following two terms depend only on coordinates and not velocities so
436 12. THE PENDULUM

they “look like” potential energies. Therefore, the effective potential is

defined as
p2φ
Vef f = + mgl cos θ, (12.24)
2ml2 sin2 θ
and Equation (12.23) can be written as:
1 2 2
ml θ̇ = E − Vef f . (12.25)
2
The effective potential depends on the value of pφ . This parame-
ter characterizes the motion of the pendulum bob with respect to the
azimuthal angle φ.
If pφ = 0, the spherical pendulum reduces to a simple planar pen-
dulum and the effective potential reduces to the potential energy (that
is, Vef f = mgl cos θ). The effective potential (or potential energy) for
this special case is plotted in Figure 12.8.

Veff = V

E3
mgl

E2
θ
π/2 π 3 /2 2
E1
-mgl E0

Figure 12.8. Effective potential (equal to the potential

energy) for a simple planar pendulum. Note that pφ = 0.

We now consider the motion for different possible values of the

total energy E = T + V . Note that T is always positive, and V varies
between ±mgl. The smallest energy occurs when the kinetic energy
is zero and V = −mgl. As illustrated in Figure 12.8 this corresponds
to E = E0 = −mgl and represents a pendulum which is not moving
and is hanging straight down (θ = π). For larger values of the total
energy, such as E = E1 < 0 in the figure, the pendulum is trapped in a
potential well and oscillates back and forth around θ = π. The angular
amplitude of these oscillations is less than π/2; the pendulum does not
rise above the z = 0 plane. For larger values of total energy such as E2 ,
the pendulum swings about θ = π, but now the angular amplitude of
the oscillations exceeds 90◦ . The bob rises above the z = 0 level on each
 12.4. THE SPHERICAL PENDULUM 437

swing. Finally, for energies such as E3 > +mgl, the pendulum actually
“goes over the top” and the angle increases (or decreases) continuously.
A more interesting situation arises when pφ 6= 0. The pendulum now
moves azimuthally. It may or may not have motion in the θ direction.
If θ is constant, the pendulum bob traces out a horizontal circle of
constant radius. This special case is called the “conical pendulum.”
If pφ 6= 0, the effective potential is given by Equation (12.24). Vef f
is plotted in Figure 12.9 for various constant but nonzero values of
pφ . The effective potential rises to infinity at θ = 0 and at θ = π,
indicating that any motion of the pendulum involves angles between
these limiting values. The effective potential has a minimum at a point
whose location depends on pφ . This minimum gets closer and closer to
π/2 as pφ increases. For a given value of pφ the pendulum swings in
a circle about the vertical axis, while undergoing oscillations in θ. The
simplest motion of this type occurs when θ is constant and equal to its
value at the minimum in Vef f . This is, of course, a conical pendulum.
To treat the general case of the spherical pendulum, it is convenient to
start the analysis with the conical pendulum.
Effective Potential vs Theta
90
Effective Potential, V (arbitrary units)

80

70
10
60

50

40 8

30

6
20

10 4

0 2

-10
0 0.5 1 1.5 2 2.5 3

Theta (radians)

Figure 12.9. Effective Potential vs θ for various values

of pφ .

Exercise 12.8. Fill in the missing steps to obtain Equations (12.18),

(12.20), and (12.21)
Exercise 12.9. For the spherical pendulum, show that ml2 sin2 θφ̇
is the generalized momentum conjugate to φ.
438 12. THE PENDULUM

12.4.1. The Conical Pendulum. The conical pendulum is a spe-

cial case of the spherical pendulum in which pφ 6= 0 and θ = constant.
The bob moves in a horizontal circular orbit.
For a given value of pφ and a constant θ, the energy of the conical
pendulum will correspond to the lowest value of Vef f . (See Equation
12.24 and Figure 12.9.) For higher energies, θ oscillates between two
values, but this is no longer a conical pendulum. The lowest value
of θ is closer and closer to π/2 as the pendulum swings around faster
and faster, as can be appreciated by noting that in Figure 12.9 the
minimum in Vef f moves towards π/2 as pφ increases. That is, as the
value of pφ increases, the pendulum swings nearer and nearer to the
horizontal plane.
Denoting the polar angle of the conical pendulum by θ0 we can de-
termine the relationship between pφ and θ0 by noting that the effective
potential has a minimum at θ0 . Therefore,

p2φ
    
dVef f d
= + mgl cos θ = 0, (12.26)
dθ θ0 2ml2 sin2 θ
dθ θ0
 2
pφ d

−2
0 = sin θ − mgl sin θ ,
2ml2 dθ θ0
p2φ cos θ0
0 = − − mgl sin θ0 . (12.27)
ml2 sin3 θ0

Consequently,

p2φ = −m2 l3 g sin3 θ0 tan θ0 . (12.28)

As pφ approaches ∞, sin3 θ0 tan θ0 → −∞, and consequently θ0 ap-

proaches π/2 from an angle greater than π/2. That is, the pendulum
bob always hangs below the horizontal (z = 0) plane. Furthermore,
squaring the definition of pφ (Equation 12.22) and using Equation
(12.28), leads to
p
φ̇ = −g/(l cos θ0 ).

The negative sign does not give an imaginary quantity because θ0 is in

the second quadrant. As the azimuthal velocity φ̇ increases, the value
of θ0 gets closer and closer to π/2, in agreement with your childhood
experiments with conical pendulums.
 12.4. THE SPHERICAL PENDULUM 439

The energy of a conical pendulum is given by Equation (12.23) with

θ̇ = 0, that is,
p2φ
Ecp = + mgl cos θ, (12.29)
2ml2 sin2 θ
 2
ml2 sin2 θφ̇
= + mgl cos θ.
2ml2 sin2 θ
Setting θ = θ0 and using the expression for φ̇ yields
1 mgl
Ecp = (2 − 3 sin2 θ0 ).
2 cos θ0

Exercise 12.10. The string of an 85 cm conical pendulum makes

an angle of 20◦ with the negative z-axis. (That is, the vertex angle of
the cone is 20◦ .) Determine the period of the circular motion of the
bob. Answer: 1.83 sec.

12.4.2. The Spherical Pendulum. We now consider the spher-

ical pendulum. We shall only consider a pendulum for which the os-
cillations in θ are of small amplitude. (This is essentially a conical
pendulum which rises above and descends below the horizontal plane
at z = l cos θ0 as it goes around azimuthally.2)
According to Equation (12.23) the energy of the spherical pendulum
is
1 2 2 p2φ 1 2 2
E = ml θ̇ + 2 + mgl cos θ = ml θ̇ + Vef f (θ).
2 2ml2 sin θ 2
Figure 12.10 is a plot of the effective potential for an arbitrarily se-
lected value of pφ . Note the turning points at θ1 and θ2 . The θ motion
of the pendulum is an oscillation between these values as it revolves
azimuthally with constant pφ . This motion is illustrated in Figure 12.11
where the pendulum bob moves between the two horizontal circles de-
noted by θ1 and θ2 .
2Coherent discussions of the spherical pendulum are found in H. C. Corben
and Philip Stehle Classical Mechanics Wiley and Sons Inc., Hoboken, NJ: 1950
(republished by Dover Press in 1994), and in Grant Fowles and George L. Cassiday,
Analytical Mechanics, 5th ed. Harcourt Brace & Co, Orlando FL: 1993.
440 12. THE PENDULUM

Effective Potential, Veff E

Angle

Figure 12.10. Effective potential for a spherical pen-

dulum with a given value of pφ . The total energy is E
and the turning points are at θ1 and θ2 .

The θ equation of motion is given by equation (12.20), which can

be written as,
θ̈ − φ̇2 sin θ cos θ − (g/l) sin θ = 0. (12.30)
This differential equation can be solved in terms of elliptic integrals
(see Problem 12.20) but if the θ variations are small the problem can
also be solved by the technique of successive approximations, which we
will now consider.
In general, the technique of successive approximations is based on
finding the exact solution of a simpler system that approximates the ac-
tual system. The difference between the simple system and the actual
system can be expressed in terms of a small parameter . The simple
system that can be solved exactly is called the zeroth order approx-
imation. For the spherical pendulum with small variations in θ, the
motion is nearly that of a conical pendulum. The conical pendulum
is the zeroth order approximation and is proportional to 0 = 1. The
solution that is linear in the small quantity (1 ) is called the first order
approximation; the solution in which we keep terms involving the small
quantity squared (2 ) is called the second order approximation, and so
on. Thus,
[spherical pend] = 0 [conical pend] +1 [f1 (θ, φ)] +2 [f2 (θ, φ)] + · · ·
| {z }
zeroth order
| {z }
first order
| {z }
second order
 12.4. THE SPHERICAL PENDULUM 441

We will limit ourselves to the first order approximation of the spherical

pendulum.

π/2
θ1

θ2

Figure 12.11. Trajectory of a spherical pendulum

showing the motion between two parallel circles at dif-
ferent values of θ. The point of support of the pendulum
is the center of the sphere.

Let a conical pendulum with polar angle θ0 be slightly perturbed.

Then the angle θ of the resulting spherical pendulum will be close to
θ0 and to first order

θ = θ0 + θ1 . (12.31)

Here  is the “smallness parameter.”

Since

pφ = ml2 sin2 θφ̇,

φ̇ can be eliminated from the equation of motion, (12.30). Then defin-

ing β 2 = p2φ /m2 l4 , Equation (12.30) can be written as:

cos θ
θ̈ − β 2 − (g/l) sin θ = 0. (12.32)
sin3 θ

Inserting (12.31) into (12.32) leads to

cos (θ0 + θ1 )

θ̈1 − β 2 − (g/l) sin (θ0 + θ1 ) = 0.
sin3 (θ0 + θ1 )
442 12. THE PENDULUM

Expanding3 and discarding terms containing  raised to powers of 2 or

more, we obtain
3 cos2 θ0
   
2 cos θ0 1
θ̈1 −β − + θ1 +(sin θ0 + θ1 cos θ0 ) (−g/l) = 0.
sin3 θ0 sin2 θ0 sin4 θ0
Next, collecting terms in various powers of  we get
3 cos2 θo
     
0 2 cos θo g 1 2 1 g
 −β − sin θo + θ̈1 + θ1 β + − θ1 cos θo = 0.
sin3 θo l sin2 θo sin4 θo l
Since  is arbitrary, terms involving the various powers of  must vanish
separately. Setting the coefficients of 0 and 1 equal to zero, generates
the following two relations:

cos θ0 g
β2 3 + sin θ0 = 0, (12.33)
sin θ0 l
and
3 cos2 θ0
  
1 2g
θ̈1 + + β − cos θ0 θ1 = 0. (12.34)
sin2 θ0 sin4 θ0 l
Equation (12.33) gives
g sin4 θ0
β2 = − .
l cos θ0
Inserting this expression into equation (12.34) yields
1 + 3 cos2 θ0 g
 
θ̈1 − θ1 = 0. (12.35)
cos θ0 l
But this equation has the form θ̈1 +ω 2 θ1 = 0 (simple harmonic motion)
where
r
g β p
ω= − (1 + 3 cos2 θ0 ) = 1 + 3 cos2 θ0 (12.36)
l cos θ0 sin2 θ0

3Note that
cos(θ0 + θ1 ) = cos θ0 cos θ1 − sin θ0 sin θ1
=
˙ cos θ0 − θ1 sin θ0 ,
and
sin(θ0 + θ1 ) = sin θ0 cos θ1 + cos θ0 sin θ1
=
˙ sin θ0 + θ1 cos θ0 ,
where we used sin θ1 =θ
˙ 1 and cos θ1 =1.
˙
 12.4. THE SPHERICAL PENDULUM 443

With this definition of ω, a solution of (12.35) is4

θ1 = cos ωt, (12.37)
and so the θ motion is
θ = θ0 +  cos ωt.
We can now determine the corresponding φ motion. Start with the
definition of β,
 2
2 2
2
pφ ml sin θ φ̇
β2 = 2 4 = 2 4
= φ̇2 sin4 θ.
ml ml
Therefore,
β β
φ̇ = 2 = 2
sin θ sin (θ0 + θ1 )
Using the approximation given in the footnote above, we can write this
as
β
φ̇ = (1 − 2θ1 cot θ0 )
sin2 θ0
β
= (1 − 2 cos ωt cot θ0 ) .
sin2 θ0
Integrating, we obtain
 
β 2
φ= t − sin ωt cot θ0 . (12.38)
sin2 θ0 ω
This shows that on average φ is equal to its unperturbed value (βt/ sin2 θ0 )
but it oscillates around this value with a frequency ω.
I have not given a physical explanation for the parameter . It was
treated merely as a mathematical device for carrying out the expansion.
Now note that according to Equation (12.37) θ(t) = θ0 + θ1 = θ0 +
 cos ωt. So  is the amplitude of the oscillations of θ about θ0 . Due to
our choice of constants in Equation (12.37),  is the initial displacement
of the bob in the θ direction.
θ(t = 0) = θ0 + .
That is,  is the initial displacement of the bob from the conical angle θ0
of the unperturbed conical pendulum. Thus, in Figure 12.11 the values
of θ1 and θ2 are such that θ2 − θ1 = 2. (If we had written Equation
4In general θ1 = A cos(ωt + γ) so θ1 = cos ωt is not the most general solution.
Nevertheless it is a solution and it does satisfy the differential equation. You will
soon discover that the amplitude of the oscillations is , but writing θ1 = cos ωt at
this stage makes the derivation less confusing.
444 12. THE PENDULUM

12.37 more generally as θ1 = A cos(ω1 t + β) then these statements

would have to be modified somewhat.)
The period of the motion in the θ “direction” is P = 2π/ω and in
time P the pendulum will have gone through an entire oscillation in
θ. You can easily determine how far the azimuthal coordinate φ has
advanced in this time. Assume that φ = 0 at time t = 0. You can
determine φ at time t = 2π/ω by plugging t = 2π/ω into Equation
(12.38). An approximate value for φ(t = 2π/ω) is obtained by setting
 = 0 in Equation (12.38) yielding
. β 2π
φ= .
sin2 θ0 ω
But p
ω = (β/ sin2 θ0 ) 1 + 3 cos θ0 , (12.39)
so
. 2π
φ= √ .
1 + 3 cos2 θ0
Now cos2 θ0 must lie between zero and one. If cos2 θ0 = 0, then
φ = 2π, and if cos2 θ0 = 1, then φ = π. Therefore,
π ≤ φ ≤ 2π.
That is, while the pendulum goes through a complete oscillation in θ,
it advances azimuthally through an angle between π and 2π.
This concludes the analysis of the spherical pendulum, but I would
like to take a moment to consider the problem from a slightly differ-
ent point of view, based on the fact that the effective potential is a
minimum at θ0 . Expanding the effective potential about θ0 in a Taylor
series, we obtain
1 2 d2 Vef f
  
dVef f 
Vef f (θ) = Vef f (θ0 ) + η + η + ···
dθ θ0 2 dθ2 θ0
where η = θ − θ0 . Since θ0 is constant, Equation (12.25)
 indicates that
dVef f 
Vef f (θ0 ) = E0 . According to Equation (12.26) dθ  = 0. It is easy
θ0
to show that
d2 Vef f 

−mgl 2


= 1 + 3 cos θ0 = κ,
dθ2 θ0 cos θ0
where κ is a constant. Therefore, for values of θ near θ0 ,
1
Vef f (θ) = E0 + κη 2 ,
2
 12.5. SUMMARY 445

and the energy equation (12.25) becomes

1
2
ml2 θ̇2 = E − E0 − 12 κη 2 .
Since θ̇ = η̇, and E − E0 is just a constant, this equation is
1
2
ml2 η̇ 2 + 12 κη 2 = E 0 ,
where E 0 = E − E0 . Note that this has the same form as the energy
equation for a simple harmonic oscillator of mass ml2 attached to a
0
spring of constant
p κ, and having energy E . The frequency of oscilla-
2
tion in θ is ω = κ/ml . Plugging in the expression (12.39) for κ this
reproduces Equation (12.36). Therefore, the motion of a spherical pen-
dulum with energy slightly greater than E0 is circular motion around
the vertical axis with small oscillations in θ of frequency ω.

Exercise 12.11. For a spherical pendulum, show that the ratio of

the
√ angular frequencies in the azimuthal (φ) and polar (θ) directions is
1 + 3 cos2 θ0 .

12.5. Summary
After studying this chapter you will probably agree that the “simple
pendulum” is not so simple after all! One of the benefits of studying
the motion of a pendulum is that it leads to a consideration of a variety
of mathematical techniques.
The period of a simple pendulum with arbitrary amplitude is
4 π/2
Z
dφ
P = p ,
ω 0 1 − k 2 sin2 φ
a complete elliptic integral of the first kind. These have been tabulated
but they can also be evaluated using a series expansion.
An analysis of the physical pendulum introduces a number of prop-
erties of extended bodies, including the radius of gyration k defined in
terms of the moment of inertia as
I = M k2.
A simple pendulum oscillating with the same frequency as a physical
pendulum has a length given by
l = k02 /h,
446 12. THE PENDULUM

where h is the distance from the axis of rotation to the center of mass,
and k0 is the radius of gyration relative to the axis. The distance l is
also the distance from the axis of rotation to the center of oscillation.
If a physical pendulum receives an impulse at the center of percussion,
there will be no reaction at the axis. The center of percussion lies at
the same point as the center of oscillation (but conceptually these two
points are quite different).
The motion of a spherical pendulum is most easily understood by
introducing an effective potential
p2φ
Vef f = + mgl cos θ.
2ml2 sin2 θ
For the special case of a conical pendulum the angle θ = θ0 is constant
and the value of pφ is also constant, leading to an azimuthal angular
velocity of p
φ̇ = −g/(l cos θ0 )
and a total energy of
1 mgl
Ecp = (2 − 3 sin2 θ0 ).
2 cos θ0
The spherical pendulum presents a number of difficulties, but it
warrants study because it can be treated as a conical pendulum sub-
jected to a small perturbation about a known solution. (This technique
was used in Chapter 10 to study the motion of a slightly perturbed
planet. It will be used again in the next chapter.) You will become
very familiar with perturbation techniques when you study quantum
mechanics. A conical pendulum which has been perturbed so the am-
plitude of oscillations in θ is , has an azimuthal angular position given
by  
β 2 2π
φ= 2 t − sin ωt cot θ0 ' √ ,
sin θ0 ω 1 + 3 cos2 θ0
and the oscillations in θ have an angular frequency
p
ω = (β/ sin2 θ0 ) 1 + 3 cos2 θ0 ,
where the parameter β is defined by
g sin4 θ0
β2 = − .
l cos θ0
12.6. Problems
Problem 12.1. The rotational analogue to F = dp dt
is N = dL
dt
.
Starting with this equation, obtain a work-energy theorem for rota-
tional motion.
 12.6. PROBLEMS 447

Problem 12.2. Prove that if the torque is a function only of the

angle, N = N (θ), the total mechanical energy is conserved.
Problem 12.3. Consider a large amplitude simple pendulum. Es-
timate the fractional error in the period (relative to the elliptic integral)
due to ignoring all terms beyond the first order term in the expansion
(12.11) for a pendulum whose amplitude is 60◦ .
Problem 12.4. Elliptic integrals can be used to calculate the length
of the arc of a curve. In this problem you are asked to obtain a for-
mula for the circumference of an ellipse whose semimajor axis is a and
semiminor axis is b. Define
r
a2 − b 2
e=
a2
and show that the circumference is given by
C = 4aE(e).
(Hint: the parametric equations for an ellipse are x = a cos θ, and
1/2
y = b sin θ. An element of length along the curve is ds = [dx2 + dy 2 ] .)
Problem 12.5. A simple pendulum is attached to a device that
causes the point of support to oscillate up and down according to y 0 =
A cos f t, so the y-coordinate of the bob is given by y = y 0 −l cos θ. Here
θ is measured counterclockwise from the lowest point of the pendulum
swing. (a) Obtain the Lagrangian, (b) obtain the equation of motion,
and (c) determine the angular frequency.
Problem 12.6. A pendulum of length l0 is attached to the top of
a vertical disk of radius R as shown in Figure 12.12. The pendulum
is long enough and the amplitude is small enough that the bob never
touches the disk. (a) Show that the Lagrangian can be written as
1
L = m(A + Bθ)2 θ̇2 + mg [(A + Bθ) sin θ + B cos θ]
2
and obtain explicit expressions for A and B.(b) Obtain the equation
of motion. (c) The equilibrium position is θ0 = π/2 (the string hangs
straight down). Determine the frequency of small amplitude oscilla-
tions about θ0 .

Problem 12.7. A child stands on the seat of a playground swing.

The child’s body can be approximated by a cylinder of radius 10 cm,
length 1 m and mass 30 kg. The swing is suspended by ropes of length
2.5 m. The mass of each rope is 2 kg. (a) Determine the period of
oscillation. (b) Where is the center of oscillation of this system?
448 12. THE PENDULUM

θ
R

Figure 12.12. A pendulum of variable length. Problem 12.6.

Problem 12.8. A physical pendulum oscillates about an axis 10

cm from its center of mass. The radius of gyration about the axis is 15
cm. (a) What is the period? (b) Where is the center of oscillation?
Problem 12.9. A solid sphere is cut in half and a homogeneous
hemisphere of radius r and mass M is set upon a table (with its flat
side up). The surface of the table is perfectly rough. The hemisphere
rocks back and forth with small amplitude excursions from equilibrium.
What is the length of an equivalent simple pendulum? Justify approx-
imations. Note that the center of mass of a hemisphere is at a distance
3r/8 below the center of the sphere.
Problem 12.10. A crane is lowering a heavy mass at a constant
rate r. The mass swings back and forth, so this is (essentially) a pen-
dulum of increasing length. Show that the equation of motion can be
written as
d2 θ dθ g
l 2 + 2 + 2 θ = 0,
dl dl r
where l is the length of the cable. Assume small oscillations.
Problem 12.11. A wheel is mounted on an axle. One end of a
massless rigid rod is attached to the rim of the wheel and the other
end is attached to a spring, so that the rod exerts a torque on the rim
of the wheel. Assume the wheel is a disk of mass M and radius R and
let the spring constant be k. Determine the period of small amplitude
oscillations.
Problem 12.12. Kater’s pendulum is a physical pendulum that
has the same period when supported from either of two points which
lie on opposite sides of the center of mass. Show that if ω is the
angular frequency of Kater’s pendulum, then ω 2 = g/d where d is the
distance between the two points of support. (This type of pendulum
can be used to make very accurate measurements of the acceleration
 12.6. PROBLEMS 449

of gravity because one does not need an accurate determination of the

moment of inertia of the body.)
Problem 12.13. Obtain an expression for the period of oscillation
of a physical pendulum if the oscillations are not restricted to small
amplitudes.
Problem 12.14. A rod of mass m and length 2b lies on a smooth
horizontal table, as shown in Figure 12.13. At one end a smooth peg
A is fixed into the table. The side of the rod rests against the peg.
An impulse J is given to the rod at end B. Find the magnitude and
direction of the impulse given to the peg.

2b
B
A
J
Figure 12.13. A rod of mass m and length 2b lies on a
smooth horizontal table.

Problem 12.15. A conical pendulum has a bob of mass 200 grams

and a string of length 75 cm. The string will break if the tension exceeds
2.5 N. Determine the maximum angular velocity (φ̇) allowed.
Problem 12.16. A conical pendulum is perturbed such that it has
an initial amplitude in the θ direction of 0.01 rad. Determine how far
the azimuth advances during one oscillation in θ. Assume θ0 = 160◦ .
(Note: Do not set  equal to zero!)
Problem 12.17. Consider a spherical pendulum. Obtain an ex-
pression for the tension in the string as a function of E and θ. For
given values of E and pφ , determine the angle at which the string will
collapse.
Problem 12.18. A bead of mass m can slide freely on a circular
hoop of radius a. The hoop is upright (the plane of the hoop is vertical)
and it is rotating at a constant angular speed ω about the vertical
diameter. The position of the bead on the hoop is specified by the
angle θ which is measured from the bottom of the vertical diameter as
illustrated in Figure 12.14. (a) Obtain the Lagrangian for this system.
(b) Obtain the equation of motion and show that it reduces to the
equation of a simple pendulum when ω = 0. (c) Determine the value
of θ for which the bead is in equilibrium.
450 12. THE PENDULUM

Figure 12.14. A bead on a rotating hoop. Problem 12.18

Problem 12.19. Write the Hamiltonian and Hamilton’s equations

of motion for a spherical pendulum.
Problem 12.20. Two constants of the motion for the spherical
pendulum are the total energy E and the generalized momentum pφ .
Show that these first integrals lead to the following expressions:
Z
pφ dθ
φ= √ ,
2ml2 sin θ [E − Vef f (θ)]1/2
2

and r
ml2
Z
dθ
t= i1/2 .
2 h p2φ
E − 2ml2 sin2 θ − mgl cos θ
(The first of these integrals can be put into the form of an elliptic
integral of the third kind and the the second can be expressed as an
elliptic integral of the first kind.)
COMPUTATIONAL PROJECTS
Computational Project 12.1. Write a program to evaluate el-
liptic integrals of the first kind.
Computational Project 12.2. Use the results of Computational
Project 12.1 to write a program that evaluates the period of a simple
pendulum with arbitrary amplitude. Vary the amplitude and obtain a
plot of period as a function of amplitude.
Computational Project 12.3. Plot position vs. time for several
points on a rod undergoing an impulse, as illustrated in Figure 12.4.
Assume the impulse is J = 10 Ns, the rod has mass M = 2 kg, and
length L = 0.5 m.
Chapter 13
Accelerated Reference Frames

By now you have heard me say many times that Newton’s laws of
motion are applicable only in inertial (non-accelerated) frames of refer-
ence. The question arises: How do we deal with motion in non-inertial
reference frames? After all, most real reference frames are accelerating,
and for many of them, the acceleration cannot be neglected. In fact,
since we live on the surface of large rotating sphere, it is important for
us to be able to solve physics problems in non-inertial systems.
In this chapter you will learn how to express Newton’s second law
in a non-inertial (accelerating) reference frame. We are, of course, par-
ticularly interested in rotating coordinate systems, which are a prime
example of accelerating reference frames. Our study of these systems
will introduce you to “fictitious” forces such as the centrifugal force and
the Coriolis force. The Coriolis force, as you may know, is responsi-
ble for a number of important geophysical processes such as hurricanes
and ocean currents. Finally, we will consider the motion of a very long
simple pendulum and show that the plane of this “Foucault” pendulum
precesses due to the rotation of the Earth.

13.1. A Linearly Accelerating Reference Frame

Although the most interesting non-inertial reference frame for our
purposes is a rotating reference frame such as the Earth, I would like
to begin by looking at a reference frame that is accelerating linearly
relative to an inertial reference frame. Figure 13.1 shows a reference
frame O that is at rest “with respect to the fixed stars” and a second
reference frame, O0 , that is accelerating at a constant acceleration a
451
452 13. ACCELERATED REFERENCE FRAMES

relative to O.1 In the figure, the position of a particle of mass m with

respect to the inertial frame is rO and the position of the particle with
respect to the accelerating frame is rO0 . The position of the origin O0
with respect to O is r. By tip to tail addition of vectors,
rO = r + rO0 . (13.1)

ro ro'

O r O'

Figure 13.1. Coordinate system O0 is accelerating at a

constant rate a with respect to coordinate system O.

Since O is an inertial coordinate system, Newton’s second law holds

in that frame and we can write
mr̈O = F, (13.2)
where F is the force acting on the particle. The acceleration of the
particle as viewed from the accelerating frame is r̈O0 . The force on the
particle and the mass of the particle are assumed independent of the
reference frame (I am specifically excluding relativistic effects).
Taking the second time derivative of equation (13.1) yields

r̈O = r̈O0 + r̈,

where r̈ is the acceleration of reference frame O0 relative to the inertial
frame O. Newton’s law (13.2) then reads
F =mr̈O =m(r̈O0 +r̈) =mr̈O0 +mr̈.
Therefore, from the point of view of the observer in the accelerated
reference frame, Newton’s second law has the form
mr̈O0 = F−mr̈.
That is, there appears to be an additional force f = −mr̈ acting on the
particle. If asked to determine the force on the particle, the observer
in the accelerated reference frame would answer F + f .
Thus the observer in the accelerated system will believe that addi-
tional forces are acting on the particle. These are called fictitious forces
1Please ignore the annoying fact that eventually O0 will be moving at the speed
of light!
 13.2. A ROTATING COORDINATE FRAME 453

because they are really just mass times acceleration terms which have
been moved to the other side of the equation. Well known fictitious
forces include the centrifugal force and the Coriolis force.

13.2. A Rotating Coordinate Frame

If a body is at rest in some reference frame R0 but is accelerating
with respect to an inertial reference frame, then R0 is an accelerated
reference frame. Consider an ant at rest on the surface of a basketball.
If the ball is rotating, the ant is moving in a circle around the axis of
rotation (even though it may think it is at rest. You may think you
are sitting still in this room, but you are actually moving around in a
large circle at over 600 miles per hour.) Now an object moving in a
circle has a continuously changing velocity. Therefore, it is accelerat-
ing. The acceleration is directed towards the center of the circle. A
coordinate system attached to a rotating sphere is an accelerated coor-
dinate system. All points at rest relative to this coordinate system are
accelerating. In such a coordinate system there will be fictitious forces
related to the acceleration.
As a very important special case, consider a coordinate system
0 0 0
(x , y , z ) rigidly connected to the rotating Earth. Suppose the ori-
gin of this coordinate system is at the center of the Earth, as shown
in Figure 13.2. The Earth and the coordinate system rotate together
about the z 0 axis at an angular velocity Ω (assumed constant). Figure
13.2 also shows an inertial coordinate system (x, y, z) that is at rest
with respect to the fixed stars. (I am ignoring the motion of the Earth
around the Sun and the motion of the Sun around the center of the
Galaxy.)

z z'

Ω
y'
y

x x'

Figure 13.2. A rotating coordinate system (x0 , y 0 , z 0 )

fixed in the Earth and an inertial system (x, y, z) at rest
with respect to the fixed stars.
454 13. ACCELERATED REFERENCE FRAMES

The common origin of those two coordinate systems is the center of

the Earth. The z and z 0 axes coincide and lie along the axis of rotation
of the Earth. (You can imagine these axes pointing from the center of
the Earth out through the North Pole.) The x, y and x0 , y 0 axes all lie
in the equatorial plane. Again, the coordinate axes (x, y, z) are fixed
in space and the coordinate axes (x0 , y 0 , z 0 ) are rigidly attached to the
Earth and rotating with it.
A person standing still on the surface of the Earth is at rest with
respect to the rotating coordinate system. The position of this person
is given by the vector r from the origin to the point P where the person
is standing. Obviously, the components of r can be specified in either
the fixed or the rotating coordinate system. Since P is at rest in the
rotating (Earth or primed) system,
 
dr
= 0.
dt rot
On the other hand, according to an observer in an inertial system, the
point P is moving with angular velocity Ω. The linear velocity of the
point in the inertial system is
 
dr
= Ω × r.
dt inertial
Next assume the person has a velocity vr with respect to the Earth, so
that  
dr
= vr .
dt rot
Then, from the point of view of the inertial frame, the person’s velocity
is  
dr
= vr + Ω × r.
dt inertial
This last equation can be written
   
dr dr
= + Ω × r.
dt inertial dt rot
It is easy to generalize this relationship to the time derivative of any
vector U. The relation between the rate of change of U in the inertial
system and the rate of change of U in the rotating system is
   
dU dU
= + Ω × U. (13.3)
dt inertial dt rot
 13.3. FICTITIOUS FORCES 455

This can be expressed as an operator equation

   
d d
= + Ω× (13.4)
dt inertial dt rot
where it is understood that the various terms are all operating on the
same vector.

Exercise 13.1. Somewhere out in space there is an inertial refer-

ence frame and (strange to say) in it there is a merry-go-round that is
rotating at angular velocity Ω. A child is sitting on a wooden horse a
distance R from the axis of the merry-go-round. (a) What is the veloc-
ity of the child in the rotating frame? (b) What is the velocity of the
child in the inertial frame? (c) What is the acceleration of the child in
the rotating frame? (d) Determine the acceleration of the child in the
inertial frame by elementary methods, that is, by using the definition
of centripetal acceleration. (e) Determine the acceleration of the child
in the inertial frame using equation (13.3). Answers (a) zero (b) ΩRφ̂
(c) zero (d) −Ω2 Rr̂.
Exercise 13.2. Prove that the rate of change of the angular ve-
locity of the rotating system is the same in both the inertial and the
rotating reference frame.

13.3. Fictitious Forces

Let us continue ignoring the motion of the Earth around the Sun
and imagine the Earth to be a rotating sphere isolated in space. It is
rotating at a nearly constant angular velocity of 2π radians/day. The
relationship between the time rate of change of the position vector (r)
of a particle in the inertial reference frame and its time rate of change
in the rotating reference frame is, by equation (13.3),
   
dr dr
= + Ω × r.
dt inertial dt rot
The velocity of the particle with respect to inertial space is dr
 
dt inertial
.
 dr 
Call it vi . Also, dt rot is the velocity of the particle with respect to
the rotating frame. Call it vr . Consequently,
vi = vr + Ω × r.
456 13. ACCELERATED REFERENCE FRAMES

This equation indicates that if a person on Earth perceives a particle

to be at rest, an observer in inertial space will perceive the particle to
have a velocity vi = Ω × r.
Acceleration is defined as the time rate of change of velocity. Hence,
 
dvi
ai = .
dt i
This is the acceleration in the inertial reference frame and is, of
course, the acceleration in Newton’s second law. However, we live in a
rotating reference frame, so we want to express the second law in terms
of quantities we can measure, that is, ar and vr .
Applying operator equation (13.4) yields
   
d d
ai = (vr + Ω × r) = (vr + Ω × r) + Ω × (vr + Ω × r),
dt i dt rot
   
d d
= vr + (Ω × r) + Ω × vr + Ω × (Ω × r),
dt r dt rot
 
dΩ dr
= ar + × r + (Ω× ) + Ω × vr + Ω × (Ω × r).
dt dt rot
dΩ
Ifdrthe
 Earth’s rotation rate is assumed constant, dt
= 0. Noting that
dt rot
= vr , the inertial acceleration is given by

ai = ar + 2Ω × vr + Ω × (Ω × r).
Newton’s second law is valid in the inertial frame, so ai = F/m where F
is the net external (“physical”) force acting on the particle. Replacing
ai and rearranging leads to

ar = F/m − 2Ω × vr − Ω × (Ω × r). (13.5)

This equation shows that the acceleration as measured in the rotating
reference frame will depend not only on the force acting on the particle,
but also on two other terms. These additional terms are a consequence
of considering the motion in a non-inertial reference frame.
One often writes equation (13.5) in the suggestive form
F − 2mΩ × vr − m [Ω × (Ω × r)] =mar . (13.6)
This equation looks like F = ma. The additional terms on the left hand
side of the equation have the units of force and they are referred to as
“fictitious forces.” You should be aware, however, that they really are
just mass times acceleration terms that have been brought over from
the other side of the equation. A “real” force is an interaction between
 13.3. FICTITIOUS FORCES 457

bodies. A “fictitious” force is a consequence of the acceleration of the

coordinate system.
The term
−2mΩ × vr ,
is called the “Coriolis force” and the term
−mΩ × (Ω × r),
is called the “centrifugal force”. The reason for the name is that this
“force” is directed away from axis of rotation, as illustrated in Figure
13.3

Ωxr
Ω -Ω x Ω x r
r

Figure 13.3. The centrifugal acceleration, −Ω×(Ω×r),

points away from the axis of rotation.

Worked Example 13.1. An old fashioned record player is

rotating with an angular velocity Ω0 =Ω0 k̂. A bug is crawling with
constant speed vb relative to the record along a groove in the record,
a distance r from the axis of rotation. Assume the bug is moving
in a circular path, so there is no radial component to its velocity.
(a) Determine the velocity of the bug relative to the room (that is
the “inertial space” ). (b) Determine the acceleration of the bug by
elementary means, using the definition of centripetal acceleration.
(c) Determine the acceleration of the bug using Equation (13.4)
and compare your answer to part (b). (Comment: This example
will help you learn the mechanics of using Equation (13.4). In
particular, you will appreciate that the same vector must be used
on
 dv both sides of that equation. That is, if you are calculating
you get the wrong answer if you put dvdtrot rot on the
 
i
dt inertial
other side. This is obvious, but it is an easy error to make.)
458 13. ACCELERATED REFERENCE FRAMES

Solution: Let ρ̂, φ̂, k̂ be cylindrical unit vectors in the rotating

frame. (a) The velocity of the bug in the rotating coordinate frame
is  
dr
= v0 φ̂.
dt rot
The velocity of the bug in inertial space by Equation (13.4) is
   
dr dr
= + Ω0 ×r
dt inertial dt rot
vi = vrot + Ω0 ×r
= vb φ̂ + Ωr(k̂ × ρ̂)
= vb φ̂ + Ωrφ̂.
For convenience, let us define Ωb = vb r and write
vi = (Ωb + Ω0 )rφ̂.
(b) By the definition of centripetal acceleration
vi2 −ρ̂
ai = (−ρ̂) = ((Ωb + Ω0 )r)2
r r
= −ρ̂r(Ωb + Ω0 )2 .
(c) Using Equation (13.4) to determine ai we have
   
dvi d
ai = = (Ωb + Ω0 )rφ̂
dt in dt in
 
d
= (Ωb + Ω0 )rφ̂ + Ω0 × (Ωb + Ω0 )rφ̂.
dt rot

Since vr = Ωb r and since the rate of change of φ̂h in ithe rotating

frame is only due to the motion of the bug, we have ddtφ̂ = −Ωb ρ̂.
rot
Furthermore, keeping in mind that | r |= r = constant, we obtain
ai = (Ωb + Ω0 )r(−Ωb )ρ̂ + Ω0 (Ω0 + Ωb )r(k̂ × φ̂)
= −ρ̂r [(Ωb + Ω0 )Ωb + Ω0 (Ωb + Ω0 )]
= −ρ̂r Ω2b + 2Ωb Ω0 + Ω20
 

= −ρ̂r(Ωb + Ω0 )2 ,
in agreement with the result of part (b).

13.4. Centrifugal Force and the Plumb Bob

Assume the Earth is a sphere. Most people probably assume that
a hanging plumb bob points directly towards the center of the Earth.
 13.4. CENTRIFUGAL FORCE AND THE PLUMB BOB 459

However, that would only be true if the Earth were not rotating. On
a rotating Earth, the centrifugal force causes the plumb bob to be de-
flected slightly away from the line to the center. The proof is fairly
simple. You may want to try it on your own before reading my deriva-
tion.
As shown in Figure 13.4, the “real” forces acting on the bob are
the tension in the string and the gravitational attraction of the Earth.
That is,
ME m
F = T−G 2 r̂,
RE
where RE is the radius of Earth. Inserting this expression for the force
into the “F = ma” equation in form (13.6) yields

ME m
T−G 2
r̂−m [Ω × (Ω × r)] =mar = 0,
RE
Here I used the fact that the velocity and acceleration of the plumb
bob in the rotating frame are zero.

Ω T
ac
fc
Fg g
resultant = ge

ac

Figure 13.4. A plumb bob in the Northern Hemisphere

points below the center of a spherical Earth. The sketch
on the left shows the “forces” acting on the bob. The
vector fc represents the centrifugal force. The sketch on
the right illustrates the vector sum of g and
ac = −Ω × (Ω × r).

Recall that the acceleration due to gravity at the surface of the

Earth is
ME
g = −G 2 r̂.
RE
For a spherical Earth this points towards the center of the Earth. The
equation for the plumb bob is, then,
T+mg−m [Ω × (Ω × r)] = 0.
460 13. ACCELERATED REFERENCE FRAMES

This relation indicates that the tension is not parallel to mg and thus
it is not directed along the line to the center of the Earth.
It is usually convenient to include the centrifugal acceleration into
the gravitational acceleration. The “effective” gravitational accelera-
tion is defined by
ge = g − Ω × (Ω × r),
and the equation reduces to
T+mge = 0.
Therefore the tension is directed opposite to ge rather than opposite to
g. See Figure 13.4. In the Northern hemisphere ge = g − Ω × (Ω × r)
points slightly below the center of the Earth while in the Southern
hemisphere it points somewhat above the center.
The centrifugal force is responsible for the fact that the Earth is not
a sphere; it has a “bulge” at the equator. To see how this comes about,
consider a particle on the surface of a perfectly spherical Earth. This
particle is subjected to the gravitational force, a normal force and the
centrifugal force. If you draw a force diagram for this situation, you
will find that the centrifugal force is an unbalanced force. Resolving
the centrifugal force into components parallel and perpendicular to the
Earth’s surface, you will find that the perpendicular component simply
adds to the normal force, but the unbalanced parallel component points
toward the equator. Thus, the centrifugal force is responsible for the
fact that the Earth is an oblate spheroid rather than a sphere. You
can easily convince yourself that if the Earth’s equatorial bulge is taken
into consideration there will be no unbalanced forces on the particle on
the surface. This argument also tells you that although a plumb line
does not point towards the center of the Earth, it is perpendicular to
the Earth’s surface, so it does define a local perpendicular.

Exercise 13.3. Draw the force diagram for a particle on the surface
of a perfectly smooth, spherical, rotating planet and show that the
particle accelerates towards the equator. Do this for both the Northern
and Southern hemispheres. Draw the same sort of sketch for a planet
with an equatorial bulge and demonstrate that in this case there is no
net force on the particle.
 13.5. THE CORIOLIS FORCE 461

13.5. The Coriolis Force

It is well known (at least to physicists) that a projectile fired near
the surface of the Earth will veer towards the right in the Northern
Hemisphere and towards the left in the Southern Hemisphere. This is
due to the so-called Coriolis force, which is also responsible for oceanic
circulation, cyclones, hurricanes and other geophysical phenomena.
The Coriolis force is the term −2mΩ × vr in equation (13.6). Note
that the direction of this fictitious force depends on the direction and
sense of the velocity vector.
When studying the effect of the Coriolis force on objects moving
near the surface of the Earth it is convenient to introduce a Cartesian
coordinate system in which the z axis points up from the surface (i.e.,
opposite to the direction of the effective gravitational acceleration ge ),
the x axis points South and the y axis points East. See Figure 13.5.
Note that the angular velocity of the Earth (Ω) has components along
the −x axis (North) and along the +z axis (up). Thus:
Ω = Ω cos λk̂ − Ω sin λı̂,
where λ is the “colatitude.”
Ω z
z
Ω

λ
λ
x

Figure 13.5. A reference frame fixed to the rotating

Earth. The z axis is along the local vertical and the x
axis points South. The angle λ is the co-latitude.

Exercise 13.4. Draw force diagrams to determine the direction of

the Coriolis force on the following objects. (a) A falling stone. (b) A
projectile fired due North in the Northern hemisphere. (c) A projectile
fired due East in the Northern hemisphere. (d) A projectile fired due
East in the Southern hemisphere.
462 13. ACCELERATED REFERENCE FRAMES

13.5.1. A Falling Body. An object of mass m is dropped from

a height h. Where does it land? It turns out that the object hits the
ground a tiny bit to the East of the spot directly below the point where
it was released. The reason is that the falling object has a downward
velocity so there is a Coriolis force acting on it, given by −2mΩ × vr .
(In the rotating reference frame of the Earth, the forces acting on the
particle are the gravitational force, the Coriolis force and the centrifugal
force.)
For simplicity, we incorporate the centrifugal acceleration into g
and write ge for the sum of these two terms. The equation of motion
of the falling body is, then
mar = mge − 2mΩ × vr .
If you think about it, you will soon realize we cannot solve this equation
for ar because we do not know vr , and we cannot determine vr until we
solve this equation! It would seem that we are faced with an unsolvable
problem.
The way out of this dilemma is to use the fact that the Coriolis
acceleration is much smaller than the gravitational acceleration, so vr
is almost (but not quite) equal to the velocity of the falling body in the
absence of the Coriolis force. That is, the Coriolis effect can be treated
as a perturbation. The technique is as follows: First solve the unper-
turbed problem and determine the unperturbed velocity and time for
the object to fall. This is called the “zeroth order solution.” Then use
the unperturbed velocity in the expression for the Coriolis acceleration
to obtain the “first order solution.” This process can be repeated as
many times as desired to get better and better approximations to the
correct answer, but usually the first or second order solution is quite
adequate.2
Let us begin by solving the problem for the unperturbed velocity and
position of the particle as a function of time (the zeroth order solution).
The only force acting on the body is the (effective) gravitational force.
Consequently,
ar = −ge k̂.
Integrating once,
vr = −ge tk̂,
and integrating again,
z = hk̂ − 12 ge t2 k̂.

2This technique is known as the method of successive approximations.

 13.5. THE CORIOLIS FORCE 463

The time required for the particle to reach the ground, tf , is the value
of t when z = 0, that is,
p
tf = 2h/ge .
Note that none of the above relations include the Coriolis effect.
Now include the Coriolis acceleration, given by:
ac = −2Ω × vr .
You cannot evaluate ac exactly because you do not know the value of vr
which itself depends on the Coriolis acceleration. However, you know
the zeroth order approximation to the velocity. It is vr = − ge tk̂. The
first order correction is obtained by including the Coriolis term in the
acceleration but using the zeroth order velocity. That is, a =ge k̂ + ac
where

ac = −2Ω × vr
h i h i
= −2 Ω cos λk̂−Ω sin λı̂ × −ge tk̂
= 2ge tΩ sin λ̂.
Note that for the falling body, the Coriolis acceleration is in the positive
y direction, that is, towards the East. The eastward component of the
velocity (vy ) is initially zero and at any later time it is obtained by
integrating the expression above. Therefore,
vy = ge t2 Ω sin λ.
The eastward deflection (y) is obtained by integrating again.
1
y = ge t3 Ω sin λ.
3
To determine the total eastward deflection whenp the object is dropped
from height h, substitute the time of flight (t = 2h/ge ) into this last
expression.
The results obtained are the first order solution. As mentioned
above, these results can be plugged back into the equation of motion
to obtain the second order solution, and so on.
Students are often mystified by this result. The eastward deflection
is obviously not due to the Earth rotating under the falling body - any-
way, that would give a westward deflection! A physical explanation for
the horizontal (eastward) deflection of a dropped body comes from the
conservation of angular momentum. Let l be the angular momentum
of the particle in the inertial frame. There is no torque acting on it, so
l is constant. Initially the object is at position r (where r is a vector
464 13. ACCELERATED REFERENCE FRAMES

from the center of the Earth). In the inertial frame the body has an
eastward linear velocity v. (Since the body is at rest with respect to
.
the surface of the Earth, this initial velocity is v = Ω × r =ΩRe sin λ̂.)
The angular momentum of the particle is
l =mr × v.
For l to remain constant while r decreases (as the particle falls), the
velocity v must increase. Therefore, in the Earth based reference frame,
the object acquires a velocity towards the East.
Another strange and interesting effect of Coriolis forces is the fact
that if you throw an object straight up, it does not land at the point
from which it started. (You will prove this in problem 13.6.)

Worked Example 13.2. To first order a falling particle is

not deflected towards the South (x-direction). However, there is a
southward deflection in the second order approximation. Estimate
this deviation and compare to the first order deviation towards the
East.
Solution: To first order the velocity is
vr =vz k̂+vy ̂ = −ge tk̂+ge t2 Ω sin λ̂
The Coriolis acceleration is
ac = −2Ω × vr
h i h i
= −2 Ω cos λk̂−Ω sin λı̂ × −ge tk̂+ge t2 Ω sin λ̂
That is,
ax = 2 Ω2 cos λ sin λge t2

ay = 2Ω sin λge t
az = 2Ω2 sin2 λge t2 − ge
Consequently,
Z
2
vx = ax dt = Ω2 cos λ sin λge t3
3
and
1
x = Ω2 cos λ sin λge t4
6
Using the first order value for the time of flight we obtain
2 h2
x = Ω2 cos λ sin λ .
3 ge
 13.5. THE CORIOLIS FORCE 465

Note that the time of flight will actually be slightly longer because
the Coriolis acceleration opposes the gravitational acceleration.
Comparing the southward and eastward deflections we have
 x  (2/3)Ω2 cos λ sin λ(h2 /ge )
 
 = Ω cos λ
= p .
 −y  (1/3)Ω sin λ(2hge )1/2 2ge /h
Since Ω is about 10−5 and cos λ is near unity and the denominator
will be in the range of 1 to 10, we appreciate that the southward
deflection is significantly smaller than the eastward deflection.
Exercise 13.5. A stone is dropped from a height of 50 meters.
How far is it deflected towards the East? Assume the latitude is 60◦
N. Answer: 0.39 cm.

13.5.2. A Projectile. To determine the motion of a projectile

fired from the surface of the Earth with arbitrary initial speed and
direction, let us return to the equation for the acceleration in a rotating
coordinate frame. A slight modification of equation (13.5) gives
ar = g−2(Ω × v).
I dropped the subscript “e” on g, but keep in mind that the centrifugal
force is incorporated into g. The velocity v at any time can be expressed
as
v =vx ı̂+vy ̂+vz k̂,
and
Ω = −Ω sin λı̂ + Ω cos λk̂.
Hence,
ar = g + 2vy Ω cos λı̂−2(vx Ω cos λ + vz Ω sin λ)̂ + 2vy Ω sin λk̂ (13.7)
or
ax = 2vy Ω cos λ,
ay = −2vx Ω cos λ − 2vz Ω sin λ,
az = −g+2vy Ω sin λ.
Given these three components of the acceleration, you could determine
the motion of the projectile by integrating twice if you knew the veloc-
ity components. Once again, you obtain the first order approximation
by using the velocity components of the zeroth order approximation,
and similarly calculating the higher order approximations. However,
even without doing any calculations, the functional dependence of the
466 13. ACCELERATED REFERENCE FRAMES

acceleration components give a good qualitative idea of the motion.

For example, in the Northern Hemisphere, if the projectile is fired due
North, then vy (the eastward component of the velocity) is zero and
there is no Coriolis acceleration in the x or z directions. There is, how-
ever, an eastward component of the acceleration and the projectile will
be deflected towards the East (to the right of its direction of motion).
To actually evaluate this deflection you must incorporate the zeroth
order approximations of vx and vz into the expression for ay and then
integrate.
Similarly, the expressions for the acceleration components indicate
that a projectile fired due East is deflected towards the South. In this
case, the velocity component vy affects az and therefore changes the
time of flight of the projectile resulting in an increase or decrease in the
range. Furthermore, ay is non-zero, affecting the range of the projectile.
It is interesting to note that the vertical velocity vz is positive during
the first part of the trajectory and negative during the second part, but
you can not assume that the second term in ay averages out to zero.
In solving for the motion of a projectile, you must use perturba-
tion techniques, but fortunately a reasonably accurate solution is usu-
ally given by the lowest order approximation that includes the Coriolis
force.
You should convince yourself that in the Northern hemisphere a
projectile will be deflected toward its right and in the Southern hemi-
sphere it will be deflected toward its left.

Worked Example 13.3. A large cannon is located at colat-

itude λ in the Northern Hemisphere. It fires a shell with velocity
v0 . The cannon is aimed at an angle φ East of North and has an
elevation angle θ. (a) Write expressions for the unperturbed ve-
locity components as a function of time. (Call them v0x , v0y , v0z .)
(b) Write expressions for ax , ay , az in terms of the unperturbed ve-
locity components. (c) Obtain first order expressions for vx , vy , vz
as functions of time and the unperturbed values of the velocity.
Ignore air resistance. (The results of this example will be found
useful when solving the problems at the end of the chapter.)
 13.6. THE FOUCAULT PENDULUM 467

Solution: (a) From elementary considerations the velocity com-

ponents in the absence of Coriolis forces
vx (t) = −v0 cos θ cos φ,
vy (t) = v0 cos θ sin φ,
vz (t) = v0 sin θ − gt.
(b) The Coriolis acceleration is
h i h i
ac = −2Ω × v = −2 −Ω sin λı̂ + Ω cos λk̂ × vx ı̂ + vy ̂ + vz k̂ ,
h i
= −2 −vy Ω sin λk̂+vz Ω sin λ̂ + vx Ω cos λ̂ − vy Ω cos λı̂ ,
or
ax = 2v0 Ω cos λ cos θ sin φ,
ay = 2v0 Ω cos λ cos θ cos φ − 2Ω sin λ(v0 sin θ − gt),
az = 2v0 Ω sin λ cos θ sin φ − g.
(c) The velocity is obtained by integrating the acceleration com-
ponents to obtain:
Z
vx = v0x + ax dt = −v0 cos θ cos φ + (2v0 Ω cos λ cos θ sin φ) t,
Z
vy = v0y + ay dt
= v0 cos θ sin φ + (2v0 Ω cos λ cos θ cos φ) t
− (2v0 Ω sin λ sin θ) t + gt2 Ω sin λ,
Z
vz = v0z + az dt = v0 sin θ − gt + (2v0 Ω sin λ cos θ sin φ) t.

Exercise 13.6. Carry out the operation Ω × v to verify equation

13.7.
Exercise 13.7. A projectile located at 55◦ S is fired due East with
an initial velocity of 300 m/s at an angle 25◦ above the horizontal. Eval-
uate the initial values of the components of the Coriolis acceleration.
Answer: ax = −3.24 × 10−2 m/s2 .

13.6. The Foucault Pendulum

As a final example of the effect of a rotating coordinate system
on the dynamics of a physical system, let us consider the Foucault
468 13. ACCELERATED REFERENCE FRAMES

pendulum. This pendulum consists of a massive bob suspended by a

very long string or thin wire. Foucault pendulums range from ten to
sixty meters in length. When set in motion the bob swings back and
forth and the plane of motion of the pendulum slowly precesses. The
precession rate is Ω cos λ, where Ω is the angular velocity of the Earth
(2π rad/day) and λ is the colatitude. A long pendulum goes through
many oscillations before the motion is damped out, and the precession
is quite noticeable. When Foucault suspended a pendulum inside the
Pantheon in Paris great crowds appeared to “observe the rotation of
the Earth.”
The behavior of the Foucault pendulum is easy to understand if
you imagine a pendulum suspended above the North Pole. The Earth
rotates under the pendulum and an observer standing on the surface
sees the plane of motion precessing at a rate of one revolution per day.
On the other hand, at the equator, a pendulum does not precess at
all. The motion at an arbitrary latitude is not intuitively obvious and
requires an analysis to understand it.

θ
T Path of pendulum bob

y
ψ
vh
x
mg
Projection of path onto xy plane

Figure 13.6. The Foucault pendulum. Note that x

points South as in Figure 13.5

Consider a long pendulum at an arbitrary latitude. Let the z axis

be oriented along the local vertical. See Figure 13.6 which shows the
projection of the path of the pendulum onto the xy plane. The pro-
jected path makes an angle ψ with the x axis. The precession of the
pendulum causes ψ to change as a function of time. The basic problem
consists in demonstrating that the rate of change of ψ is
dψ
= Ω cos λ.
dt
 13.6. THE FOUCAULT PENDULUM 469

The pendulum bob is acted upon by two real forces: The tension in
the string (T) and gravity (mg). It is also acted upon by two fictitious
forces: the Coriolis force (−2mΩ × v) and the centrifugal force. For
convenience assume the centrifugal force is absorbed into mg.
Therefore,
ma = T+mg−2m(Ω × v).
In component form this relationship becomes
max ı̂+may ̂+maz k̂ = −T sin θ cos ψı̂−T sin θ sin ψ̂ + T cos θk̂−mg k̂
h i h i
−2m Ω cos λk̂ − Ω sin λı̂ × vx ı̂+vy ̂+vz k̂ .
That is,

−T sin θ cos ψ + 2mΩ cos λvy = max , (13.8)

−T sin θ sin ψ − 2mΩ cos λvx − 2mΩ sin λvz = may , (13.9)
T cos θ − mg + 2mΩ sin λvy = maz . (13.10)
Consider equations (13.8) and (13.9). Since the vertical displace-
ment is small, the vertical velocity vz is much smaller than the horizon-
tal velocity. Therefore, we are justified in ignoring the term containing
vz in equation (13.9), leading to the set of coupled differential equations
−T sin θ cos ψ + 2mΩ cos λvy = max ,
−T sin θ sin ψ − 2mΩ cos λvx = may .
The pendulum is very long so the angle θ is quite small and the string
is essentially vertical at all times (Figure 13.6 is significantly out of
scale). To a high degree of accuracy the tension in the string is equal
to mg. Furthermore, x = l sin θ cos ψ and y = l sin θ sin ψ. Therefore,
the coupled equations above can be formulated as:
g 
− x + 2vy Ω cos λ = ax ,
 gl 
− y − 2vx Ω cos λ = ay .
l
Noting that g/l = ω 2 where ω is the oscillatory frequency of the pen-
dulum, and letting K = Ω cos λ, these equations become
ẍ − 2K ẏ + ω 2 x = 0, (13.11)
ÿ + 2K ẋ + ω 2 y = 0.
The problem now is to solve these coupled linear differential equations.
I will do this in three ways. First I will use a “trick” then I will solve
the equations in a straightforward, but more tedious manner. Then, in
Worked Example 13.6, I will show you a third way to find the solution.
470 13. ACCELERATED REFERENCE FRAMES

The reason for doing this is because I want you to see several techniques
for solving coupled differential equations.
A Neat Trick
The set of coupled equations (13.11) can be solved very nicely by
introducing the complex quantity
ζ = x + iy.
√
Multiplying the second equation by i (= −1 ) and adding the two
equations gives
ζ̈ + 2iK ζ̇ + ω 2 ζ = 0,
which resembles the equation of motion for a damped harmonic oscilla-
tor. This linear differential equation is solved in the usual way, assum-
ing ζ = Aept and obtaining the “secular” or “characteristic” equation

p2 + 2iKp + ω 2 = 0,
with solution √
p = −iK ± i K 2 + ω 2 .
Now p K = Ω cos λ where Ω is the rotation rate of the Earth, whereas
ω = g/l is the angular frequency of the pendulum. So K << ω and
the K 2 under the root can be neglected. There are two possible values
for p, so the solution is
ζ = Ae−i(K+ω)t + Be−i(K−w)t (13.12)
where A and B are complex constants. Writing the complex constants
in the form (a + ib) and (c + id) the solution is
ζ(t) = (a + ib)e−i(K+ω)t + (c + id)e−i(K−w)t . (13.13)
This complex solution is expressed in terms of real quantities, but it
is still not obvious what the motion looks like. The following example
will help you to visualize the motion described by equation (13.13).

Worked Example 13.4. At time t = 0 the Foucault pendu-

lum is released from rest at position x =A, y = 0. Obtain expres-
sions for x(t) and y(t).
 13.6. THE FOUCAULT PENDULUM 471

Solution: Let us write the constants in Equation (13.12) as

1
2
Ce±iθ ,
thus
1 −iθ −i(ω+K)t 1 iθ i(ω−K)
ζ = Ce e + Ce e t
2 2
1 −iKt  −i(ωt+θ)
+ e+i(ωt+θ)

= Ce e
2
1
= C [cos Kt − i sin Kt] [2 cos(ωt + θ)]
2
= C [cos Kt cos(ωt + θ) − i sin Kt cos(ωt + θ)]
Since x = Re(ζ) and y = Im(ζ) we conclude that
x = C cos(tΩ cos λ) cos(ωt + θ),
y = −C sin(tΩ cos λ) cos(ωt + θ).
The constants C and θ are now easily obtained from the initial
conditions, yielding C = A and θ = 0, so
x(t) = A cos ωt cos(tΩ cos λ),
y(t) = −A cos ωt sin(tΩ cos λ).
The motion is a rapid oscillation of frequency ω, modulated by a
low frequency variation in amplitude (Ω cos λ). The modulated x
and y oscillations have time varying amplitudes that are 180◦ out
of phase. The resultant motion is a rotation of the plane of motion
with a frequency Ω cos λ. This is illustrated in Figure 13.7

Figure 13.7. Amplitudes of oscillations in the x and y

directions for the Foucault pendulum.

Exercise 13.8. A Foucault pendulum is observed to have a ve-

locity v0 in the +x direction at the instant it passes through the ori-
gin. Obtain expressions for x and y as functions of t. The moment
472 13. ACCELERATED REFERENCE FRAMES

the pendulum passed through the origin is taken as t = 0. Answer:

x = Re(ζ) = − vK0 sin ωt cos Kt, y = Im(ζ) = vK0 sin ωt sin Kt.

A More General Way to Solve the Problem. Return again

to equations (13.11). They are a pair of coupled second order linear
differential equations whose solution is desired. By this time you are
very familiar with the technique of assuming a solution of the form Aept .
In this case there are two unknowns and the procedure is to assume
two solutions, but to let both of them have the same time dependence.3
In other words, assume the solutions
x = Aept ,
y = Bept .
Plugging these solutions into (13.11) leads to the following set of
secular equations
p2 A − 2KpB + ω 2 A = 0,
p2 B + 2KpA + ω 2 B = 0.
or
A(p2 + ω 2 ) − B(2Kp) = 0,
A(2Kp) + B(p2 + ω 2 ) = 0.
This is a system of homogeneous linear algebraic equations in the
unknowns A and B. Recall that the condition for a non-trivial solution
is  2 
 p + ω 2 −2Kp 
  = 0.
 2Kp p2 + ω 2 
This leads to √
p = −iK ± i K 2 + ω 2 . (13.14)
The rest of the solution is straightforward.4

3You probably remember that this is the procedure we used to solve the coupled
oscillators problem. (See section 11.5.)
4It is interesting to note that the Foucault pendulum problem has many par-
allels to a seemingly completely unrelated problem, namely the violation of CP
symmetry in the decay of an elementary particle called the neutral kaon. This
illustrates the basic unity of physics. To explore further, see the paper “Classical
Illustration of CP Violation in Kaon Decays” by Jonathan L. Rosner and Scott A.
Slezak, Am. J. Phys., 69, 44-49, 2001.
 13.6. THE FOUCAULT PENDULUM 473

Worked Example 13.5. Solve for the precession rate of a

Foucault pendulum by considering the problem in a rotating refer-
ence frame in which the pendulum does not precess.
Solution: Using a rotating reference frame makes the problem
easier to solve. Imagine looking down on the pendulum from a co-
ordinate system rotating at an angular velocity K = Ω cos λ, equal
to the precession rate. In this coordinate frame the bob moves
back and forth in a straight line. Let the axes in this coordinate
frame be x∗ and y ∗ and define the x∗ axis to lie along the path of
the bob. Then y ∗ is always zero. Denoting the distance from the
origin to the bob along the x∗ axis by s, Figure 13.8 shows that
the Earth based coordinates x and y are
x = s cos Kt
and
y = −s sin Kt.
The sum of the equations of motion (equations 13.11) is
ẍ + ÿ + 2K(ẋ − ẏ) + ω 2 (x + y) = 0 (13.15)
Plugging in the expressions for x and y in terms of s gives
s̈ + (K 2 + ω 2 )s = 0. (13.16)
Again, K << ω so, to an excellent approximation,
s̈ + ω 2 s = 0.
This is just the SHM equation. Thus, in the rotating coordinate
system the Foucault pendulum simply oscillates back and forth, or,
from the point of view of the observer on the surface of the Earth,
the plane of the pendulum rotates at a constant rate K = Ω cos λ.

Exercise 13.9. Fill in the missing steps between equations (13.15)

and (13.16).
Exercise 13.10. Show that equation (13.14) leads to the same
result as obtained in Worked Example 13.5.
Exercise 13.11. A 20 m long Foucault pendulum is mounted at
San Jose, California (latitude 37.33◦ ). Determine the precession rate
of its plane of motion. Answer: 9.1◦ /hour.
474 13. ACCELERATED REFERENCE FRAMES

y*
s cos(Kt)
y x

s sin(Kt)
Kt
s x*

Figure 13.8. The starred coordinate system rotates at

an angular velocity K relative to the unstarred system.
The motion of the pendulum bob is along the x∗ axis.

13.7. Summary
A rotating reference frame is an accelerated reference frame. The
time derivative of a vector in a rotating frame is related to the time
derivative in an inertial frame by the operator equation
   
d d
= +Ω×.
dt inertial dt rot
The acceleration ar relative to a rotating frame is obtained by applying
the operator equation twice to the position vector, yielding:

ar = F/m − 2Ω × vr − Ω × (Ω × r),
that is often written as

F − 2mΩ × vr − m [Ω × (Ω × r)] =mar .

The term −2mΩ × vr is called the Coriolis force and −m [Ω × (Ω × r)]

is called the centrifugal force.
The centrifugal force causes a plumb bob to point slightly away
from the center of the Earth.
The Coriolis force causes a falling body to be deflected towards the
East and a projectile to be deflected to the right of its initial velocity.
Solving for the motion under the action of the Coriolis force usually
involves applying the method of successive approximations.
The Foucault pendulum is a long simple pendulum which swings
for a long enough time to allow one to appreciate the precession of the
plane of oscillation at Ω cos λ.
 13.8. PROBLEMS 475

13.8. Problems
Problem 13.1. A child runs on a merry-go-round with velocity V
with respect to the merry-go-round. Show that
 
dr
= V + (ωe +ω mgr ) × r
dt I
where ωe is the angular velocity of Earth and ωmgr is the angular ve-
locity of the merry-go-round.
Problem 13.2. A wedge of mass M and angle α is resting on a
frictionless horizontal plane. A rectangular box of mass m is on the
wedge. Since all surfaces are frictionless, the box will slide down the
wedge and the wedge will slide in the opposite direction on the plane.
So far, this is the same as Problem 4.7 which you may have solved.
There you used the Lagrangian technique to solve the problem and
found that the horizontal acceleration of the wedge is
!
m g sin α
Ẍ = − m cos α.
m + M 1 − m+M cos2 α
Here you are asked to find the horizontal component of the accelera-
tion of the box relative to the wedge. Solve the problem by applying
Newton’s second law directly, but using the fact that the wedge is ac-
celerating so the motion of the box relative to the wedge is motion in
an accelerated reference frame.
Problem 13.3. Obtain an equation analogous to equation (13.5)
but including the orbital motion of the Earth around the Sun. You may
assume the two rotation vectors are constant and parallel and that the
Earth’s orbit is circular. Compare the magnitude of any additional
term to the Coriolis term in equation (13.5).
Problem 13.4. Assume the Earth is an inertial coordinate system.
A child on a rotating platform runs (a) radially towards the edge. (b)
Runs in the azimuthal direction. Determine the fictitious forces acting
on the child in both cases.
Problem 13.5. A wheel of radius a rolls at constant speed V in
a circular path of radius R. Assume the wheel is perpendicular to the
plane of the path and that the surface of Earth is an inertial reference
frame. Consider a point at the top of the wheel. Determine the Coriolis
acceleration, and the centrifugal acceleration this point .
Problem 13.6. A particle is projected vertically upward with an
initial velocity v0 . Determine, to first order, the point where it lands
relative to the point where it was thrown.
476 13. ACCELERATED REFERENCE FRAMES

Problem 13.7. A NASA research center in Cleveland, Ohio has an

underground vacuum drop tank in which an object can fall 132 meters.
The latitude of Cleveland is 41.5◦ N. Determine the Coriolis deflection
to second order.
Problem 13.8. A 5 kg stone is dropped from a height of 100 me-
ters above the surface of the Earth at a location where the latitude
is 40◦ . Assume the radius of the Earth is 6378 km. (a) What is the
linear velocity of the stone just before it is dropped? (b) What is its
angular momentum at that moment? (c) Determine the difference in
its eastward velocity by the time it hits the ground by noting that the
angular momentum of the stone does not change.
Problem 13.9. A projectile is fired due East with an initial ve-
locity of 2000 m/s and at an elevation of 25o . Due to Coriolis effects
it misses its target. Determine how far away from the target it lands
to first order, assuming that the target is at the zeroth order location.
Obtain the correction to the time of flight. Assume λ = 30◦ .
Problem 13.10. A low pressure area develops over the North At-
lantic. Draw a vector diagram to show that the air mass will tend to
move in a circle centered on the low pressure region. Is the air moving
clockwise or counterclockwise? Do the same thing for an air mass in
the South Atlantic.
Problem 13.11. A low pressure region in the North Atlantic is
located at 30◦ N . A weather ship is located a distance of 80 km from
the low. The atmospheric pressure at the ship is 1 atm, but it decreases
towards the low with a pressure gradient of 10−2 Pa/m. Evaluate the
wind velocity at the position of the ship. (The density of air is 1.3
kg/m3 .)
Problem 13.12. An object is dropped from 2 km above the Earth’s
surface at colatitude 45◦ . Determine how far it has deviated horizontally
from a line directed towards the center of Earth by the time it hits the
surface due solely to the effect of the centrifugal force. Although they
are much greater influences, neglect both the Coriolis force and air
resistance.
Problem 13.13. John and Mary (who live at 45◦ N ) go sky diving.
They jump out of the airplane together. His free-fall velocity is 8 m/s
and hers is 10 m/s. If the jump started at an altitude of 5 km, how far
apart will they be when they land? (Assume the Coriolis force is the
only factor determining this difference in final position.)
 13.8. PROBLEMS 477

Problem 13.14. An airplane is flying due East along the 30◦ N

latitude circle at 800 km/hr. (a) Determine the Coriolis acceleration
acting on it. (b) If the Coriolis force is not compensated for, how far
off course will the airplane be after one hour?
Problem 13.15. A stationary pendulum is hanging from the ceil-
ing of an airplane flying at a constant velocity of 800 km/hr, following
a meridian of longitude. Determine the deviation of the string of the
pendulum relative to a line to the center of the Earth at the moment
the airplane flies over the North Pole.
Problem 13.16. Suppose the Earth is perfectly spherical. An ec-
centric millionaire in Seattle builds an ice rink that is perfectly flat
and tangent to the surface of the Earth. Furthermore, the ice is per-
fectly frictionless. The millionaire is surprised to find that a puck set
in motion at the point of tangency slides back and forth like the bob
of a Foucault pendulum. Find the frequency of oscillation and the
frequency of precession of the trajectory of the puck.
Problem 13.17. It has been suggested that a “space habitat” in
the shape of a giant rotating doughnut (a torus) would allow people to
live inside in an environment with artificial gravity due to the centrifu-
gal force. (This would be analogous to being in the inner tube of a large
rotating bicycle wheel.) If the torus were big enough (say 3 km radius)
and rotated fast enough (at about 30 rev/hour), the centrifugal force
on the “outside” wall would be approximately equal to g. Suppose the
inhabitants of such a space habitat decided to play basketball. In the
toroidal cavity the playing court would be perpendicular to the radial
direction. (That is, “down” is radially outward and “up” is radially
inward.) Define the z axis along the direction of Ω, the rotational ve-
locity of the torus, and φ perpendicular to r and z such that r̂ × φ̂ = k̂.
Determine the Coriolis force on a ball that is thrown at speed v in the
following directions: (a) “up” (b) the azimuthal direction (c) parallel
to Ω. (d) Derive an expression for the point where a ball will land if it
is thrown upward at speed v0 . (Note: this means the initial velocity is
−v0 r̂.)
Problem 13.18. A dirigible crossing the North Atlantic at 50◦
latitude was flying at 200 mph due East. The navigator went to sleep
and the Coriolis force caused the dirigible to move in a circular path.
Determine the radius of the circle.
Problem 13.19. During the Falkland Islands war, the Argentine
Air force fired a number of Exocet missiles at British ships. Assume
478 13. ACCELERATED REFERENCE FRAMES

an airplane flying horizontally due East at 1000 km/hr at an altitude

of 6 km drops a missile. The rocket motor of the missile accelerates
it at 10 m/sec2 horizontally. Determine how far the Coriolis force will
have caused the missile to deviate from “straight ahead” by the time it
hits the water. You may assume the incident took place at a latitude
of 55◦ S. Ignore air resistance.
Problem 13.20. A particle of mass m is on a frictionless rotating
turntable. It is acted upon by a force directed towards the axis of
rotation and given by F = −kr. We wish to obtain an expression for
the position of the particle as a function of time in the rotating reference
frame. To do so, first find the rotation rate Ω for the turntable such that
the particle remains at rest. Next perturb the body radially (r = a + η,
where a = unperturbed value of r). Obtain expressions for r(t) and θ(t)
and then for x(t) and y(t) in the rotating reference frame. Describe
the motion. Hint: the θ equation can be easily integrated.
Problem 13.21. A mass m is acted upon by a force directed to-
wards the origin and given by F = −kr. Analyze the motion in a
rotating coordinate system. (a) Determine the rotation rate of a coor-
dinate system such that the unperturbed body is at rest. (b) Allow the
body to be perturbed so that now it undergoes radial oscillations. Ob-
tain expressions for r = r(t) and θ = θ(t) where r and θ are measured
in the rotating coordinate system.
COMPUTATIONAL PROJECTS
Computational Project 13.1. Write a computer program to
determine the impact point for a shell from a long range cannon. In-
clude the Coriolis effect to third order. (Neglect air resistance.)
Computational Project 13.2. Generate Figure 13.7 assuming
a pendulum of length 40 m at 40◦ N.
Computational Project 13.3. A stone is dropped from a height
of 100 meters at latitude 40◦ N. Determine the distance it is deflected
horizontally by the Coriolis force in the absence of air resistance. Next
include air resistance and compare the results.
 Part 4

The Mechanics of Extended Bodies

Chapter 14
Statics (Optional)

In this part of the book (Chapters 14 through 18) you will study
various aspects of extended material bodies. This chapter involves
a number of advanced concepts in statics. We will begin with a few
definitions and two simple theorems concerning systems of forces acting
on rigid bodies, then go on to analyze the statics of freely deformable
bodies such as a string or cable hanging from stationary supports.
This is followed by definitions of stress and strain and a generalization
of Hooke’s law. An important application is an investigation of the
properties of a fluid in equilibrium. The last topic is d’Alembert’s
Principle and the concept of virtual work. You will see how this is
related to the equilibrium of an extended body, and how the principle
can be used to derive Lagrange’s equations.

14.1. Basic Concepts

Statics is the study of bodies that are not accelerating, that is,
bodies that are at rest or moving at constant velocity. It is always
possible to find an inertial coordinate system in which the body is at
rest, so the linear velocity is usually assumed to be zero.
If a body is not accelerating, the net force and the net torque act-
ing on it must be zero. Consequently, the basic conditions for static
equilibrium are:
X
Fi = 0,
X
Ni = 0.
There are two useful theorems related to the equilibrium of a rigid
body.
481
482 14. STATICS (OPTIONAL)

Theorem 1. If a body is in equilibrium under the action of three

forces then the lines of action of the three forces all lie in the same
plane and intersect at one point.
Proof. Two force vectors define a plane. If the third force does
not lie in that plane, the tip-to-tail sum of the vectors do not form a
closed triangle. Therefore, for equilibrium, the third force must also lie
in the plane. Furthermore if two of the forces intersect at some point,
they contribute zero torque about that point. For the body to be in
equilibrium, the third force must also contribute zero torque about that
point and consequently its line of action must also pass through the
point. 
Theorem 2. If a force Fc can be considered the sum of two other
forces, Fc = Fa + Fb , then the torque about a point due to Fc is the
sum of the torques due to Fa and Fb .
Proof. The torque due to Fc is Nc = r × Fc . But
Nc = r × Fc = r× (Fa + Fb ) = Na + Nb ,
and the theorem is proved. This is sometimes referred to as the Theo-
rem of Varignon. 
The following examples and simple exercises will refresh your knowl-
edge on how to solve statics problems.

T1

T2

Mg

2M

2Mg

Figure 14.1. A system in equilibrium.

Worked Example 14.1. A rod of length L and mass M

is supported from the ceiling by two massless vertical strings of
 14.1. BASIC CONCEPTS 483

lengths L/2 and L, as shown in Figure 14.1. A weight of mass 2M

hangs from the lower end of the rod. Determine the tensions in the
strings.
Solution: The net force and the net torque must be zero. All
the forces act vertically so
X
F =0 =⇒ T1 + T2 = 3M g.
Taking torques about the lower end of the rod,
X L
N = 0 =⇒ T1 L cos θ = M g cos θ,
2
so
1
T1 = M g,
2
and consequently
T2 = 3M g − T1 = 2.5M g.

N

μN

m g cosθ

m g s in θ
θ mg

Figure 14.2. A spool of thread lies on its side on an in-

clined plane. As seen in the figure, the thread is wrapped
clockwise around the spool so it cannot roll without slip-
ping.

Worked Example 14.2. A spool of thread rests on its side

on an inclined plane. See Figure 14.2. It does not roll down the
plane because the string is wrapped over the top of the spool and
the thread cannot simply unwind. In fact, when the spool slips, it
will have to rotate counter clockwise to unwind the string. Assume
the higher end of the inclined plane is slowly raised until the spool
484 14. STATICS (OPTIONAL)

slips. Let the coefficient of static friction between spool and plane
be µ = 0.2. Determine the angle at which the spool begins to slip.
Solution: Forces up the plane equal forces down the plane so
if N is the normal force,
mg sin θ = T + µN.
The forces perpendicular to the plane also add to zero,
N = mg cos θ.
Clockwise torques equal counterclockwise torques. Taking torques
about the point of contact between spool and plane:
T (2R) = mgR sin θ,
1
T = mg sin θ,
2
where R is the radius of the spool. Since T = mg sin θ − µN =
mg sin θ − µmg cos θ we have
1
mg sin θ = mg sin θ − µmg cos θ,
2
tan θ = 2µ = 0.4,
θ = 21.8◦ .

Exercise 14.1. Prove that a building such as the leaning Tower of

Pisa will not topple over if a vertical line from its center of mass passes
through its base.
Exercise 14.2. A yo-yo of mass m and radius r hangs from a nail
on a wall such that its curved side is in contact with the wall. The nail
is a distance L above the point of contact. Determine the force exerted
by the wall on the yo-yo. You may assume the string is attached to
the center of mass of the yo-yo, as shown in Figure 14.3. Answer:
N = M gr/L.

Exercise 14.3. A ladder of length 2 m and mass 10 kg rests against

a smooth wall. The coefficient of static friction between ladder and floor
is 0.2. How far from the wall is the foot of the ladder when it begins
to slip? Answer: 0.74 m.

14.2. Couples, Resultants and Equilibrants

 14.2. COUPLES, RESULTANTS AND EQUILIBRANTS 485

Figure 14.3. A yo-yo hangs from a nail in the wall.

Assume the inner axle has diameter zero so the string is
attached to the center of mass. (Exercise 14.2)

Here are a few definitions and some concepts that are useful when
treating statics problems. Couple: A couple is a system of forces
P
whose sum is zero, i.e., Fi = 0. Clearly a couple cannot generate a
linear acceleration, but it can exert a torque on a body, as is obvious
from an inspection of Figure 14.4. Although a couple may be a system
of many forces, two forces are sufficient to describe it.

Figure 14.4. A couple

Equivalent Systems of Forces: If two systems of forces exert

the same total force and the same total torque about every point, then
the two systems of forces are called equivalent.
Resultant: If a force R acting at point P is equivalent to a system
of forces Fi acting at points Pi , then R is called the resultant of the
system of forces.
Equilibrant: If a force R acting at P is the resultant of the system
of forces Fi acting at several points Pi , then applying the force R at P
will cause the same linear and angular acceleration as all of the forces
Fi acting at points Pi . Consequently, if one applies a force −R at P, the
system of forces is balanced. For this reason, −R acting at P is called
the equilibrant. A system that is not in static equilibrium is brought
486 14. STATICS (OPTIONAL)

into static equilibrium by applying an equilibrant. The practical uses

of the equilibrant in designing structures is obvious.
A useful fact about couples is given by the following theorem.
Theorem 3. A couple exerts the same torque about every point.
Proof. The torque exerted about point Q by a couple is
X
NQ = (rQi × Fi ) ,
where rQi is a vector from Q to the point of application of force Fi .
The torque exerted by this couple about a different point Q0 is given
by X
NQ0 = (rQ0 i × Fi ) .
But rQ0 i = rQi +d where d is the vector from Q to Q0 . So
X X
NQ0 = ((rQi +d) × Fi ) = NQ + d × Fi .
i
P
By the definition of a couple, i Fi = 0. Therefore NQ0 = NQ , and
the theorem is proved. 

Exercise 14.4. A bridge is collapsing when Superman flies under

it and holds it up by applying a single force at one point. This is
possible because Superman is very good at determining the equilibrant
of a system. Suppose the bridge has a mass of 100,000 kg, and a length
of 20 meters. Other forces acting on the bridge (just before Superman
appeared) were 300,000 N at the left end and 150,000 N at the right
end, both acting vertically upward. The weight of the bridge acts at its
geometric center. Determine the equilibrant. Answer: R = 5.3 × 105
N at 12.8 m from the left end.
Exercise 14.5. A rod of length 1 meter has strings attached at
either end and at the middle. The string attached to the left end
makes an angle of 30◦ to the rod and exerts a force of 10 N on it. The
string attached to the middle makes an angle of 270◦ to the rod and
exerts a force of 15 N. The string on the right end makes an angle of
90◦ and exerts a force of 10 N. All the forces lie in the same plane.
Determine the resultant and the equilibrant. Answer: R = 8.66 N at
90◦ to the rod acting at 0.79 m from the left end.
 14.4. THE HANGING CABLE 487

14.3. Reduction to the Simplest Set of Forces

It is often convenient to reduce a complicated set of forces to the
smallest number of forces that are equivalent to the original system,
that is, to the smallest number of forces that yield the same total force
and same total torque.
Theorem 4. Any system of forces is equivalent to a single force
applied at an arbitrary point plus a couple.
Proof. Consider a systemP of forces, Fi , i = 1, 2, · · · n, applied at
points P1 , P2 , · · · , Pn . Let F = Fi , so F is the total force exerted by
the system of forces. Let NQ be the torque about some given point Q
due to the system of forces. Assume the net force F acts at an arbitrary
point P. This gives rise to a torque about Q, but in general it is not
equal to NQ . Adding a couple to the system does not change the net
force, but it allows one to obtain the desired net torque NQ . Let the
couple be composed of the two equal and opposite forces Fc and −Fc .
Let +Fc act at P (the same place where F is acting). Then the original
system of forces is equivalent to a force F + Fc acting at P plus a force
−Fc acting at some other point such that the torque is equal to that of
the original system. Thus the system of forces is reduced to one force
and a couple and the theorem is proved. 

Exercise 14.6. A 1 meter rod lying along the x axis is acted upon
by the following three forces: A force F1 = 10̂ N at the left end, a
force F2 = −20̂ N at the midpoint, and a force F3 = 5ı̂ N at the right
end. Determine an equivalent system of two forces.

14.4. The Hanging Cable

A string (and to a lesser extent, a cable or a chain) is a “freely
deformable” body. This section is an analysis of the physics of an ideal
rope or cable that is suspended from two fixed points. We will study
two problems: (1) A “bridge” suspended from cables, and (2) A rope
hanging under its own weight.
14.4.1. A Suspension Bridge. We begin by considering a sus-
pension bridge, such as shown in Figure 14.5. The suspension cables
are assumed massless and the only load is the weight of the roadway
488 14. STATICS (OPTIONAL)

which is distributed uniformly in the horizontal direction. The points

of support (A,B) are at the same level. It is desired to find the shape
of the curve formed by the cables.

A B

Figure 14.5. An idealized suspension bridge.

In a problem of this type it is convenient to consider a portion of a

cable starting with the lowest point (C), as illustrated in Figure 14.6.
Since C is the lowest point of the sagging cable, the tension at C, (Tc ),
is horizontal. The tension at a nearby point D, (Td ), is at an angle θ
with respect to the horizontal. The section of roadway supported by
this section of the cable has length x and weight wx where w is the
weight per unit length of the roadway.

Td
y D θ

Tc C

wx
x
O

Figure 14.6. A portion of the cable of a suspension bridge.

The lines of action of the three forces (Tc , Td and wx) pass through
the same point (by Theorem 1 of section 14.1). By symmetry this point
is at x/2. The sum of the forces is zero so
Td sin θ = wx,
Td cos θ = Tc .
Dividing one equation by the other,
wx
tan θ = .
Tc
 14.4. THE HANGING CABLE 489

In the limit of D being very close to C, this reduces to

dy wx
= .
dx Tc
Integrating this differential equation gives an expression for y as a
function of x. But y = y(x) is the equation for the curve formed by the
sagging cable. Integrating:
wx2
y= + c.
2Tc
This is the equation of a parabola.

Exercise p
14.7. Show that the tension at any point in the cable is
given by T = Tc2 + w2 x2 .

14.4.2. A Hanging Rope. Next we consider a slightly more dif-

ficult problem, namely a rope hanging under its own weight. Consider
a rope (or cable or chain) suspended from two ends which may or may
not be at the same height. Again, we want to determine the shape of
the rope.

y
T
B
θ
A
s
T0 W=ws
C
x

Figure 14.7. A portion of a rope sagging under its own weight.

Consider a segment AB of length s as shown in Figure 14.7. For

convenience s is measured from the lowest point in the sagging rope.
The segment is subjected to three forces, namely the horizontally di-
rected tension T0 at A, the tension T acting at angle θ at B, and the
weight W = −ws acting downward at the center of mass of the seg-
ment. Note that if the force of gravity on the rope is expressed as a
490 14. STATICS (OPTIONAL)

force per unit length w thenRthe total “body force”1 acting on the piece
s
of rope of length s will be 0 wds. In equilibrium, the vector sum of
the three forces acting on the segment is zero. Equating the vertical
and the horizontal components of these forces leads to
T cos θ = T0 ,
T sin θ = ws.
Therefore
w
tan θ = s,
T0
or
T0
s= tan θ = c tan θ, (14.1)
w
where c = w/T0 and has the units of length. (In a little while I will
give you a geometric interpretation of c as the vertical distance from
the origin to the lowest point in the rope.)
If you are very good at analytical geometry, you might have recog-
nized equation (14.1) as the intrinsic equation of a catenary. In this
sense, the problem is solved, but most people would prefer to have a
formula for the curve in Cartesian coordinates x, y.
Note that the Cartesian coordinates shown in Figure 14.7 have the
y axis passing through the lowest point of the curve, but the origin is
displaced vertically from this lowest point.
Equation (14.1) can be written as
dy s
tan θ = = .
dx c
Consequently,
p
d2 y 1 ds 1 dx2 + dy 2
= = ,
dx2 csdx c dx
 2
1 dy
= 1+ .
c dx
dy
Let p = dx
and write this relation as
dp 1p
= 1 + p2 ,
dx c
dp 1
p = dx.
1 + p2 c
1A
“body force” is a force that is proportional to the amount of matter being
considered. Thus, for example, the tension in the rope would not be a body force,
but the weight of the rope depends on the amount of rope and is a body force.
 14.4. THE HANGING CABLE 491

Integrating
x
= sinh−1 p + constant.
c
Since the slope (p) is zero at x = 0, the constant of integration is zero.
Therefore, x
p = sinh .
c
But p = dy/dx, so
dy x
= sinh .
dx c
Integrating again leads to
x
y = c cosh + constant.
c
If the origin is placed a distance c below the lowest point, then y = c
at x = 0 and the constant of integration is zero. Thus the equation of
the curve formed by the hanging rope is
x
y = c cosh , (14.2)
c
which is the equation of a catenary in Cartesian coordinates.
The tension at any point in the rope is obtained from
T0 T0 ds
T = = dy = T0 .
cos θ ds
dy
But s = c sinh(x/c) so
x
T = T0 cosh .
c
Furthermore, y = c cosh xc so

T0
T = y,
c
and finally
T = wy.
That is, the tension at any point in the rope is directly proportional to
the height of that point. The maximum tension occurs at the highest
points, that is, at the end points. For a rope whose end points are at
the same height and are separated by a horizontal distance a (called
the “span”) the end points are at x = ±a/2, and y = h + c, where h
(the “sag”) is the vertical distance from the points of support to the
lowest point in the rope. The maximum tension is then
a
Tm = w(h + c) = wc cosh .
2c
492 14. STATICS (OPTIONAL)

Worked Example 14.3. Obtain an expression for the length

of a hanging rope in terms of the parameter c.
Solution: Assume the rope is hanging between points 1 and
2. If L is the length of the rope, then
Z 2
L= ds = s2 − s1 .
1
But
 
dy d   x  x
s = c tan θ = c =c c cosh = c sinh .
dx dx c c
Consequently,
h x   x i
2 1
L = c sinh − sinh
c c
If points 1 and 2 are not at the same height then the difference in
their heights is
h x   x i
2 1
H = y2 − y1 = c cosh − cosh .
c c
Consequently,
 a  a
2 2 2 2 2
L = H + 2c cosh − 1 = H + 4c sinh .
c 2c
This relationship can be inverted to determine the parameter c of
a catenary in terms of its length L, its span a, and the vertical
displacement of the end points H.

Exercise 14.8. Show that

 the length of a sagging rope is given by
L = c sinh xc2 − sinh xc1 .


Exercise 14.9. A rope of length 3 meters and mass per unit length
of 0.2 kg/m is hanging under its own weight. The endpoints are lo-
cated at (-1,1.6) and (1,1.6). Numerically determine the parameter c.
Determine the maximum value of the tension. Answer: c = 0.62.

14.5. Stress and Strain

Let us now turn our attention to another aspect of real bodies:
they are not completely rigid. When acted upon by forces, real bodies
undergo deformations. These deformations are due to forces that act
 14.5. STRESS AND STRAIN 493

across surfaces. For example, to describe what happens when you

squeeze a tennis ball, you need to specify not only the magnitude and
the direction of the force, but also the area of the surface across which
it is transmitted. A force that acts across a surface is called a stress. A
stress is always expressed as a force divided by an area (F/A). Stresses
are characterized by their effect. If you squeeze something, you tend to
compress it. The stress is then called a pressure. If the force tends to
stretch the body upon which it acts, as when you pull on the ends of a
rubber band, the stress is called a tension. If the stress tends to change
the shape of the body, as for example, when a force acts parallel to a
surface of the body, it is called a shear. These three types of stress are
illustrated in Figure 14.8.

Tension θ
Compression Shear

Figure 14.8. Examples of three types of stress. A ten-

sion is represented by a force exerted across the end of a
rod, tending to stretch it. A compression is represented
by forces perpendicular to the faces of a cube, tending
to squeeze it. A shear is represented by forces parallel to
the top and bottom faces of a cube, tending to deform
it.

The deformation of the body is called the strain. As with the

stress, the strain depends on the situation. Thus, a rod of length l
subjected to a force across an end (a tension) will undergo a change
in its length (∆l). The deformation will be characterized by ∆l/l. A
cube subjected to a compression will undergo a change in volume. This
deformation is characterized by ∆V /V. A parallelepiped subjected to
a shear might have its top surface slide relative to the bottom surface.
This is characterized by the tangent of the angle θ shown in Figure
14.8.
The relation between stress and strain is given by an empirical
relation known as Hooke’s law:
Strain is proportional to stress.
This law is true for most material objects as long at the strain is small.
The proportionality constant is a characteristic of the material and is
494 14. STATICS (OPTIONAL)

tabulated in various places, such as the CRC Tables.2 Hooke’s law for
a long thin rod of length l is

F ∆l
=Y , (14.3)
A l
where F is the tension, A is the cross-sectional area of the rod and Y is
a quantity characteristic of the composition of the rod called Young’s
modulus.
For a compression, Hooke’s law is
F ∆V
= −B , (14.4)
A V
where B is called the bulk modulus. In this case, the quantity F/A is
usually called the pressure, as it corresponds to the notion of pressure
as a force acting across an area that tends to cause a compression. The
negative sign in the law emphasizes that the stress is a compression.
Finally, for a shear, Hooke’s law is
F
= n tan θ, (14.5)
A
where n is called the shear modulus.
The Stress Tensor
You have certainly noticed that the left hand side of the Hooke’s
law equations all have the same form although they represent different
situations, specifically the direction of the force and the orientation of
the surface relative to the force. It turns out that we can put all of
these into a single expression by defining a quantity called the stress
tensor.
We are considering forces that act across surfaces. In general, the
material on one side of a real or imaginary surface exerts a force on the
material on the other side. For example, the molecules on one side of an
imaginary plane in a material body will exert forces on the molecules
on the other side. Of course, we are usually interested in real surfaces
with different materials on either side. The forces exerted across a
surface, are called stresses. They can be a tension, a compression or a
shear.
The force acting across a surface can have an arbitrary direction
and in general is given by
F =Fx ı̂+Fx ̂+Fz k̂.
2CRC Tables of Chemistry and Physics, 87th Edition, CRC Press, Boca Raton,
Fl. 2006.
 14.6. THE CENTROID (OPTIONAL) 495

The orientation of a surface is usually specified by a vector perpen-

dicular to the surface. How can we describe a force in the x direction
acting across a surface oriented perpendicular to the y direction? As
you might guess, this could be written in a variety of ways, such as
(Fx )y , or some such convention. Since the stress is a force divided by
an area let us use the letter P to represent the possible stresses and
denote them by Pxy . This will be the force per unit area acting in the x
direction across a surface whose normal is in the y direction. Obviously
there are nine such terms and they can be expressed quite nicely as a
matrix, thus:  
Pxx Pxy Pxz
 Pyx Pyy Pyz 
Pzx Pzy Pzz
This is called the stress tensor. A matrix is an array of numbers. It
need not have any physical significance. A tensor can be represented
as a matrix, and as we shall see, a tensor obeys the rules of matrix
algebra. But a tensor is not a matrix. A tensor is a physical quantity
which is often represented as a matrix, but it can be represented in
other ways as well. A tensor has much in common with a vector. In
fact, a vector is a special kind of tensor. Note that a vector is a physical
quantity with three terms. A tensor is a physical quantity with nine
terms. We will come back to the question of tensors in Chapter 16, but
for now, I just wanted to mention the concept to make you aware that
in general stress is a tensor.

Exercise 14.10. Write the stress tensor for a fluid in equlibrium.

14.6. The Centroid (Optional)

A quantity closely related to center of mass is the centroid of a
geometrical figure. For an object of constant density, the two points
coincide. However, the centroid is a purely geometrical concept and
does not depend on the density or the density distribution of the body.
The centroid is a quantity that depends only on the size and shape of
a geometrical figure. If R represents the location of the centroid, then
it is defined for a volume, a surface and a curve as follows:
For a volume, Z
1
R= rdτ.
V
496 14. STATICS (OPTIONAL)

The centroid of a surface of area A is given by

Z
1
R= rdA,
A
and the centroid of a curve of length S is given by
Z
1
R= rds.
S
Some theorems concerning centroids have come down from antiq-
uity and are known as the Theorems of Pappus.
Theorem 5. A “surface of revolution” is generated by revolving a
plane curve about a line. If a plane curve is revolved about an axis in
its own plane and does not intersect the axis, the area of the surface
of revolution is equal to the product of the length of the curve and the
distance traced out by the centroid of the curve.
Proof. Consider a small portion of the curve, ds, a distance y
from the axis. See Figure 14.9. As the curve is revolved about the
axis, Rit sweeps out an
R area dA = 2πyds. The total area swept
R out is
A = 2πyds = 2π yds. But the centroid is at Y = (1/s) yds, so
A = 2πY s, and the theorem is proved. 

A s

Figure 14.9. A curve of length s is revolved about an

axis to generate a surface of revolution.

Theorem 6. A “solid of revolution” is generated by revolving a

planar surface around an axis in its own plane. If the surface does
not intersect the axis, the generated solid has a volume equal to the
product of the area of the surface and the distance traced out by the
centroid of the surface.
The proof of this theorem is left as an exercise.
As an example of the usefulness of the theorems of Pappus consider
the problem of determining the centroid of a semicircular disk. Let the
 14.7. THE CENTER OF GRAVITY (OPTIONAL) 497

disk be rotated about an axis lying on the straight side of the disk, as
shown in Figure 14.10. The volume of revolution generated is a sphere
of volume V = (4/3)πa3 . But according to theorem 6, this volume is
equal to the area of the disk times 2πY where Y is the distance from
the axis to the centroid of the shaded disk. That is
 
3 1 2
(4/3)πa = (Area of disk) 2πY = πa 2πY
2
4a
∴ Y = .
3π

Figure 14.10. A semicircular disk is rotated about an

axis to generate a sphere.

Exercise 14.11. Prove the second theorem of Pappus (Theorem

6).

14.7. The Center of Gravity (Optional)

The center of gravity is often confused with the center of mass. In
a uniform gravitational field, the center of gravity and the center of
mass lie at the same point. Often engineering drawings will have a
point denoted “CG”, but almost invariably this point is the center of
mass rather than the center of gravity. The difference between these
two concepts becomes quite clear when you consider their definitions.
You know the definition of the center of mass (Equation 5.16). It
is related to how the mass is distributed in an extended body. On the
other hand, the center of gravity is defined to be that point at which
498 14. STATICS (OPTIONAL)

all of the mass of the body can be concentrated to experience the same
gravitational force as the extended body.
As an example, think of a very long rod. The center of mass of the
rod does not depend on its orientation. But the center of gravity of
the rod will depend on whether the rod is held horizontally, parallel to
the surface of the Earth, or vertically, perpendicular to the surface of
the Earth. If the rod is horizontal, the center of mass and the center of
gravity will coincide; if the rod is vertical, the center of gravity will be
a little bit below the center of mass. This is because the gravitational
force decreases with distance from the center of the Earth and those
parts of the rod nearest the Earth are attracted a bit more strongly
than the points of the rod that are further away. Of course, the rod
would have to be extremely long for the distance between the two points
(CG and CM) to be at all noticeable.
The center of gravity of an extended body of mass M can be defined
in terms of its interaction with a particle (say m). The extended body
and the particle are illustrated in Figure 14.11. Suppose the extended
body shrinks to a point. Where should this point mass be located to
experience the same force as the extended body? The location of this
hypothetical point mass is called the center of gravity of M relative to
m. In Figure 14.11 it is indicated by the symbol CG.

m m
-F

CG M
M CG

Figure 14.11. The center of gravity of the extended

body is at the point CG. The left hand panel shows that
every element in M is attracting m. The net force is the
sum of all the force vectors.The right hand panel shows
the action-reaction forces assuming that the extended
body can be replaced by a point mass M at CG.

To determine the location of the center of gravity of M, note that

the force vectors (call them Fi ) due to all the mass elements in M are
directed along lines that go through m. This is illustrated in the right
 14.8. EQUILIBRIUM OF FLUIDS 499

hand side of Figure 14.11. The net force on m is

X
F=− F.
i
By Newton’s third law, the force on M is then +F and it must act
along the same line of action. This line goes through point CG in the
body. Let the distance from m to CG be denoted by rcg . Now,
X ∆Mi Z
X  ρdV
Fi.  = −Gm = −Gm
 
2
ri r2

2
P
But | Fi. | = −G(M m/rcg ), so
 Z −1/2
1 ρdV
rcg =
M r2
The point CG will coincide with the center of mass of M only if the
body has spherical symmetry.
We have defined the center of gravity of an extended body relative
to a particle. For two arbitrarily shaped bodies one cannot, in general,
define a unique center of gravity for either body.
14.8. Equilibrium of Fluids
By definition, a fluid (such as a gas or liquid) is a substance that
cannot support a shear. Even a very viscous fluid like tar will slowly
deform under a shear until it reaches an equilibrium state in which
the shear is zero. Therefore the only stress forces acting on a fluid in
equilibrium are forces perpendicular to the surface, and are pressures.
When I said that in a fluid in equilibrium the forces are perpendicu-
lar to the surface, I did not just mean the exposed surfaces of the fluid,
but any surface, including surfaces entirely within the fluid. For exam-
ple, you can imagine a surface lying between two layers of molecules in
the fluid. The material on one side of this (imaginary) surface exerts a
force on the material on the other side, and the force is perpendicular
to the surface. If not, there would be a shear and by definition the
fluid would not be in equilibrium.
Imagine a large tank full of some fluid, say water. You could de-
termine the pressure at any point in the tank by lowering a pressure
sensor gauge into the tank. The fluid exerts a force on the surface
of the pressure sensor and the force per unit area is the pressure at
that point in the fluid. It is an interesting fact that the pressure in
a fluid is the same in all directions. If you rotate the pressure sensor
so that it faces a different direction, it reads the same value. That is,
the pressure is the same regardless of the orientation of the surface.
This fact is easily proved by considering a region of the fluid that is
500 14. STATICS (OPTIONAL)

assumed (for simplicity) to have the shape of an equilateral prism. See

Figure 14.12. Let this volume element be infinitesimal. The fluid is at
rest so the net force on it is zero, and the three forces have the same
magnitude. (See Figure 14.12c.) The faces all have the same area. If
the forces are equal and the surface areas are equal, then the pressure
is the same for each face and consequently the pressure is independent
of the orientation of the faces.

(a) (b) (c)

Figure 14.12. (a) An infinitesimal volume element of

fluid in the shape of an equilateral prism. (b) The forces
on the rectangular faces of the prism are perpendicular
to the surfaces. (c) The forces add up to zero, so the
magnitudes of the forces are all the same.

The prism in the previous argument was assumed infinitesimal to

avoid including the weight of the fluid in the prism.3 However, if gravity
(or indeed any body force) acts on the fluid in equilibrium, then in
addition to the pressure there is a force on a finite volume element ∆V
given by
F =mg =(ρ∆V )g,
where ρ is the density of the fluid. The body force, f , here defined as
the force per unit mass, is
f = ρg = −ρg k̂.
To obtain an expression for the pressure as a function of position, you
can construct an (imaginary) infinitesimal upright cylinder at some
arbitrary point in the fluid. Let the axis of the cylinder be vertical,
that is, in the direction of the gravitational force. The top and bottom
faces have area dA and the distance between the faces is dz. There
are three forces acting on the fluid in the cylinder, namely: (1) the
downward force due to the pressure on the top surface, (2) the upward
3The body force (weight) is proportional to the volume whereas the surface
force is proportional to the area. As the fluid element shrinks to zero, the volume
decreases faster than the area. (Volume ∝ r3 , and Area ∝ r2 .) Therefore, the body
force can be neglected for an infinitesimal volume element.
 14.8. EQUILIBRIUM OF FLUIDS 501

force due to the pressure on the bottom surface, and (3) the downward
force of gravity on the fluid. Let the bottom of the cylinder be at z
and the top at z + dz. Then the equilibrium condition
force up = force down
is expressed by
p(z)A = p(z + dz)A + ρgdzdA.
Using the definition of derivative, this expression can be written
dp
= −ρg. (14.6)
dz
That is,
dp
k̂ = f .
dz
It is easy to generalize this expression to obtain an expression for the
change in pressure with respect to a translation in an arbitrary direc-
tion, leading to
∇p = f . (14.7)
That is, the gradient of the pressure is equal to the body force. This
implies that the surfaces of constant pressure are perpendicular to the
body force.
Using the vector definition of differential, we have
Z Z Z
∇p · dr = dp = f ·dr,

and so Z r
p(r) = p(r0 ) + f ·dr, (14.8)
r0
where the integral is along any path from r0 to r in the fluid. Note
that for a fluid in equilibrium, equation (14.8) defines a pressure field
(the pressure is defined at every point in the fluid). Furthermore, the
equation states that pressure changes are balanced by the body force,
f.
Equation (14.8) also tells us that if the body force is constant,
increasing the pressure at one point in a fluid increases the pressure by
an equal amount at all other points in the fluid. This is called Pascal’s
law. For example, if p(r0 ) is the pressure at the surface of the fluid,
then an increase in the surface pressure is communicated to all points
in the fluid.
To determine the pressure at any point in a fluid in equilibrium,
you can apply equation (14.8). For example, if a fluid is acted upon
502 14. STATICS (OPTIONAL)

only by the gravitational force, the total force on a volume element V

is Z Z Z Z Z Z
FW = f dV = gρdV = mg = weight,
V V
where “weight” is the weight of the fluid in the volume. The force
due to the pressure can be obtained by integrating over the surface
enclosing V, thus I
FP = − n̂pdA,
A
where n̂ is the outward normal. (That is the reason for the minus sign.)
For equilibrium
FP = −FW .
This does not tell you anything unexpected. The weight of the fluid
is equal to the pressure force on it. Now, however, assume the volume
of fluid is replaced by a solid body of the same shape and size. The
pressure force on it is the same as before, as it is due to the fluid outside
of the body. The force exerted by the fluid does not depend on the
nature of the material in the volume. But now FW is the weight of the
solid body, rather than the weight of the fluid. Therefore the upward
(buoyant) force on the body is equal to the weight of the displaced
fluid. This is Archimedes’ principle.

Worked Example 14.4. Determine the rate of change of pres-

sure in a moving fluid.
Solution: In a fluid the pressure is a function of both position
and time, thus
p = p(x, y, z, t).
There are, consequently, two time rates of change to be considered:
the change in p with time at a fixed point and the change in p with
position as the fluid element moves to a new position. Therefore,
by the definition of the differential of a function of several variables,
dp ∂p ∂p dx ∂p dy ∂p dz
= + + + ,
dt ∂t ∂x dt ∂y dt ∂z dt
∂p ∂p ∂p ∂p
= + vx + vy + vz .
∂t ∂x ∂y ∂z
That is,
dp ∂p
= + v·∇p.
dt ∂t
 14.8. EQUILIBRIUM OF FLUIDS 503

This is called the “convective derivative.”

Exercise 14.12. Prove that the pressure is independent of direc-
tion by adding the Cartesian components of the forces acting on the
rectangular faces of the prism of Figure 14.11.
Exercise 14.13. Generalize the argument above to show that if
the pressure is a function of position it is related to the body force by
∇p = f. (That is, derive equation 14.7.)
Exercise 14.14. Show that equation (14.7) implies that the body
force is conservative.
Exercise 14.15. Show that for a fluid with constant density the
pressure increases with depth as p = p0 + ρgh where h is the depth.
Determine the pressure a distance 3 m below the surface of a lake.
Answer: 1.3 × 105 pascal.
Exercise 14.16. A cylinder of radius 50 cm with an open nozzle
on the top, is filled with water to a height of 1.5 meters. A plug of
radius 1 cm is at the bottom of the cylinder. (a) Determine the net
force on the plug if the air pressure is 1 atmosphere (1 atm = 1.01X105
Pascal and 1 Pascal = 1N/m2 .) (b) The nozzle is now attached to a
pump and air is pumped in until the pressure reaches 3 atm. What is
the force on the plug now? Answer: (a) 36.4 N.
Exercise 14.17. A hydraulic automobile lift is made up of two
interconnected cylindrical reservoirs filled with hydraulic fluid. One
of the cylinders, of radius 5 cm, is attached to an air pump so the
surface pressure can be increased from 14.7 lbs/in2 to 44.1 lbs/in2 . The
other cylinder has a radius of 25 cm. How heavy an automobile can
be supported by the hydraulic fluid in the second cylinder? Answer:
39660 N.

An Ideal Gas
The volume of a gas depends on the pressure exerted on it. The
basic relation, based on Hooke’s law (stress ∝ strain) is given by equa-
tion(14.4). If the pressure is increased by an infinitesimal amount dp
the volume will decrease by dV and equation (14.4) can be expressed
as
dV dp
− = .
V B
504 14. STATICS (OPTIONAL)

We prefer to formulate the expression in terms of density rather than

volume. Since ρ = m/V, for constant mass,
dρ = −mV −2 dV,
so
dρ mdV V dV
=− 2 =− .
ρ V m V
Consequently
dρ dp
= .
ρ B
Integrating yields Z p 
dp
ρ = ρ0 exp .
po B
Most gases under standard conditions of temperature and pressure are
in reasonably good agreement with the ideal gas law,
pV = nRT,
where T is the temperature, n is the number of moles and R is the
universal gas constant. Since ρ is defined as the mass divided by the
volume and the mass is nMW (where MW is the molecular weight), you
can replace n/V by ρ/MW and write the ideal gas law as
MW
ρ= p
RT
Using equation (14.6) leads to the barometric equation
dp gMW
=− p.
dz RT
As an example, consider an “isothermal atmosphere,” that is, an at-
mosphere in which the temperature is constant. This is, of course, not
at all like the real atmosphere of Earth which has a rather complicated
temperature structure. Nevertheless it is a “zeroth order” approxima-
tion that gives some useful information. Integrating the equation for
dp/dz leads to  
gMW
p = p0 exp − (z − z0 ) .
RT
Here z0 is usually taken to be sea level and set equal to zero, and p0 is
the pressure at sea level (usually assumed to be 1 atm).

Worked Example 14.5. Suppose the temperature of the at-

mosphere decreases linearly with altitude: T = T0 − αz. (This is
 14.9. D’ALEMBERT’S PRINCIPLE AND VIRTUAL WORK (OPTIONAL) 505

somewhat more realistic than assuming an isothermal atmosphere.)

Obtain an expression for the pressure as a function of altitude (z).
Solution: Given the barometric equation:
dp gMW gMW p 1
=− p− .
dz RT R T0 − αz
So
dp gMW dz
= − ,
p R T0 − αz
Z p
gMW z dz
Z
dp
= − ,
p0 p R 0 T0 − αz
   
MW g T0 − αz MW g α
ln(p/p0 ) = ln = ln 1 − z ,
αR T0 αR T0
and hence  MW g/αR
α
p = p0 1 − z .
T0

Exercise 14.18. The pressure gradient is dp/dz = −(gMW /RT )p.

Integrate to obtain the expression given above for the pressure as a
function of altitude in an isothermal atmosphere.

14.9. D’Alembert’s Principle and Virtual Work (Optional)

An important technique used by mechanical engineers to determine
equilibrium conditions and to evaluate forces of constraint is based on
the principle developed by the French mathematician and philosopher,
Jean d’Alembert (1717-1783).
Recall that a constraint is an equation relating generalized coordi-
nates (see Section 4.4). A constraint that involves only the generalized
coordinates (and not the generalized velocities or any other quantities)
is called a “holonomic” constraint. A constraint can be used to express
one coordinate in terms of the others, thus reducing the number of co-
ordinates required to describe the motion (and, consequently, reducing
by one the number of Lagrange equations that need to be solved).4
4Ifthe constraints acting on a system are holonomic they can be determined by
using a technique called the method of Lagrange multipliers. This topic is usually
reserved for graduate courses in mechanics and we will not consider it further.
506 14. STATICS (OPTIONAL)

The principle of d’Alembert is, in a sense, a re-statement of New-

ton’s second law, but it is expressed in a way that makes it very useful
in advanced mechanics. For a system of N particles, d’Alembert’s
principle is
XN
Fext


i − ṗi · δri = 0, (14.9)
i=1
where Fext
i is the external force acting on particle i and ṗi is the rate
of change of its momentum. By Newton’s second law it is clear that
the term in parenthesis is zero. Therefore, multiplying this term by δri
has no effect. Furthermore, summing over all particles is just a sum
of terms that are all equal to zero. Thus the fact that the expression
is equal to zero is obviously true. What is not so obvious (yet) is why
this particular formulation is of any value.
It is convenient to express d’Alembert’s principle in terms of Carte-
sian coordinates. Then the d’Alembert principle in scalar form is
n=3N
X
Fiext − ṗi δxi = 0.


(14.10)
i=1

The quantity δri in (14.9) or δxi in (14.10) is called a virtual displace-

ment. (See Section 4.6.) A virtual displacement is one that has the
following properties: It is instantaneous (that is, dt = 0), it is infini-
tesimal, and it is consistent with any constraints acting on the system.
This means that time is kept constant and the forces and constraints
acting on the system are “frozen” during the virtual displacement. The
work done during a virtual displacement is called the “virtual work.”
The principle of virtual work states that for a system in equilibrium,
the work done in any virtual displacement is zero, that is,
X
δW = Qi δqi = 0, (14.11)
i
where Qi is the generalized force, defined by
X ∂xi
Qj = Fi .
i
∂qj
Since δW can also be expressed as ΣFi δxi , we appreciate that ΣFi δxi =
ΣQj δqj . This is a useful relationship for determining the equilibrium
conditions for complicated systems.
 14.9. D’ALEMBERT’S PRINCIPLE AND VIRTUAL WORK (OPTIONAL) 507

m1 m2

θ1 θ2

Figure 14.13. Two masses connected by a string on

inclined planes. Find the equilibrium condition.

Worked Example 14.6. Two smooth inclined planes are joined

at the top. The planes form angles of θ1 and θ2 with the horizontal
as shown in Figure 14.13 Two masses, m1 and m2 are connected by
a string over an ideal pulley. Using the principle of virtual work,
determine the equilibrium condition.
Solution: Assume a virtual displacement δx in which mass m1
moves up one plane and mass m2 moves down the other plane.
The force of gravity acting on m1 in this direction is −m1 g sin θ1
and the force on the second mass is +m2 g sin θ2 . According to the
principle of virtual work,
−m1 g sin θ1 δx + m2 g sin θ2 δx = 0.
Therefore, equilibrium requires that
m1 sin θ2
= .
m2 sin θ1

Exercise 14.19. Consider an Atwood’s machine with two equal

masses. Show that equation (14.11) holds for this system.

14.9.1. A Derivation of Lagrange’s Equations. An interest-

ing aspect of d’Alembert’s principle is that it leads to a different way
to derive Lagrange’s equations. We begin by considering the second
term in d’Alembert’s principle, namely ṗi δxi . The sum of ṗi δxi over all
particles is
n n k n k
X X X ∂xi X X ∂xi
ṗi δxi = ṗi δqj = mi ẍi δqj , (14.12)
i=1 i=1 j
∂qj i=1 j=1
∂qj
508 14. STATICS (OPTIONAL)

which uses the fact that if xi = xi (q1 , · · · , qk ) then

k
∂xi ∂xi ∂xi X ∂xi
δxi = δq1 + δq2 + · · · + δqk = δqj .
∂q1 ∂q2 ∂qk j=1
∂qj

Furthermore,
 
d ∂xi ∂xi d ∂xi
mi ẋi = mi ẍi + mi ẋi . (14.13)
dt ∂qj ∂qj dt ∂qj
The first term in this equation contains ∂xi /∂qj . A useful relationship
involving generalized coordinates is
∂xi ∂ ẋi
= . (14.14)
∂qj ∂ q̇j
The last term in equation (14.13) involves the expression

d ∂xi ∂ ẋi
= . (14.15)
dt ∂qj ∂qj
Using equations (14.13), (14.14) and (14.15), equation (14.12) becomes,
X X d  ∂xi

∂ ẋi

ṗi δxi = mi ẋi − mi ẋi δqj ,
i i,j
dt ∂q j ∂q j

X d  ∂ ẋi

∂ ẋi

= mi ẋi − mi ẋi δqj . (14.16)
i,j
dt ∂ q̇j ∂qj

Now the kinetic energy is T = Σ(1/2)mẋ2i , so

  X  
∂T ∂ X 1 2 ∂ ẋi
= mi ẋi = mi ẋi ,
∂qj ∂qj i 2 i
∂q j

and
  X  
∂T ∂ X 1 2 ∂ ẋi
= mi ẋi = mi ẋi .
∂ q̇j ∂ q̇j i 2 i
∂ q̇j

Comparing with the right hand side of equation (14.16) we appreciate

that

X X  d ∂T ∂T

ṗi δxi = − δqj .
i j
dt ∂ q̇j ∂qj
 14.10. SUMMARY 509

Finally, d’Alembert’s principle reads

N X  d ∂T 
X X ∂T
Fiext


− ṗi δxi = Qj δqj − − δqj = 0,
i=1 j j
dt ∂ q̇j ∂qj
X d ∂T ∂T

0 = Qj − + δqj .
j
dt ∂ q̇ j ∂q j

Using the fact that the δqj are independent it is clear that each term
in the summation must individually be equal to zero, so
d ∂T ∂T
− = Qj . (14.17)
dt ∂ q̇j ∂qj
This derivation gives Lagrange’s equations in a form involving the gen-
eralized forces and the kinetic energy. You have seen this alternate
form of Lagrange’s equations previously. (See equation 4.12 in Worked
Example 4.6.)

Exercise 14.20. Starting with equation (14.17) and assuming the

generalized forces are derivable from a potential that is independent of
the generalized velocities, derive the Lagrange equations in their usual
form (equation 4.9).

14.10. Summary
This chapter on the statics of extended bodies started with a simple
analysis of the conditions for static equilibrium, namely
X
Fi = 0,
X
Ni = 0.
That is, the sum of the forces is zero and the sum of the torques is
zero. You were then exposed to a number of theorems and definitions
to familiarize you with the concepts and terms used in the study of
statics. Thus, for example, you were given the definitions of a couple,
a resultant, and an equilibrant. You found out that any system of
forces can be reduced to a single force plus a couple.
The next topic was more advanced, being a study of a hanging
cable. There were two applications, namely, a suspension bridge and a
510 14. STATICS (OPTIONAL)

rope hanging under its own weight. The shape of the cable suspending
a bridge is a parabola,
w 2
y= x + c,
2Tc
and the shape of a rope hanging under its own weight is a catenary
x
y = c cosh .
c
Hooke’s law states that stress and strain are proportional. Defining
stress as force per unit area, leads to the three expressions
Law Stress Proportionality Constant
F ∆l
A
=Y l tension Y = Young’s modulus
F ∆V
A
= −B V pressure B = Bulk modulus
F
A
= n tan θ shear n = Shear modulus
A consideration of the orientation of the surface across which the
force acts led to the concept of the stress tensor which was described
but not utilized in the analysis.
The centroid is a geometrical concept similar to the physical concept
of center of mass. The centroids of a volume, a surface and a line are
defined as
Z
1
Rvol = rdτ,
V
Z
1
Rsurf = rdA,
A
Z
1
Rline = rds.
S
Some properties of centroids were summarized in the two theorems of
Pappus.
The center of gravity is the point in an extended body of mass M
at which a point mass M exerts the same gravitational force on a point
mass m as does the extended body.
The next topic was a brief study of the equilibrium of a fluid, in
which you found that the pressure field is related to the body force
f =ρg by
f = ∇p.
Two important applications are Archimedes’ principle and Pascal’s law
which were applied to fluids in equilibrium and to the atmosphere of
Earth.
 14.11. PROBLEMS 511

Finally you were exposed to d’Alembert’s principle and virtual

work, a rather advanced topic in the study of statics. We defined virtual
work and noted that for a system in equilibrium the virtual work done
in a virtual displacement is zero. We also showed how d’Alembert’s
principle can be used to derive Lagrange’s equations.
14.11. Problems
Problem 14.1. A work of art consists of a rod of length L and mass
M. It is suspended by thin wires at either end. The wires are connected
to two walls and rod “floats” horizontally between the walls. The angles
formed by the wires with the horizontal are θ and φ. Determine the
angle formed by the rod with the horizontal. Determine the tensions
in the wires.
Problem 14.2. A rigid rod of mass M and length L is supported
from a single point by two strings attached to the ends of the rod. Each
string has length L. A weight, also having mass M is hung from one
end of the rod. Find the angle the rod makes with the horizontal and
determine the tensions in the strings. See Figure 14.14.

T1 LL
L
T2
α

Mg Mg

Figure 14.14. A suspended rod. See problem 14.2.

Problem 14.3. A smooth rod that is nailed to the floor and to the
wall, makes an angle of 30o to the horizontal, as shown in Figure ??.
A ring of mass m can slide on the frictionless rod. A string is passed
through the ring; one end of the string is attached to a point on the
floor and the other end of the string supports a weight of mass 3m.
Determine the angle φ between the two segments of the string. Note,
the rod does not move, but the ring can slide along it.
Problem 14.4. A rigid flat plate 2 meters on a side is acted upon
at its corners by the following forces: F1 = 10̂ acting at (-1,1), F2 =
5ı̂−5̂ acting at (1,-1), F3 = 10̂ acting at (-1,-1), and F4 = 10ı̂ at
(1,1). The forces are in newtons. The center of the plate is at (0,0).
Determine the resultant and the equilibrant.
512 14. STATICS (OPTIONAL)

m 3mg
30o
ϕ 3m 4mg

Figure 14.15. The sketch on the left shows the setup

and the sketch on the right shows the forces on the ring
of Problem 14.3.

Problem 14.5. An open cylindrical pipe of mass M and radius a

is set upright on a table. Two billiard balls of mass m and radius r are
dropped into the pipe. Assume r > a/2. Show that the pipe will not
tip over if M ≥ 2m(1 − ar ).
Problem 14.6. A stepladder is made from two ladders of length
l and mass m by joining them at the top with a frictionless pivot. A
horizontal string of length s is attached at points S, S 0 , a third of the
way up the ladders. The system rests on a smooth floor. What is the
tension in the string? See Figure 14.16.

S S'

Figure 14.16. A stepladder.

Problem 14.7. Consider the rod of Figure 14.1. If the string with
tension T2 breaks, the rod will rotate and translate until it is hanging
straight down. Determine the equilibrant of the forces acting on the
rod at the instant the string breaks.
Problem 14.8. A ladder of weight W and length l stands on a
smooth floor and leans against a smooth wall. The angle between the
ladder and the floor is α. As you might expect, since there is no friction,
the ladder is slipping. What is the equilibrant? (The expression you
will obtain includes the normal forces)
 14.11. PROBLEMS 513

Problem 14.9. An equilateral triangle, 1 meter on a side, is sub-

jected to three 10 newton forces as shown in Figure 14.13. (a) Describe
the motion of the object. (b) Find the equilibrant.

Figure 14.17. An equilateral triangle subjected to

three forces of equal magnitude.

Problem 14.10. A cube of side a has one corner at the origin and
the diagonally opposite corner at (a, a, a). The sides are oriented along
the axes. The forces acting on the cube are all equal in magnitude to
F. The force on the corner at (0, 0, 0) is Fı̂, the force on the corner at
(0, a, a) is F̂ and the force on the corner at (a, a, a) is F k̂. Find an
equivalent single force and couple. (See Figure 14.14.)

z
a

a
x

Figure 14.18. A cube subjected to forces of equal mag-

nitude but different directions at some of its corners.

Problem 14.11. Expanding the equation for a catenary (equation

14.2) show that if the tension is large, y ' c + x2 /2c.
Problem 14.12. Consider a suspended cable in which the cross
sectional area of the cable varies, being smallest at the midpoint and
greatest at the points of support (assumed to be at the same height).
The thickness of the cable is designed such that the tension per unit
area of the cross section is constant along the cable. That way it will
have equal probability of breaking at all points. Note that for such a
cable the tension is proportional to the area and hence to the weight per
unit length, w. That is, T = kw where k is a constant of proportionality
514 14. STATICS (OPTIONAL)

and w varies along the length of the cable. (a) Show that such a cable
forms a curve given by
x
y = k log sec + c
k
where c is a constant of integration. (b) Show the maximum possible
span for such a cable is πk.
Problem 14.13. Using the Taylor’s series expansion for the po-
tential, show that it leads to Hooke’s law for small displacements from
equilibrium. (For simplicity, you may assume a one dimensional defor-
mation.)
Problem 14.14. Obtain a formula for the volume of a torus. (Hint:
Use a theorem of Pappus.)
Problem 14.15. Two uniform spheres of mass M are located at
x = ±a. Determine the center of gravity relative to a point P on the
y-axis. Show that the center of gravity and the center of mass approach
one another as the distance to P increases.
Problem 14.16. A satellite consists of two spheres (which we ap-
proximate as particles) each having mass m. They are connected by a
massless rod of length l. The satellite is in a circular orbit of radius
r about the Earth with both masses lying along the same radial line
from the center of Earth. Determine the distance h between the center
of mass and the center of gravity of this satellite.
Problem 14.17. A cylindrical rod of length 20 km is held per-
pendicular to the surface of Earth. Determine the distance between
its center of mass and center of gravity. (Assume the Earth is a point
mass located at its center.) Hint: Use the binomial expansion.
Problem 14.18. The Earth is an oblate spheroid so its equato-
rial radius is somewhat greater than its polar radius. The following
“thought experiment” will allow you to estimate the difference between
these two radii. Imagine a tunnel is drilled from the North Pole to the
center of the Earth where it makes a ninety degree turn and emerges
at the equator. The tunnel is filled to the brim with water. How much
longer is the equatorial tunnel than the polar tunnel if the density of
the Earth is assumed constant? Next solve the problem if the density
of the Earth varies linearly with depth and that the density at the
center is ρ0 .
Problem 14.19. Under some circumstances the lower atmosphere
can be considered adiabatic. Assuming an adiabatic atmosphere, the
 14.11. PROBLEMS 515

relation between temperature and pressure is

dT 2T
=
dP 7P
(Roughly speaking, air is an ideal diatomic gas with molecular weight
29.) Obtain an expression for the lapse rate, that is the rate at which
the temperature decreases with altitude.
Problem 14.20. Prove the relationship given in equation (14.14),
that is, show that for any set of generalized coordinates,
∂xi ∂ ẋi
= .
∂qj ∂ q̇j
Problem 14.21. Show that
X X
Fi δxi = Qj δqj
i j

Problem 14.22. Assume that the two planes of Figure ?? are not
smooth, and the coefficient of friction is µ. The system is in equilib-
rium. Use the principle of virtual work to obtain an expression for the
coefficient of friction.

COMPUTATIONAL PROBLEMS
Computational Techniques: Solving a System of Linear Equations
Statics problems often involve solving a system of linear equations.
Computer Algebra Systems, such as Matlab, usually have built in func-
tions that allow one to solve such equations very simply. For example,
consider the system of three equations in three unknowns
a11 x1 + a12 x2 + a13 x3 = b1 ,
a21 x1 + a22 x2 + a23 x3 = b2 ,
a31 x1 + a32 x2 + a33 x3 = b3 .
Here the a0 s and b0 s are known quantities and one wishes to determine
the x0 s. Using matrix notation, these equations can be written
    
a11 a12 a13 x1 b1
 a21 a22 a23   x2  =  b2 
a31 a32 a33 x b3
or
Ax = B.
Then, in the syntax of Matlab,
x =A\B.
516 14. STATICS (OPTIONAL)

(Note the backslash. Without going into the details, the process of
solving for the vector x uses a technique called Gaussian elimination
that basically solves the first equation for x1 and substitutes it into the
remaining equations. Then it solves the second equation for x2 and
substitutes it into the remaining equations, and so on, until the last
x is obtained in terms of the coefficients. Finally, “back substitution”
generates the values of all the x0 s.)
Computational Project 14.1. In Figure 14.14 assume the lengths
of the rod and strings are 80 cm. Let the angle α be 15◦ and the let
the hanging rod have mass 12 kg. Determine T1, T2 , and the mass of
the rod.
Computational Project 14.2. Meteorologists often plot the
variation of meteorological quantities as a function of the logarithm
of the pressure, decreasing with altitude. Generate such a plot for the
temperature as a function of pressure of the atmosphere assuming the
temperature decreases adiabatically, that is that the pressure and tem-
perature are related by P 1−γ T γ = constant. You can let γ = 1.4 and
assume the pressure and temperature at the surface are 1 atmosphere
and 298 kelvin.
Computational Project 14.3. We found
the equation for a rope
x
hanging under its own weight is y = c cosh c . Recall that
1 u
e + e−u .


cosh u =
2
u −u 1 u −u
Plot e and e and 2 (e + e ) to generate a catenary. Show that
the lowest point in the catenary is a distance c above the origin.
Chapter 15
Rotational Kinematics

This chapter and the next are a study of the general motion of a
rigid body. This is a fairly complicated topic which involves mathe-
matical concepts that you may not have encountered before.
We have already considered the rotation of a symmetrical body
about a fixed axis. We now generalize to rotations about a fixed point.
For example, consider a spinning top. A top is a symmetrical body that
rotates about its axis of symmetry. When it is spinning very fast, the
top remains upright, and as long as the top is not slipping, its center of
mass is at rest. But as the top slows down, it leans over. The center of
mass traces a circle around the vertical. The axis of rotation precesses
about this vertical. Eventually, the top will begin to nutate or “nod”
up and down as it precesses. If the top is not sliding, it has one fixed
point, namely the point that is in contact with the ground.
Another example of rotation about a fixed point is illustrated in
Figure 15.1 which shows a disk with an axle passing through it. In this
case, the axle is not perpendicular to the disk. If the system is not
supported in any way (imagine it is a poorly designed satellite) this
nonsymmetrical body will wobble while it rotates. The axis of rotation
does not have a fixed direction. If the center of mass is at rest, this is
a rotation about a fixed point.
Analyzing the motion of a body rotating about a fixed point involves
developing a system of rotational kinematics. Recall that in Chapter 2
you studied the kinematics of a point particle. This entailed learning
how to describe the position (as well as the velocity and acceleration) of
a particle in terms of Cartesian, cylindrical and spherical coordinates.
In a similar manner, you will now learn how to describe the orientation
of a rigid body and how this changes as the body rotates.

517
518 15. ROTATIONAL KINEMATICS

Figure 15.1. A body that is not symmetric about an

axis will tend to wobble when it rotates.

From a mathematical point of view, the rotation of a rigid body

can be considered an orthogonal transformation. The analysis of or-
thogonal transformations, as you will see, leads to a consideration of
Euler angles and Euler’s theorem. Euler’s angles represent three rota-
tions that carry a body from any given initial orientation to any final
orientation. Euler’s theorem goes one better and tells us that we can
carry a body from any initial orientation to any final orientation with
a single rotation. (I feel obliged to warn you that some of this material
is pretty tough going!)
15.1. Orientation of a Rigid Body
All the points in a rigid body maintain a fixed relative distance from
one another. This means that the distance between any two points i
and j is given by the constraint equation
|rj −ri | = rij = constant.
Suppose the rigid body is rotating about a fixed point. That is,
there is one point in the body that is fixed in space and all the other
points in the body can move about it in any way that is consistent
with the set of constraints above.1 Our first task is to describe the
orientation of the body relative to the fixed in terms of the smallest set
of independent (generalized) coordinates.
Six coordinates are sufficient to completely describe the position
and orientation of a rigid body. (I will now show you that this state-
ment is true; note that the explanation will require you to do some
abstract reasoning.) To specify the orientation of the body you need
to know the relative positions of three non-collinear points. Three co-
ordinates are required to locate a particular point P in the body - this
point will often be the center of mass. You need to know the posi-
tions of two more points relative to P. The second point is at some
1For
a body rotating about a fixed axis, there is the additional constraint that
all the points on the axis are fixed in space and all off-axis points move in circles
about the axis. However, the present problem considers a body with only one fixed
point (possibly the center of mass).
 15.1. ORIENTATION OF A RIGID BODY 519

given distance d from P so no matter how the body is rotated about

P, the second point must lie somewhere on the surface of a sphere of
radius d with center at P. Two coordinates are sufficient to specify the
position of a point on the surface of the sphere so you need two coor-
dinates to determine relative position of this second point. You still
need to specify the position of a third point relative to the first two.
Draw a line between the first two points. Now the body can be rotated
about an axis through these points without changing their positions.
Therefore, the third point must lie on a circle centered on this line;
to locate it requires a single coordinate. Consequently, a total of six
independent coordinates are sufficient to specify both the position and
the orientation of the body.
If some point, such as the center of mass, is fixed at some known
position, you only need three coordinates to describe the orientation of
the body. Let the center of mass be the origin of an inertial Cartesian
coordinate system denoted by x, y, z. Let x0 , y 0 , z 0 be another Carte-
sian coordinate system that is fixed in the body and rotates with the
body. The two systems are illustrated in Figure 15.2. The orienta-
tion of the body is given by the orientation of the rotating (primed)
system relative to the inertial (unprimed) system. The orientation of
x0 , y 0 , z 0 relative to x, y, z can be described using just three independent
quantities. Another way to say this is to point out that x, y, z can be
transformed into x0 , y 0 , z 0 with three rotations.

z' z
y'

x'

y
x
Figure 15.2. An inertial reference frame (x, y, z) and a
reference frame (x0 , y 0 , z 0 ) fixed in the body and rotating
with it.
0
Denote the unit vectors for the two systems by ı̂,̂, k̂ and ı̂0 ,̂0 , k̂ .
The direction of the x0 -axis with respect to the unprimed axes is given
by its direction cosines α1 , α2 , α3 , defined by
α1 = cos(x0 , x) = ı̂0 ·ı̂, (15.1)
α2 = cos(x0 , y) = ı̂0 · ̂,
α3 = cos(x0 , z) = ı̂0 · k̂.
520 15. ROTATIONAL KINEMATICS

Similarly, the directions of the y 0 - and z 0 -axes are determined by their

direction cosines. The nine direction cosines (three for each of the
primed axes) give the orientation of the primed (or “body”) axes with
respect to the fixed (or inertial) axes. However, nine quantities are not
needed to describe the relative orientation of the two sets of axes; as
you know, three are sufficient. This means that not all of the direction
cosines are independent.
Just as αi (i = 1, 2, 3) denotes the direction cosines for the x0 -axis,
the directions cosines for the y 0 - and z 0 -axes can be denoted by βi and γi
with definitions analogous to Equations (15.1). Using the orthogonality
of the unit vectors, it is easy to show that
αl αm + βl βm + γl γm = δlm (l, m = 1, 2, 3), (15.2)
where δlm is the Kronecker delta and is equal to unity if l = m and
equal to zero otherwise. Equations (15.2) are known as the “orthogo-
nality conditions.”2

Exercise 15.1. Write expressions analogous to Equation (15.1) for

βi and γi .
Exercise 15.2. Prove relations (15.2).

15.2. Orthogonal Transformations

Consider the axes illustrated in Figure 15.2. The position of a point
in space is given by (x, y, z) in the inertial system and by (x0 , y 0 , z 0 ) in
the rotating system. The relation between these coordinates is
x0 = α1 x + α2 y + a3 z, (15.3)
y 0 = β1 x + β2 y + β3 z,
z 0 = γ1 x + γ2 y + γ3 z.
To make the notation easier to handle, it is traditional to write x, y, z,
as x1 , x2 , x3 , and instead of x0 , y 0 , z 0 , one writes x01 , x02 , x03 . Furthermore,
instead of α1 we write a11 , instead of α2 we write a12 , instead of β3 we
2The word “orthogonality” implies that one thing is perpendicular to another.
In this case, we are just noting that the unit vectors are mutually perpendicular. In
the case of orthogonal matrices the geometric concept of perpendicular quantities
is harder to visualize.
 15.2. ORTHOGONAL TRANSFORMATIONS 521

write a23 , and so on, so that the transformation equations (15.3) are
expressed as
x01 = a11 x1 + a12 x2 + a13 x3 , (15.4)
x02 = a21 x1 + a22 x2 + a23 x3 ,
x03 = a31 x1 + a32 x2 + a33 x3 .
It is convenient to use the Einstein summation convention to express
equations (15.4) in the very compact form
x0i = aij xj . (15.5)
The Einstein summation convention states that if a term contains the
same subscript
P twice, a summation over that subscript is implied. Thus
aij xj = j aij xj .
The orthogonality relationships (Equations 15.2) are then simply
aij aik = δjk . (15.6)
Note that the transformation from the unprimed to the primed coor-
dinates is carried out using the coefficients aij . The aij are related to
each other by the orthogonality conditions given by Equation (15.6).
Therefore, the transformation (15.4) is called an “orthogonal transfor-
mation.”
You will show in Problem 15.1 that the orthogonality relation (15.6)
leads to
x0i x0i = xi xi . (15.7)
But xi xi = x2 + y 2 + z 2 is the square of the length of the vector
xı̂+y̂+z k̂. So Equation (15.7) tells us that an orthogonal transforma-
tion conserves the length of a vector.
The coefficients aij lend themselves quite naturally to being ex-
pressed as a matrix, thus
 
a11 a12 a13
A =  a21 a22 a23  . (15.8)
a31 a32 a33
It is convenient at this time to express vectors as column matrices. For
example, the position vector in the inertial system is denoted x and is
expressed as a column matrix as follows:
   
x x1
x =  y  =  x2  .
z x3
522 15. ROTATIONAL KINEMATICS

In the rotating system the position vector is denoted by x0 where

 0   0 
x x1
x0 =  y 0  =  x02  .
z0 x03
Note that x and x0 are, in this case, the same vector, but expressed in
terms of two different coordinate systems. That is, the components of
x are (x, y, z) whereas the components of x0 are (x0 , y 0 , z 0 ).
The transformation equation (15.4) relating x and x0 can now be
written in the compact matrix form,
x0 = A x. (15.9)

Worked Example 15.1. Determine the transformation ma-

trix A for a two dimensional rotation.
Solution: Consider point P in Figure 15.3. The coordinates
of P in the unprimed system are (x, y) and the coordinates of P in
the primed system are (x0 , y 0 ). The transformation equations are
x0 = x cos θ + y sin θ, (15.10)
y 0 = −x sin θ + y cos θ.
If we write x0 = A x, then
 
cos θ sin θ
A= . (15.11)
− sin θ cos θ

y
y' P
x'
r
θ
θ x
Figure 15.3. The coordinates of point P in the two
reference frames.

Exercise 15.3. Carefully draw axes such as those shown in Figure

15.3 and determine the inverse relations to equation (15.10), that is,
obtain x = x(x0 , y 0 ) and y = y(x0 , y 0 ).
 15.2. ORTHOGONAL TRANSFORMATIONS 523

Exercise 15.4. Show that if we represent vectors by row and col-

umn matrices, the dot product of A and B can be written
 
Bx
( Ax Ay Az )  By  .
Bz

Interpretation
The matrix A describes the transformation from the unprimed sys-
tem to the primed system. Let us change the notation of the position
vector back to r. If (x, y) and (x0 , y 0 ) represent the components of the
vector r in the two coordinate systems, then A tells us how the two
sets of components are related. That is, A describes the result of a
counterclockwise rotation of the coordinate system through the angle
θ. This is called the “passive” interpretation of A, and is illustrated in
Figure 15.3.
There is, however, a different interpretation in which A is considered
to be an operator that rotates the vector r in a clockwise manner to
generate a new vector r0 . In this interpretation, the coordinate system
is fixed and (x0 , y 0 ) are the components of the new vector, as shown in
Figure 15.4. This is, of course, called the “active” interpretation.

r
y

y' r'

x x'
Figure 15.4. The “active” interpretation assumes the
vector rotates in a clockwise sense rather than the coor-
dinate system rotating in the counter-clockwise sense.

15.2.1. Orthogonal Matrices (Optional). Let us consider in

more detail the transformation matrix A . Start with
r0 = A r. (15.12)
This can be written as X
x0i = aij xj , (15.13)
j
524 15. ROTATIONAL KINEMATICS

or, using the summation convention,

x0i = aij xj . (15.14)
(I will be using the summation convention frequently in what follows:
watch for repeated indices.) The inverse transformation from r0 to r
can be obtained by multiplying both sides of Equation (15.12) by the
inverse of A, denoted A−1 . By definition,
AA−1 = A−1 A =1,
where  
1 0 0
1= 0 1 0 
0 0 1
is the unit matrix. Thus
A−1 r0 = A−1 (Ar) = A−1 Ar = r,
or
r =A−1 r0 .
Now by the rules of matrix multiplication, if C = A·B, then
cjk = aji bik . (15.15)
Consequently, if
A−1 A =1,
then
a−1
ji aik = δjk . (15.16)
But recall that the orthogonality condition (Equation 15.6) can be
written as
aij aik = δjk . (15.17)
Comparing (15.17) with (15.16) you will appreciate that
a−1
ji = aij .

That is, the elements of the inverse matrix are obtained by transposing
the rows and columns of the original matrix. The transposed matrix is
denoted by AT . Thus, for an orthogonal matrix,
AT = A−1 , (15.18)
and hence,
AT A = 1. (15.19)
In physics we are often interested in the matrix obtained by taking the
transpose of the original matrix and then replacing each element by its
 15.2. ORTHOGONAL TRANSFORMATIONS 525

complex conjugate. The resulting matrix is called the adjoint matrix

and is usually denoted by A† .

∗
A† = AT . (15.20)
If A† is the inverse of A, that is, if
A† A = 1, (15.21)
then A is said to be unitary.
If A† is equal to A , that is, if
A† = A, (15.22)
then A is said to be hermitian or “self adjoint.”
One of the properties of the orthogonal matrices we are using to
represent rotations is the fact that they have a determinant of +1.
This important property is easily proved. Taking the determinant of
Equation (15.19) yields3
 T
A  |A| = |1| .

But the determinant of the transpose is the same as the determinant

of the original matrix because the determinant is not affected by the
interchange of rows and columns. Therefore,
|A|2 = 1,
and hence |A| = ±1.
The table below summarizes the names and properties of these var-
ious kinds of orthogonal matrices

Orthogonal Matrices
Name Symbol Element Property
Orthogonal A aij |A|2 = 1
Inverse A−1 a−1
ij = aji A−1 A =1
Transpose AT aTij = aji AT A =1
Adjoint A† aij = a∗ji
Unitary A† = A−1
Hermitian A† = A

3The determinant of a product of matrices is the product of the determinants

of the matrices. (Say that three times quickly!) In other words,
|AB| = |A| |B| .
526 15. ROTATIONAL KINEMATICS

The determinant of A is either plus or minus unity. The value +1

corresponds to a rotation and the value −1 corresponds to an inversion
of the coordinate axes. (An inversion is a reflection of all the axes.)
This can be easily appreciated by considering the matrix S that has a
determinant −1 :  
−1 0 0
S =  0 −1 0 .
0 0 −1
Then, if r0 = Sr,
 0      
x −1 0 0 x −x
 y 0  =  0 −1 0   y  =  −y  .
0
z 0 0 −1 z −z
That is,
x0 = −x; y 0 = −y; z 0 = −z.
Thus, the operator S inverts the axes, changing a right-handed coor-
dinate system into a left-handed coordinate system. Any matrix with
determinant −1 generates an inversion because it can be written as the
product of S and a matrix with determinant +1. Therefore, an orthog-
onal matrix representing a rotation must have a determinant equal to
+1.
If the matrix B is considered an operator that generates a rotation of
the coordinate system, the representation of a vector V in the rotated
coordinate system is
V0 = B V.
In the passive interpretation, this is the transformation law giving the
relationship between the components of the vector in the rotated coor-
dinate system and the components of the same vector in the original
system.
A matrix (call it A) is described by a set of elements aij and these
will, in general, be different in the rotated coordinate system than in
the original system. Let us determine the transformation law for a
matrix under the rotation of the coordinate system.
Assume that in some coordinate system the matrix A acting on the
vector V produces the vector W.
W =AV. (15.23)
Let B represent a rotation of the coordinate system. In terms of
the rotated set of axes, the vector W is given by
W0 = BW =BAV (15.24)
 15.2. ORTHOGONAL TRANSFORMATIONS 527

But in the rotated coordinate system V is given by V0 = BV so it is

desirable to multiply V by B. This can be accomplished by introducing
the unit matrix B −1 B into (15.24) thus:
W0 = BW =BA(B −1 B)V = (BAB −1 )BV = (BAB −1 )V0 . (15.25)
−1
This equation shows that the matrix BAB transforms the vector
V0 into W0 . That is, Equation (15.25) describes the same operation
as Equation (15.23), but referred to the rotated coordinate system.
Therefore, the operator A in the rotated system is represented by
A0 =BAB −1 . (15.26)
This is the transformation law for matrices. It is called a similarity
transformation. In general, you can consider a similarity transforma-
tion as a way to generate a new matrix that has certain properties;
there is no need to interpret it in geometrical terms as the result of
a rotation. For example, you can use a similarity transformation to
transform a matrix into diagonal form. It is left as a problem to show
that a similarity transformation preserves both the determinant and
the trace of a matrix.

Worked Example 15.2. Consider a matrix B that represents

a rotation about the z-axis through an angle φ. Using B carry out a
similarity transformation of the matrix A of Equation (15.11) and
interpret the result.
Solution: Write
 
cos φ sin θ
B=
− sin φ cos φ
and  
cos θ sin θ
A= .
− sin θ cos θ
Then,
−1
A0 = BAB
   
cos φ sin θ cos θ sin θ cos φ − sin θ
=
− sin φ cos φ − sin θ cos θ sin φ cos φ
where we used the fact that B is orthogonal so B −1 = B T . Carrying
out the matrix multiplications and simplifying we end up with
 
0 cos θ sin θ
A =
− sin θ cos θ
528 15. ROTATIONAL KINEMATICS

which tells us that the rotation matrix is unchanged by this rotation

of the axes.
Exercise 15.5. Consider the transformation matrix, Equation (15.11).
(a) Is this matrix orthogonal? (b) Is it unitary? (c) Is it hermitian?

15.3. The Euler Angles

We now return to the question of describing the orientation of a
rigid body. As usual we let x0 , y 0 , z 0 represent a coordinate system
embedded in the body and rotating with it, and we let x, y, z represent
an inertial coordinate system. For convenience, we assume the two
systems have a common fixed origin.

Figure 15.5. The first Euler angle: a rotation through

an angle φ about the z-axis.

The orientation of x0 , y 0 , z 0 with respect to x, y, z can be described

in terms of the nine direction cosines denoted aij . These are related by
the orthogonality conditions (15.6), so they are not all independent.
As discussed above, if the origins of the two systems coincide, there are
only three independent coordinates. The Euler angles are a set of three
independent coordinates that describe the orientation of x0 , y 0 , z 0 with
respect to x, y, z. Specifically, these angles describe three rotations that
carry the axes x, y, z into x0 , y 0 , z 0 .
The first Euler angle is a rotation about the z-axis through φ. This
causes x and y to be transformed to a new pair of axes, call them ξ and
η. This rotation is illustrated in Figure 15.5. This procedure defines a
new coordinate system ξ, η, ζ. Note that ζ lies along the original z-axis
and ξ and η lie in the same plane as x, y.
 15.3. THE EULER ANGLES 529

Next, carry out a rotation through θ about the ξ-axis. This causes
the ζ-axis and the η-axis to change direction as indicated in Figure
15.6. The axes resulting from this rotation are denoted ξ 0 , η 0 , ζ 0 . The
old xy-plane is now the ξ 0 η 0 -plane and is inclined to the original plane
by an angle θ. (The line from the origin along the ξ- (or ξ 0 -) axis is
called the line of nodes.)

ϕ y
x

Figure 15.6. The second Euler angle: a rotation

through θ about the ξ-axis.

Finally carry out a rotation about the ζ 0 -axis through ψ. This oper-
ation leaves ζ 0 unchanged but carries ξ 0 to x0 and η 0 to y 0 , as indicated
in Figure 15.7. The new coordinate system is x0 , y 0 , z 0 .
It is easy to appreciate that the three Euler angles φ, θ, ψ can be
used to carry the original axes x, y, z to any desired final orientation
x0 , y 0 , z 0 .

y
x

Figure 15.7. The third Euler angle: a rotation through

ψ about the ζ 0 axis.

I described the three Euler rotations in words. Figures 15.5, 15.6

and 15.7 illustrate the rotations graphically. Now I will show you how to
530 15. ROTATIONAL KINEMATICS

represent these rotations mathematically. A rotation about a fixed axis

is described by a matrix such as given by Equation (15.8). Therefore,
the first rotation (through φ about z) can be expressed mathematically
as  
cos φ sin φ 0
D =  − sin φ cos φ 0  . (15.27)
0 0 1
(Note the similarity to the rotation matrix of Equation 15.11.) You
can easily verify that,
   
ξ x
 η  = D y . (15.28)
ζ z
For simplicity let us write this as
ξ =Dx.
The next rotation is through θ about ξ. This rotation leaves the ξ-axis
unchanged, and it is expressed by the matrix
 
1 0 0
C= 0 cos θ sin θ  . (15.29)
0 − sin θ cos θ
Letting
 
ξ0
ξ 0 =  η0  ,
ζ0
we can write
ξ 0 = Cξ.
Finally, the last rotation is through ψ about ζ 0 . This has the same
structure as D, so we write
 
cos ψ sin ψ 0
B =  − sin ψ cos ψ 0  . (15.30)
0 0 1
Again, letting
 
x0
x0 =  y 0  ,
z0
we have
x0 = Bξ 0 .
 15.3. THE EULER ANGLES 531

The transformation from x, y, z to x0 , y 0 , z 0 is a sequence of three

rotations described by the three matrices D, C, B. Defining the matrix
A by
A = BCD
the total transformation is given by
x0 = Ax. (15.31)
The transformation matrix A is the matrix product of three matrices
and is rather complicated. As you can verify by some rather tedious
matrix multiplication, it is given by
 
cos φ cos ψ− sin φ cos θ sin ψ sin φ cos ψ+ cos φ cos θ sin ψ sin θ sin ψ
A =  − cos φ sin ψ− sin φ cos θ cos ψ − sin φ sin ψ+ cos φ cos θ cos ψ sin θ cos ψ 
sin φ sin θ − cos φ sin θ cos θ
(15.32)
Note that A involves only the independent coordinates φ, θ, ψ.
An alternate set of angles describing rotations about three different
axes is often used in aeronautical engineering. Those three angles are
called roll, pitch, and yaw.

Worked Example 15.3. A student is told to determine the

orientation of the coordinate axes if the Euler angles are 30◦ , 45◦ ,
and 60◦ . The student does the rotations in reverse order. What
happens?
Solution: The appropriate angles are φ = 30◦ , θ = 45◦ and
ψ = 60◦ . So,  
0.13 0.78 0.61
A =  −0.61 −0.13 0.35  .
0.35 0.61 0.71
However, if the rotations are taken in reverse order, then φ = 60◦ ,
θ = 45◦ , and ψ = 30◦ , and
 
0.13 0.93 0.35
A0 =  −0.78 −0.13 0.61  .
0.61 −0.35 0.71
Thus, the correct directions for the new coordinates axes are given
by
x0 = 0.13x + 0.78y + 0.61z,
y 0 = −0.61x − 0.13y + 0.35z,
z 0 = 0.35x + 0.61y + 0.71z.
532 15. ROTATIONAL KINEMATICS

and the wrong directions are given by

x0 = 0.13x + 0.93y + 0.35z,
y 0 = −0.78x − 0.13y + 0.61z,
z 0 = 0.61x − 0.35y + 0.71z.

Exercise 15.6. Verify Equation (15.28)

Exercise 15.7. Verify Equation (15.32).
Exercise 15.8. Describe in words the three rotations denoted roll,
pitch, and yaw.
Exercise 15.9. Write the 3×3 matrix for a rotation about the z-
axis through 30◦ , followed by a rotation about the new x-axis through
45◦ .  
0.866 0.500 0
Answer:  −0.354 0.612 0.707  .
0.354 −0.612 0.707

15.4. Euler’s Theorem

So far we have been studying the rotation of a rigid body with one
fixed point and you have seen that such a rotation is described by the
matrix A where
x0 = Ax.
In other words, if the position of a point P in the body is given by x
before the rotation, it is given by x0 after the rotation.4
As you know, any desired orientation of a rigid body can be achieved
through three rotations (the Euler angles). But experience suggests
the possibility of reaching any given orientation by a rotation through
a single angle about some appropriately chosen axis. This concept is
formulated more precisely by Euler’s theorem of rigid body motion:
Theorem 7. (Euler’s Theorem): The general displacement of a
rigid body with one fixed point is a rotation about some axis.
This theorem states that as long as one point in the body is fixed,
it is possible to find some axis such that a single rotation about that
axis will carry the body to the desired orientation. (Note that you still
need three parameters to describe the rotation. Two of these give the
4Notethat A can be a function of time so A = A(t) and A(t = 0) = 1. Thus,
A(t) leads to a description of the orientation of the body at any instant of time.
 15.4. EULER’S THEOREM 533

direction of the axis of rotation and the third gives the angle through
which the body is rotated.)
A rotation has two important properties: (1) The length of a vec-
tor is unchanged by a rotation, and (2) A vector lying on the axis of
rotation is unchanged in direction and magnitude. If the matrix A
describes a rotation, then property (1) will be satisfied by requiring A
to be an orthogonal matrix (see Equation 15.7). To find the conditions
for property (2) to be satisfied, consider a vector R lying along the
axis of rotation. Then R0 = R, where R0 is the vector in the rotated
system. But R0 = AR. Consequently,
AR = R. (15.33)
The matrix A represents a rotation from an initial orientation to some
final orientation. A may have been generated by the three rotations
through Euler angles, for example. What Euler’s theorem states is
that the same final orientation can be achieved by a single rotation
about some axis. If this is so, then a vector that is unchanged by the
rotation must lie on the axis of rotation. That is, there exists a vector
R satisfying Equation (15.33).
But (15.33) is just a special case of the more general equation
AR =λR. (15.34)
where λ is a real or complex scalar constant. Equation (15.34) is called
an eigenvalue equation; λ is called the eigenvalue and R is called the
eigenvector. The eigenvalue problem consists in finding the eigenvalues
and eigenvectors for a given matrix (or operator) A.
Euler’s theorem requires that AR = R. Therefore, the theorem is
proved if we can show that any orthogonal matrix representing a rota-
tion will have at least one eigenvalue equal to +1.
Let us write Equation (15.34) as
(A − λ1) R =0, (15.35)
and express R as a column vector,
 
X
R = Y .
Z
Expanding Equation (15.35),
(a11 − λ)X + a12 Y + a13 Z = 0, (15.36)
a21 X + (a22 − λ)Y + a23 Z = 0,
a31 X + a32 Y + (a33 − λ)Z = 0,
534 15. ROTATIONAL KINEMATICS

where X, Y, Z are the components of the vector R. The problem of

finding a vector that is unchanged by the rotation is solved by deter-
mining a vector R corresponding to the eigenvalue +1. Actually, it
is not necessary to find the vector itself, but only its direction. This
means that it is sufficient to find the relative values of the components
of R. Note from equation (15.34) that any multiple of an eigenvector
is also an eigenvector. Therefore, the magnitude of an eigenvector is
not a particularly important property. However, the direction of the
eigenvector is important.
Homogeneous Algebraic Equations. Equations (15.36) for the
components of R are a set of three coupled linear homogeneous alge-
braic equations. You ran into the problem of solving for a set of two
such equations while studying coupled oscillators. I will review some
of the main points in obtaining a solution, but if you are unclear on
the concepts you should reread the material in Section 11.5.
Using Cramer’s rule to solve coupled homogeneous equations leads
to the so-called “trivial solution,” X = 0, Y = 0, Z = 0 unless the
determinant of the coefficients in equation (15.36) is zero. That is, a
nontrivial solution is possible if and only if
 
 a11 − λ a12 a13 
 
D=
 a21 a22 − λ a23  = 0.

 a31 a32 a33 − λ 
If the determinant of a matrix is zero, it means that one (or more) of
the rows or columns is not linearly independent of the others. That is,
D = 0 indicates that one of the three homogeneous equations can be
obtained from the other equations. For example, one equation might
simply be a multiple of another equation or it might be the sum of
the other two equations. Then one of the equations does not give any
information that is not contained in the other two equations. That
equation can be discarded. Suppose, for the sake of argument that the
last equation in the set is the one which is not linearly independent, and
that it has been discarded. You started out with three homogeneous
equations in three unknowns. You now have two equations in three
unknowns (X, Y, Z). This is not a good situation to be in! If, however,
you divide each equation by Z, you end up with the following two
equations
X Y
(a11 − λ) + a12 = −a13 , (15.37)
Z Z
X Y
a21 + (a22 − λ) = −a23 ,
Z Z
 15.4. EULER’S THEOREM 535

which is a set of two inhomogeneous equations in the two unknown

ratios that you can solve using Cramer’s rule! (This technique also
works for much larger systems of linear homogeneous coupled algebraic
equations. If the new set of equations are also homogenous just repeat
the process.)
What is the point of all this? The point is that the equations only
have the trivial solution X = Y = Z = 0 unless the determinant of
the coefficients is zero. That is, a nonzero solution requires that
 
 a11 − λ a 12 a 13

 
 a21
 a 22 − λ a 23
 = 0.
 (15.38)
 a31 a32 a33 − λ 
This condition guarantees that a nontrivial solution for the ratios X/Z
and Y /Z can be obtained. Furthermore, Equation (15.38) can be solved
for λ. The cubic equation for λ is called the secular equation. There
will, of course, be three roots; let us denote them λ1 , λ2 , and λ3 . These
can each be substituted into Equations (15.37) to solve for the ratios
X/Z and Y /Z. Note that in general you will obtain three solutions,
each corresponding to one of the three values of λ. (Remember that
you want to show that one of these eigenvalues is +1.)
Knowing the ratios X/Z and Y /Z (or equivalently X : Y : Z)
does not actually specify the vector with components (X, Y, Z), but
it does give the direction of this vector and this is all you really need
to know. Specifically, you have determined a vector with components
(X/Z, Y /Z, 1). This vector may not have the same magnitude as the
vector you set out to find and it might even be pointing in the opposite
sense, but at least it has the correct direction, and all you set out to
determine was the axis about which to rotate the body.
For each of the three roots λ1 , λ2 , and λ3 (the three eigenvalues) you
will get a different set of components. Therefore, in general, you get
three different vectors corresponding to the three different eigenvalues.
The vector obtained using λ1 , for example, is called the eigenvector
corresponding to λ1 .
Since the ratios of the components give the direction of the eigen-
vector, you can use this information to generate a unit vector êi in
that direction. Similarly, you can obtain unit vectors in the other two
directions. These unit vectors can be denoted by

ê1 , ê2 , ê3 ,

and they specify the directions of the eigenvectors. The three lines that
lie in these directions are called the principal axes of the matrix A.
536 15. ROTATIONAL KINEMATICS

The eigenvectors of A can be used to generate a useful matrix by

defining S to be the matrix whose columns are the eigenvectors of A.
For example, suppose that the eigenvector corresponding to λ1 , the
first eigenvalue of A, is
 (1)   (1) 
X R
(1) (1)   1(1) 
R =  Y =  R2  .
(1) (1)
Z R3
Similarly, let R(2) and R(3) be the eigenvectors corresponding to eigen-
values λ2 and λ3 . Then
  
(1) (2) (3) 
R1 R1 R1 ↑ ↑ ↑
S =  R2(1) R2(2) R2(3)  =  R(1) R(2) R(3)  . (15.39)
 
(1)
R3 R3 R3
(2) (3) ↓ ↓ ↓

The matrix S (called the “modal matrix”) itself describes a rotation, or

more generally, an orthogonal transformation. Using S to transform A
via a similarity transformation generates a transformed matrix A0 . This
transformed matrix is diagonal and the elements along the diagonal are
the eigenvalues. That is,
 
λ1 0 0
A0 = S −1 AS =  0 λ2 0  . (15.40)
0 0 λ3
Thus, the modal matrix S “diagonalizes” the matrix A. (This will be
proved later.)
Application: The preceding discussion has been rather abstract,
so let us consider a specific example using the matrix that represents
a rotation of 45◦ about the z-axis. This matrix is
√
   
cos 45◦ sin 45◦ 0 1 1 0
2
A =  − sin 45◦ cos 45◦ 0  = −1 1 0√  .
0 0 1 2
0 0 2/ 2
The first task is to find the eigenvalues and the eigenvectors of this
matrix. The eigenvalue equation is
AC = λC,
where C is an eigenvector that can be written in the form of a column
matrix, thus  
c1
C =  c2  .
c3
 15.4. EULER’S THEOREM 537

That is,     
1 1 0 c1 c1
 −1 1 2
0√   c2  = √ λ  c2  .
0 0 2/ 2 c3 2 c3
Carrying out the matrix multiplications we obtain the following three
equations:
√
(1 − 2λ)c1 + c2 = 0, (15.41)
√
−c1 + (1 − 2λ)c2 = 0,
(1 − λ)c3 = 0.
The third equation yields λ = 1. The first two are coupled homogeneous
algebraic equations and have a nontrivial solution if and only if the
determinant of their coefficients is zero:
 √ 
 (1 − 2λ) 1
√

  = 0.
 −1 (1 − 2λ) 
This leads to the secular equation for λ, namely
√
2λ2 − 2 2λ + 2 = 0.
Solving for λ yields
1
λ = √ (1 ± i) .
2
The three eigenvalues are
λ(1) = 1,
1
λ(2) = √ (1 + i),
2
1
λ(3) = √ (1 − i).
2
There is one eigenvector associated with each eigenvalue. Inserting
eigenvalue λ(1) = 1 into the first of equations (15.41) gives (for the
components of C(1) )
√ (1) (1)
(1 − 2)c1 + c2 = 0,
from which
(1) (1)
c2 = 0.41c1 .
where the superscripts are reminders of which eigenvalue is being used.
The second equation of (15.41) gives
(1)
√ (1)
−c1 + (1 − 2)c2 = 0,
538 15. ROTATIONAL KINEMATICS

which yields
(1) 1 (1)
c2 = c .
0.41 1
(1) (1)
The two relations between c1 and c2 are contradictory. They can
(1) (1)
only be satisfied if c1 = 0 and c2 = 0. Now insert eigenvalue λ = 1
into the third equation of (15.41) to obtain
(1)
c3 (1 − λ) = 0,
(1) (1)
which is satisfied for any value of c3 . We usually set c3 = 1 for
convenience. This completes the determination of the first eigenvector,
giving
 
0
(1) 
C = 0 .
1
Next use λ(2) = √12 (1 + i) in the three equations. The first equation
of (15.41) then reads
√
  
1
1 − 2 √ (1 + i) c1 + c2 = 0
2
from which
(2) (2)
c2 = ic1 .
Plugging λ(2) into the second equation of (15.41) gives the same rela-
tion, so the equations only give the relative values of the components.
(1)
At this point, one usually simply lets c1 = 1. Lastly, plugging λ(2) into
(2)
the third equation gives c3 = 0. So the second eigenvector is
 
1
(2) 
C = i .
0

Similarly, using λ(3) leads to the following expression for the third eigen-
vector  
1
C(3) =  −i  .
0
Once you have obtained the eigenvalues and the related eigenvec-
tors you can write down the modal matrix S by remembering the the
columns of S are the eigenvectors. That is, the columns of S are
 15.4. EULER’S THEOREM 539

C(1) , C(2) , and C(3) thus:

 
0 1 1
S =  0 i −i  .
1 0 0
Now I want to show you that the similarity transformation S −1 AS
will generate a diagonal matrix with the eigenvalues along the diagonal.
But to do so, I first need to determine S −1 , the inverse of S. You may
not recall the technique for obtaining the inverse of a matrix. One
method is to first generate a matrix whose elements are the cofactors
of the elements of S. (The cofactor of an element of a matrix is the
determinant formed from what is left after scratching out the row and
column of the element. Thus, the cofactor of S11 is 0. The cofactor
of S12 is −i (because of the sign change associated with the rows and
columns of determinants). The matrix of the cofactors of S is
 
0 −i −i
D =  0 −1 1  .
−2i 0 0
Next we take the transpose of D. Exchanging rows and columns gives
 
0 0 −2i
DT =  −i −1 0  .
−i 1 0
Finally, the inverse of S is given by
1
S −1 = DT
det S
where  
 0 1 1 
 
det S = |S| =  0 i −i  = −2i.
 1 0 0 
Consequently,  
0 0 −2i
1 
S −1 = −i −1 0  .
−2i −i 1 0
Finally, we can evaluate S −1 AS to find
 
1 0 0
−1 1
S AS = 0  √
2
(1 + i) 0 ,
1
√ (1 − i)
0 0 2

and indeed the diagonal elements are the eigenvalues of A.

540 15. ROTATIONAL KINEMATICS

Exercise 15.10. Obtain C(3) .

Exercise 15.11. Verify that S −1 S =1 where 1 is the unit matrix.
Exercise 15.12. Verify the expression obtained for S −1 AS

Proof of Euler’s Theorem

After that rather long discussion of homogeneous algebraic equa-
tions, we return to Euler’s theorem. I mentioned previously that the
theorem is proved if you can show that the real orthogonal matrix rep-
resenting the rotation of a rigid body with one point fixed always has
at least one eigenvalue equal to +1.
Although I showed you in the example above that the matrix S
diagonalizes A and that the diagonal elements are the eigenvalues, I
will now prove it in general. Recall that
AR =λR
represents the three equations (one for each eigenvalue/eigenvector
pair),
AR(1) =λ(1) R(1) , AR(2) =λ(2) R(2) , AR(3) =λ(3) R(3) . (15.42)
These equations are expressed more neatly by defining the matrix λ as
 (1) 
λ 0 0
λ =  0 λ(2) 0  . (15.43)
0 0 λ(3)
Then, since the modal matrix S is
 
↑ ↑ ↑
S =  R(1) R(2) R(3)  ,
↓ ↓ ↓
the three equations (15.42) can be expressed as the single matrix equa-
tion,
AS = Sλ. (15.44)
−1
Multiplying equation (15.44) on the left by S you obtain
S −1 AS = S −1 Sλ = λ
Therefore, the similarity transformation S −1 AS does indeed generate a
diagonal matrix whose elements are the eigenvalues. Furthermore, the
determinant of λ is +1 because A is an orthogonal matrix and under
 15.4. EULER’S THEOREM 541

a similarity transformation the determinant of a matrix is conserved.

Consequently,
|λ| = +1. (15.45)
From Equation (15.43) you can see that this means
λ(1) λ(2) λ(3) = +1. (15.46)
The proof of Euler’s theorem is almost complete, but it is convenient
at this point to consider some simple facts about the nature of the
eigenvalues of orthogonal rotation matrices. These are:
Fact 1: All the eigenvalues of an orthogonal rotation matrix have
unit magnitude. This is a consequence of the fact that a rotation
preserves the length of a vector. That is, |R0 | = |R| , or
x02 + y 02 + z 02 = x2 + y 2 + z 2 .
However, things are not quite as simple as they might appear because
the secular equation could have imaginary or complex roots. In that
case the eigenvectors are complex. The magnitude of a complex vector
is obtained from
|R|2 = R∗ · R.
Therefore, the “rotated” vector R0 has the property
R0∗ · R0 = R∗ · R.
But R0 = λR, so
R0∗ · R0 = λ∗ λR∗ · R.
Consequently, complex eigenvalues, must satisfy the condition
λ∗ λ = +1.
In other words, if λ is real, then λ = +1 but if λ is complex, then
λ∗ λ = +1.
Fact 2: A real orthogonal 3 × 3 matrix has at least one real eigen-
value. This is proved from the fact that the secular equation is cubic
so it has the form
f (λ) = λ3 + aλ2 + bλ + c = 0.
This function is illustrated in Figure 15.8.
Note that if λ = −∞, f (λ) is negative and if λ = +∞, f (λ) is
positive. Therefore, the function crosses the axis at least once and
there must be at least one real root. You have seen that a rotation
requires that |A| = +1 so the real root must be λ = +1.
Fact 3: The complex conjugate of an eigenvalue is also an eigen-
value. This is a consequence of Fact 1. Also, it is a consequence
of the fact that the elements of A are real because in that case the
542 15. ROTATIONAL KINEMATICS

Figure 15.8. A plot of the cubic secular equation.

coefficients of the secular equation are real. So if λ is a solution of

λ3 + aλ2 + bλ + c = 0, we can take the complex conjugate of the equa-
tion and see that λ∗ is also a solution. Therefore, if λ1 is complex, then
λ2 = λ∗1 .
The proof of Euler’s theorem is now in hand. You know that |λ| =
λ1 λ2 λ3 = +1. There are two possibilities, namely that all the roots are
real or that one root is real and the two complex roots are conjugates.
Consider the first possibility. All the roots are real. Recall that all
real roots must equal +1. Then A =1. This represents a null rotation.5
That leaves the second possibility. One root is real and equal to
+1. The other two roots are complex conjugates. For example,
λ(1) = eiφ λ(2) = e−iφ λ(3) = +1.
This corresponds to a rotation through φ about the z axis.
Conclusion: For any nontrivial situation, one and only one of the
eigenvalues is +1. Euler’s theorem is proved.
You can use Euler’s theorem to determine the direction of the axis
of rotation and the angle of the single rotation that takes a rigid body
from any given initial orientation to any desired final orientation. Begin
with the matrix A using, for example, the Euler angles, as in Equation
(15.32). Assume λ = +1 and solve the eigenvalue equations (Equa-
tions 15.36) for X, Y, Z. This gives you the direction of the desired axis
of rotation. The magnitude of the single rotation about this axis is
obtained by evaluating the trace of the matrix A. To appreciate this,
note that you can always use a similarity transformation to transform
A to a set of axes in which the single rotation is about the z axis. Call
5Another
possibility is λ1 λ2 λ3 = (−1)(−1)(+1) but this just represents a rota-
tion through π about one axis or an inversion of the other two axes.
 15.5. INFINITESIMAL ROTATIONS 543

the angle of rotation φ. The transformed matrix will have the form
 
cos φ sin φ 0
A0 =  − sin φ cos φ 0  .
0 0 1
The trace of A0 is
Tr A0 = 1 + 2 cos φ.
(Recall that the trace is the sum of the diagonal elements.) The trace
of an orthogonal matrix is invariant under a similarity transformation,
so the trace of the original matrix A is also 1 + 2 cos φ. Therefore, you
can determine the rotation angle by simply evaluating the trace of A.

Exercise 15.13. Show that the two eigenvalues of A obey the

relation λ∗ λ = +1.
Exercise 15.14. Assume a, b, c are real. Show that if λ is a solution
of λ3 + aλ2 + bλ + c = 0, then λ∗ is also a solution.

15.5. Infinitesimal Rotations

According to Euler’s theorem, any rotation or series of rotations
can be described as a single rotation about a suitably selected axis.
Furthermore, as you know, a rotation can be described in terms of three
independent quantities. A vector has three independent components.
Therefore, it would seem that a rotation could be described by a vector.
Indeed, we already did this in previous chapters when we described
rotations about a fixed axis in terms of the vector Ω.
There is, however, a problem in using a vector such as Ω to describe
a finite rotation about a fixed point (rather than a fixed axis). The
problem is that finite rotations about different axes do not commute.
The reason for this is that rotations are described by matrices and in
general AB = 6 BA. For example, if an airplane undergoes a 90◦ roll
followed by a 90◦ yaw, it ends up in a different orientation than if it
undergoes a 90◦ yaw followed by a 90◦ roll. If these two rotations are
represented by two vectors, say Ω1 and Ω2 , then, since vector addition
is commutative, you would expect that Ω1 + Ω2 = Ω2 + Ω1 , which
is clearly not true for the situation just described. Finite rotations
cannot, in general, be represented by vectors.
544 15. ROTATIONAL KINEMATICS

Nevertheless, infinitesimal rotations do commute. An infinitesimal

rotation only changes the components of a vector slightly so that, for
example,
x01 = x1 + 11 x1 + 12 x2 + 13 x3 .
In this equation the ij are infinitesimals so we are justified in keeping
only first order terms in the ’s. Using the summation convention, an
infinitesimal rotation is expressed
x0i = (δij + ij )xj ,
or in matrix notation
x0 = (1 + )x. (15.47)
The matrix (1 + ) is the matrix describing an infinitesimal rotation.
It is almost equal to the identity transformation.
It is now easy to prove that infinitesimal rotations commute. If
1 + 1 and 1 + 2 represent two different infinitesimal rotations, then
.
(1 + 1 )(1 + 2 ) = 1 + 1 + 2 + 1 2 = 1 + 1 + 2
Here 1 2 has been neglected because it is a second order infinitesimal.
The product of the matrices taken in reverse order obviously gives the
same result.
Other Properties of Infinitesimal Rotations
If A = 1 + , then A−1 = 1 −  because
AA−1 = (1 + ) (1 − ) = 1 +  −  = 1.
Furthermore, since the matrix representing a rotation is orthogonal,
AT = A−1 and therefore
AT = 1 + T = 1 − ,
so
T = −.
This means that  is an antisymmetric matrix. That is, the elements
on the diagonal are zero and the off-diagonal elements are such that
ij = −ji . Therefore the matrix  has the form
 
0 dΩ3 −dΩ2
 =  −dΩ3 0 dΩ1  . (15.48)
dΩ2 −dΩ1 0
Going back to Equation (15.47), note that the differential change
in the vector x due to the rotation is
dx = x0 − x = x.
 15.6. SUMMARY 545

Therefore, using (15.48) the components of dx are

dx1 = x2 dΩ3 − x3 dΩ2 ,
dx2 = x3 dΩ1 − x1 dΩ3 ,
dx3 = x1 dΩ2 − x2 dΩ1 .
These equations are the familiar expressions for the components of the
cross product of the vectors r and dΩ where
 
dΩ1
dΩ =  dΩ2  .
dΩ3
That is,
dx = x × dΩ.
The vector dΩ is a vector lying along the axis of rotation. It’s length
is equal to the angle through which the rotation takes place.

Exercise 15.15. Draw a picture of an airplane. Draw the same

airplane after it has undergone a pitch of 90◦ followed by a yaw of
90◦ . Now draw the same airplane after it has undergone a yaw of 90◦
followed by a pitch of 90◦ .

15.6. Summary
The orientation of a rigid body is described in terms of the ori-
entation of a set of axes (x0 , y 0 , z 0 ) fixed in the body relative to a set
of inertial axes (x, y, z). Mathematically, the orientation is represented
by the set of nine direction cosines aij giving the angles between the
primed and the unprimed axes. Since three parameters are sufficient
to describe the orientation of a rigid body, the direction cosines are
not all independent; indeed they are related by the six orthogonality
conditions,
aij aik = δjk .
The rotation (or transformation) matrix A is a 3 × 3 matrix with ele-
ments aij . (Equation 15.8). The position vector in the rotated frame is
x0 and is related to the position vector in the inertial frame by equation
(15.9).
x0 = Ax.
546 15. ROTATIONAL KINEMATICS

The transformation matrix A can be interpreted in an active sense

as a rotation of the vector x, or in the passive sense as an opposite
rotation of the coordinate system. The transformation matrix is or-
thogonal, that is,
AT = A−1 .
If A represents a rotation, its determinant is +1. If B represents a
rotation of a coordinate system, then the elements of A will change.
The change in A under such a rotation is given by the similarity trans-
formation,
A0 = BAB −1 .
A body can be transformed from any initial orientation to any final
orientation by applying three successive rotations through the Euler
angles as follows: (1) about z through φ, (2) about the new x-axis
through θ, and (3) about the new z-axis through ψ. These rotations
are represented by the 3 × 3 matrices D, C, B (given in Equations 15.27,
15.29, and 15.30). The total transformation matrix A = BCD is given
in Equation (15.32).
Euler’s Theorem states that the transformation from any given ini-
tial orientation to any given final orientation can be achieved by a single
rotation about an appropriately chosen axis. Points on the axis will
not be affected by the rotation, so if R lies along the axis
AR = R,
or, in terms of an eigenvalue equation
AR = λR.
To prove Euler’s theorem requires showing that one eigenvalue of A is
+1. This involves defining the eigenvectors R(1) , R(2) , and R(3) asso-
ciated with the three eigenvalues λ1 , λ2 , and λ3 . The modal matrix S
whose columns are the eigenvectors, diagonalizes the transformation
matrix A by a similarity transformation (Equation 15.40):
 
λ1 0 0
A0 = S −1 AS =  0 λ2 0  = λ.
0 0 λ3
This implies that |λ| = λ1 λ2 λ3 = +1. By showing that two of the
eigenvalues are complex (and complex conjugates) it follows that one
eigenvalue is +1, thus proving Euler’s theorem.
Rotations about a fixed point are represented by matrices. They
cannot, in general, be represented by vectors because vectors commute
 15.7. PROBLEMS 547

whereas finite rotations do not commute. However, infinitesimal ro-

tations do commute and these can be represented by an infinitesimal
vector dΩ.

15.7. Problems
Problem 15.1. Show that the length of a vector is unchanged by
an orthogonal transformation; that is, prove that
x0i x0i = xi xi .
Problem 15.2. Show that the determinant of the Euler transfor-
mation matrix, Equation (15.32) is +1.
Problem 15.3. Prove that the determinant and the trace of a
matrix are unchanged by a similarity transformation.
Problem 15.4. A “natural” coordinate system for the Earth has
the origin at the center of the planet, the z-axis passing through the
North Pole, and the x-axis in the equatorial plane and going through
the Greenwich meridian. (Where is the y-axis?) Suppose that you
wanted to change to a system with the z-axis going through London
(0◦ E, 51◦ 310 N ), and the x-axis also passing through the Greenwich
meridian. Determine the Euler angles to generate this transformation.
Problem 15.5. People in Fresno felt neglected and asked us to
determine an axis of rotation such that a single rotation would put
Fresno on the top of the world, that is, at the North Pole. Determine
the direction of the axis of rotation as well as the angle required to put
Fresno where it wants to be. At present, Fresno is at 119◦ 470 W and
36◦ 440 N.
Problem 15.6. The Earth is magically rotated through the fol-
lowing Euler Angles: φ = 30◦ , θ = 45◦ , and ψ = 30◦ . Determine
the latitude and longitude of the new directions of the x, y, z axes.
At present, the z-axis points through the North Pole, the x-axis goes
through (0◦ E, 0◦ N ) and the y-axis goes through (90◦ E, 0◦ N )
Problem 15.7. Determine the eigenvalues and eigenvectors for the
matrix  
1 3 0
 3 −2 −1  .
0 −1 1
Problem 15.8. Prove that the product of two orthogonal matrices
is an orthogonal matrix.
548 15. ROTATIONAL KINEMATICS

Problem 15.9. Assume the following matrices act on  a two dimen-

  −1 0 0
1 0
sional vector R. Describe the result. (a) , (b) 0 −1 0  .
0 −1
0 0 1
Problem 15.10. A vector r is reflected in the xy-plane. (a) Write
a vector equation for r0 , the vector after reflection. (b) Generalize to a
reflection in a plane perpendicular to the unit vector n̂. (c) Write the
transformation as a matrix equation. (d) Show that the tranformation
matrix is an improper orthogonal matrix. (Answers: (a) r0 = r −
2(r · k̂)k̂. (d) |A| = −1.)
 
1 4
Problem 15.11. Consider the matrix T = . Obtain its
1 1
eigenvalues and eigenvectors. Obtain the modal matrix S. Show that
S −1 T S diagonalizes T and that the diagonal elements are the eigen-
values.
Problem 15.12. Determine the eigenvectors for the matrix
 
cos 60◦ sin 60◦ 0
 − sin 60◦ cos 60◦ 0  .
0 0 1
Problem 15.13. Assume A is an orthogonal matrix. Let each
column of A represent a vector. Show that these vectors are orthogonal.
Problem 15.14. Consider the angular velocity vector ω. Its com-
ponents along the x, y and z-axes are ωx , ωy and ωz . Show that these
components can be expressed in terms of the Euler angles as follows:
ωx = θ̇ cos φ + ψ̇ sin θ sin φ,
ωy = θ̇ sin φ − ψ̇ sin θ cos φ,
ωz = ψ̇ cos θ + φ̇.

COMPUTATIONAL PROJECTS
Computational Project 15.1. Develop a program to determine
the eigenvalues of a 2 × 2 matrix.
Computational Project 15.2. Write a computer program to
find the inverse of a 2 × 2 matrix.
Chapter 16
Rotational Dynamics

In the previous chapter we described the kinematics of a rigid body.

In this chapter we study the dynamics of a rigid body that is rotating
in an arbitrary manner about a fixed point. To make the analysis easier
I will introduce tensors and dyads, physical quantities that may be new
to you. (You will probably be using tensors for the rest of your life, so
this is a good chance to learn them!) Once the necessary mathematical
tools have been developed they are used to obtain expressions for the
energy and angular momentum of rotating bodies. Next we develop
the equations of motion for rotating bodies; these are called Euler’s
equations. Finally, these concepts are used to study two problems,
a freely rotating body and a rotating body acted upon by a torque.
As examples we consider the Earth and a spinning top. The Earth is
approximately a torque-free spinning body and a top1 (or gyroscope)
is an example of a spinning body subjected to an external torque.
It is convenient to split the dynamical problem into two parts, one
corresponding to the translation of the body and the other correspond-
ing to the rotation of the body about some point. I will not prove
it, but it turns out that one can always do this.2 Therefore, in the
remainder of this chapter I will assume that the body is rotating about
a fixed point. As you saw in the previous chapter, the generalization
1Rotational motion is a particularly rich field of study. Klein and Sommerfeld
wrote a four volume study on the motion of a spinning top. F. Klein and A.
Sommerfeld, Uber die Theorie des Kreisels B. G. Teubner, Leipzig, 1897-1910.
vols 1-4. As you might expect, the book is wide ranging, discussing rotating bodies
in general with applications to geophysics, astronomy, and technology.
2Chasles’ Theorem states that the most general displacement of a rigid body
is a translation plus a rotation. This is actually a corollary of Euler’s Theorem of
the previous chapter.
549
550 16. ROTATIONAL DYNAMICS

from rotation about a fixed axis, to rotation about a fixed point leads
to a much more complicated formulation of the problem. (That is why
we will need to use tensors.) The study of rotational dynamics is of
particular value to physics students because many of the mathematical
techniques will carry over into the analysis of small oscillations about
stable equilibrium points and into quantum theory.

16.1. Angular Momentum

In our previous consideration of the angular momentum of rotating
bodies, in Chapter 7, we assumed not only that the rotation was about
a fixed axis, but furthermore that the body was symmetrical about
the axis of rotation. We are now going to generalize and obtain an
expression for the angular momentum of a body of arbitrary shape
rotating about a fixed point.
The angular moment li of a particle of mass mi and linear momen-
tum pi is
li = ri × pi = mi (ri × vi ) ,
where ri is the position of particle i with respect to the origin. (The
origin is usually the center of mass, or a point in the body that is fixed
in space.) The angular momentum of a rigid body with respect to the
origin will be X X
L= li = mi (ri × vi ).
i i
The summation is over all the particles in the body and vi is the velocity
of particle i due to the rotation of the body. The velocity of particle i
in a body rotating with angular velocity ω is
vi = ω × ri ,
so X
L= mi [ri × (ω × ri )] .
i
Then, using the BAC-CAB rule,
X
L= mi [ω(ri · ri ) − ri (ri · ω)] . (16.1)
i
By expanding this expression, it is easy to show that the x-component
of the angular momentum of the body is
X 
mi ri2 ωx − xi (xi ωx + yi ωy + zi ωz ) ,

Lx =
X 
mi ωx ri2 − x2i − ωy xi yi + ωz xi zi ,


=
X   X   X 
= mi (ri2 − x2i ) ωx + − mi xi yi ωy + − mi xi zi ωz ,
 16.1. ANGULAR MOMENTUM 551

and similarly for Ly and Lz . Therefore, the components of L have the

form
Lx = Ixx ωx + Ixy ωy + Ixz ωz , (16.2)
Ly = Iyx ωx + Iyy ωy + Iyz ωz ,
Lz = Izx ωx + Izy ωy + Izz ωz ,
where the moments of inertia are defined by expressions of the form
X Z
2 2
Ixx = mi (ri − xi ) or Ixx = ρ(r)(r2 − x2 )dτ, (16.3)
V

and the products of inertia are defined by expressions of the form

X Z
Ixy = − mi xi yi or Ixy = − ρ(r)xydτ. (16.4)
V

These expression can be written very economically as follows:

Z
Ijk = ρ(r)(r2 δjk − rj rk )dτ,
V

where the subscripts j, k take on the values 1, 2, 3, and δjk is the Kro-
necker delta. Note that Ijk = Ikj so there are only six independent
elements in our expressions for the moments and products of inertia.3
The nine quantities Ijk defined above can be represented as a 3 × 3
array. This array is called the inertia tensor. It is denoted I.
Finally, using matrix multiplication, the angular momentum can be
expressed as
L = Iω. (16.5)

Exercise 16.1. Expand Equation (16.1) to obtain an expression

for Ly .
Exercise 16.2. Consider a ring of mass M and radius R. Determine
the moments and products of inertia. Assume the z-axis goes through
the center of the ring and the x- and y-axes are in the plane of the ring.
Answer: Ixx = 12 M R2 .

3In this context it is somewhat more natural to write r1 , r2 , r3 rather than

x1 , x2 , x3 as we did in the previous chapter. The expression ri2 − x2i is often written
in the equivalent form yi2 + zi2 , and similarly for the other two moments of inertia.
552 16. ROTATIONAL DYNAMICS

Tensors
A scalar (mass, charge, etc.) is a physical quantity having one
component. If you rotate the coordinate system, the value of a scalar
does not change. A scalar is a tensor of rank zero.
A vector (position, velocity, etc.) is a physical quantity having three
components. If you rotate the coordinate system, the components of
a vector change according to the rule Vi0 = aij Vj where the a’s obey
condition aij aik = δjk . (See Equation 15.6.) A vector is a tensor of rank
one.
A tensor of rank two is a physical quantity having nine components4
that transform according to the rule,
Tij0 = aik ajl Tkl .
Note that the coefficients aik can be considered elements of a matrix A
that transforms T into T 0 by the similarity transformation
T 0 = AT AT .
A rank two tensor can always be represented as a square matrix. Note,
however, that a tensor and a matrix are not the same thing. A matrix
is simply an array of numbers. There are no a priori conditions on the
transformation of a matrix. A matrix usually has no physical signifi-
cance, whereas a tensor is a physical quantity that can be represented
as an orthogonal matrix. A second rank tensor is often represented by
a 3 × 3 matrix. For computational purposes there is no difference be-
tween a tensor and a matrix; if you represent a tensor by a matrix then
the tensor obeys all the rules of matrix algebra. (There are, however,
ways to represent tensors that do not involve matrices.)
A tensor of rank N is a quantity having 3N components which
transform in a particular way under orthogonal transformations. The
general transformation law for a tensor undergoing an orthogonal trans-
formation is:
0
Tijk··· (r0 ) = ail ajm akn · · · Tlmn··· (r),
where the summation convention is implied. Here T 0 (r0 ) is the ten-
sor represented in terms of the primed coordinate system and T (r) is
the tensor represented in terms of the unprimed coordinate system.
In mathematical terms, a tensor is often defined as a quantity that
transforms according to this relation.
You may be thinking that this is very complicated, but you will see
that in practice the only tensors we use are those of rank zero, rank
4I
am assuming three-dimensional space. In a four-dimensional space a rank
two tensor has sixteen elements.
 16.1. ANGULAR MOMENTUM 553

one, and rank two (and you are very familiar with tensors of rank zero
and one).
In summary:
A tensor of rank zero (N = 0) has one component and it is invariant
under a coordinate transformation. We call such an object a scalar.
A tensor of rank one (N = 1) has three components and transforms
according to Ti0 = ail Tl . Such an object is called a vector.
A tensor of rank two (N = 2) has nine components and transforms
according to Tij0 = ail ajm Tlm .
We shall not be concerned with tensors of rank higher than two.
Dyads and Dyadics
Another useful quantity is a dyad, defined in terms of two vectors.
It is the object you would obtain if you multiplied two vectors together
in the most naive way possible. For example, the dyad formed from
the vectors A and B and denoted AB is defined as5
  
AB = Ax ı̂ + Ay ̂ + Az k̂ Bx ı̂ + By ̂ + Bz k̂
= Ax Bx ı̂ı̂ + Ax By ı̂̂ + Ax Bz ı̂k̂
+Ay Bx ̂ı̂ + Ay By ̂̂ + Ay Bz ̂k̂
+Az Bx k̂ı̂ + Az By k̂̂ + Az Bz k̂k̂.
A dyadic is a linear polynomial of dyads, as for example
AB + CD + · · · .
The unit dyadic is
1 =ı̂ı̂+̂̂+k̂k̂.
A dyad operates on a vector by the dot product, thus
AB · C = A(B · C),
and
C · AB = (C · A)B.
In dyadic notation the inertia tensor can be written as
I = mi (ri2 1 − ri ri ),
where the summation is implied, or as
Z
I = ρ(r) r2 1 − rr dτ.
 

5This kind of product of two vectors is also called the “outer product”, the
“direct product” and the “tensor product.”
554 16. ROTATIONAL DYNAMICS

Referring to Equation (16.2), it is clear that in tensor notation the

angular momentum is given by
L =I·ω. (16.6)
This differs from Equation (16.5) only in the dot between I and ω.
indicating that we are considering I as a tensor rather than as a matrix.
(This makes no practical difference.)
Keep in mind that if I were a scalar, then the angular momentum L
would have the same direction as the angular velocity ω, but since I is
a tensor, the angular momentum and angular velocity will, in general,
not have the same orientation.
Worked Example 16.1. : Obtain an expression for L from
L =I·ω assuming I is (a) a matrix and (b) a dyadic.
Solution: As a matrix
    
Lx Ixx Ixy Ixz ωx
L =  Ly  =  Iyx Iyy Iyz   ωy 
Lz Izx Izy Izz ωz
 
Ixx ωx + Ixy ωy + Ixz ωz
=  Iyx ωx + Iyy ωy + Iyz ωz  .
Izx ωx + Izy ωy + Izz ωz
As a dyadic
L = I·ω = (Ixx ı̂ı̂ + Ixy ı̂̂ + Ixz ı̂k̂ + Iyx ̂ı̂ + Iyy ̂̂ + Iyz ̂k̂
+Izx k̂ı̂ + Izy k̂̂ + Izz k̂k̂) · (ωx ı̂ + ωy ̂ + ωz k̂)
= Ixx ωx ı̂ + Ixy ωy ı̂ + Ixz ωz ı̂ + Iyx ωx ̂ + Iyy ωy ̂ + Iyz ωz ̂
+Izx ωx k̂ + Izy ωy k̂ + Izz ωz k̂
= (Ixx ωx + Ixy ωy + Ixz ωz )ı̂ + (Iyx ωx + Iyy ωy + Iyz ωz )̂
+ (Izx ωx + Izy ωy + Izz ωz ) k̂.

Exercise 16.3. (a) Evaluate 1·ı̂. (b) Evaluate 1 · (5ı̂+6̂+7k̂). An-

swer: (a) ı̂.
Exercise 16.4. Evaluate AB · C if A = 3ı̂, B = 2ı̂+5̂, and
C =ı̂+2̂+3k̂. Answer: 36ı̂.

16.2. Kinetic Energy

Having determined the angular momentum of a rotating body, let
us now obtain an expression for the kinetic energy of a rigid body
 16.2. KINETIC ENERGY 555

rotating about a fixed point. The kinetic energy of a particle is

1
Ti = mi vi2 .
2
Therefore, the kinetic energy of a rigid body is
X1
T = mi vi2 .
i
2

But vi2 = vi · vi and vi = ω × ri so,

X1
T = mi (ω × ri ) · vi
i
2
X1
= mi ω · (ri × vi )
2
i
1 X
= ω· ri × mi vi
2 i
1
= ω · L.
2
Using Equation (16.6) for the angular momentum, we obtain
1
T = ω·I·ω. (16.7)
2
The work-energy theorem for rotating bodies is
dT
= ω · N. (16.8)
dt

Worked Example 16.2. : Prove that the work-energy theo-

rem for a rotating body is given by Equation (16.8).
Solution: The work-energy theorem states that the net work
done on a body is equal to the body’s increase in kinetic energy.
For linear motion,
dT = dW = F · dx
so
dT dx
= F =F·v
dt dt
= F · (ω × r) = (ω × r) · F = ω · (r × F)
= ω·N (QED)
556 16. ROTATIONAL DYNAMICS

16.3. Properties of the Inertia Tensor

16.3.1. Eigenvalues and Principal Axes. The inertia tensor
can be represented by a symmetric matrix, i.e., Ixy = Iyx . The com-
ponents of the inertial tensor depend on the orientation of the body
relative to the coordinate axes, but since the tensor is symmetric, only
six of these components are independent.6 Furthermore, the inertia
tensor is real because all the elements are real. From these two prop-
erties it follows that
∗
I † = I T = I,
which show that the inertia tensor is hermitian.
The definition of the moments and products of inertia (Equations
16.3 and 16.4) indicate that the components of I depend on the choice
of the origin and on the orientation of the axes. As I will prove in a
bit, by making an appropriate choice of the axes you can obtain an
expression for I in which all of the off-diagonal terms are zero. I will
denote this diagonalized inertia tensor by I 0 . In matrix form it looks
like this:  
I1 0 0
I 0 =  0 I2 0  . (16.9)
0 0 I3
Note that I1 , I2 , and I3 are the elements of the diagonalized tensor.
Generally speaking they will not be equal to Ixx , Iyy , and Izz .
The diagonalized inertia tensor can also be written in dyadic form,
thus:
I 0 = I1 ı̂ı̂+I2 ̂̂+I3 k̂k̂. (16.10)
It is always helpful to express the inertia tensor in diagonalized form. It
is a clean and simple way to express the tensor and it leads to very nice
expressions for the angular momentum and the kinetic energy, namely:
L1 = I1 ω1 L2 = I2 ω2 L3 = I3 ω3 ,
1 1 1
T = I1 ω12 + I2 ω22 + I3 ω32 .
2 2 2

Exercise 16.5. Show that Equations (16.9) and (16.10) are equiv-
alent.

6As
we shall see, there is a particular set of axes in which the inertia tensor has
only three non-zero components.
 16.3. PROPERTIES OF THE INERTIA TENSOR 557

We now consider how to diagonalize the inertia tensor. Recall that

when a matrix is diagonalized, the diagonal elements are its eigenvalues.
Thus, the quantities I1 , I2 , and I3 are the eigenvalues of the inertia
tensor. Furthermore, once you have found the eigenvalues, you can
find the eigenvectors. The directions of the eigenvectors give the three
axes of the coordinate system in which I is diagonal. (These are called
the principal axes.) To diagonalize the inertia tensor you will first
determine the eigenvalues and eigenvectors. (You learned how to do
that in the previous chapter.)
Except for the trivial case, the orthogonal matrices considered in the
previous chapter had only one real eigenvalue. For the inertia tensor, all
three eigenvalues are real and the three corresponding eigenvectors are
mutually orthogonal. While proving that this is true we will formulate
the procedure for finding the principal axes.
Let the three eigenvectors of I be denoted Rj (j = 1, 2, 3) and let
Ij be the corresponding eigenvalue. The eigenvalue equations are:
I · Rj = Ij Rj j = 1, 2, 3. (16.11)
Multiply from the left by R∗l . (In matrix notation R∗l will have to be
a row vector; otherwise you could not carry out the matrix multiplica-
tion.)
R∗l · I · Rj = Ij R∗l · Rj . (16.12)
The complex conjugate of Equation (16.11) is
I ∗ · R∗l = Il∗ R∗l . (16.13)
(Note the change in the dummy index.) By the rules of matrix multi-
plication, I · R = R·I T , so Equation (16.13) can be written as
R∗l · I † = Il∗ R∗l .
Multiplying from the right with Rj gives
R∗l · I † · Rj = Il∗ R∗l · Rj . (16.14)
But I = I † because I is hermitian. Therefore, the left-hand sides of
Equations (16.12) and (16.14) are identical and
(Ij − Il∗ )R∗l · Rj = 0. (16.15)
Either l = j or l 6= j. If l = j,
(Ij − Ij∗ )R∗j · Rj = 0.
But R∗j · Rj = |Rj |2 > 0, and hence, Ij = Ij∗ , thus proving the eigenval-
ues are real. This result followed from the hermitian nature of I, so it
implies that in general the eigenvalues of a hermitian matrix are real.
(This fact is very important in quantum mechanics where operators are
558 16. ROTATIONAL DYNAMICS

represented by matrices and observables are represented by the eigen-

values of these operators. Since observables must be real quantities,
the operators in quantum mechanics must be represented by hermitian
matrices.)
Now consider the possibility l 6= j. In that case, if the eigenvalues
are not equal, R∗l · Rj = 0 proving that the eigenvectors corresponding
to different eigenvalues are orthogonal.
However, if Il = Ij this proof does not hold. In that case, the
eigenvalues are identical and are said to be degenerate. Assume, for
example, that eigenvalues 2 and 3 are degenerate, i.e., I2 = I3 , but
that I1 is distinct. Then there is an eigenvector R1 corresponding to
I1 and at least one eigenvector (call it R2 ) corresponding to the double
eigenvalue. This eigenvector is orthogonal to R1 by Equation (16.15).
If the coordinate axes are selected with the x-axis along R1 and the
y-axis along R2 , then
I ·ı̂ = I1 ı̂,
and
I · ̂ = I2 ̂.
Writing these equations out in full immediately demonstrates that
I12 = I13 = I23 = 0. But since I is a symmetrical tensor it follows
that I31 and I32 are also zero. Therefore k̂ is also an eigenvector with
eigenvalue I2 .
The problem of degenerate eigenvalues can also be addressed by
noting that the equality of two eigenvalues implies a symmetry of the
rigid body about the axis with the distinct eigenvalue (say I1 ). A plane
that is perpendicular to R1 must contain the other two eigenvectors.
Since Equations (16.11) are linear, any linear combination of eigenvec-
tors in that plane is also a solution. Therefore, any two orthogonal
directions in the plane perpendicular to R1 can serve as the other two
principal axes.7
A third approach to the problem of degenerate eigenvalues is famil-
iar to you if have already studied quantum mechanics. There you used
the Gram-Schmidt procedure to generate a set of orthogonal eigenvec-
tors from a set of nonorthogonal eigenvectors.8
If all three of the eigenvalues of the inertia tensor are equal, then
all directions in space are eigenvectors and any set of three orthogonal

7See,
for example, Florian Scheck, Mechanics: From Newton’s Laws to Deter-
ministic Chaos, Springer Verlag, Berlin, 1990, pp 172-173.
8See, for example, David J. Griffiths, Introduction to Quantum Mechanics,
Prentice Hall, Englewood Cliffs, N.J.,1995, pp 79 - 80.
 16.3. PROPERTIES OF THE INERTIA TENSOR 559

axes can be considered principal axes. In that case the inertia tensor
will be diagonal to begin with.
To determine the eigenvalues (or “principal moments of inertia”)
recall that the eigenvalue equation (16.11) can be written as
(I − I1) · R = 0.
A nontrivial solution can be found if and only if the determinant of the
coefficients vanishes, that is, iff
 
 I11 − I I12 I13

 
 I21
 I22 − I I23  = 0.

 I31 I32 I33 − I 
This “secular equation” is a cubic in I and yields the three roots which
are the principle moments of inertia. Each one of these roots can then
be substituted into Equation (16.11) to obtain the directions of the
principal axes.

Exercise 16.6. Show by writing the equation in matrix form that

if I·ı̂ = I1 ı̂ then I21 and I31 are zero.

16.3.2. Evaluating the Inertia Tensor. To show you how to

evaluate the inertia tensor and how to determine the eigenvalues and
principal axes I will go through a few examples in detail. However,
to learn this material you will have to solve a number of problems on
your own. Each problem has its own pitfalls and complications. I have
selected some rather simple systems because I am trying to show you
the technique. You will get a chance to consider more complicated
systems when you solve the problems.
Evaluating the inertia tensor basically involves nothing more than
doing some (often unpleasant) integrations to determine the moments
and products of inertia. There are, however, some rather nice labor
saving devices that you should use whenever possible. The four labor
saving devices you may find most useful are:
Labor Saving Device 1 (LSD 1): The Parallel Axis Theorem. If
you know the inertia tensor about the center of mass, you can determine
the inertia tensor with respect to parallel axes through a different origin
O by using the relation
IO = ICM + M R2 1 − RR


(16.16)
560 16. ROTATIONAL DYNAMICS

where R is the coordinate of the center of mass relative to O.

Labor Saving Device 2 (LSD 2): Rotating the Tensor. If it is
easy to evaluate the tensor about one set of axes (say xyz) then you can
obtain the expression for the inertia tensor with respect to another set
of axes (x0 y 0 z 0 ) having the same origin by a similarity transformation,
thus:

I 0 = AIAT . (16.17)

The elements of A are the direction cosines (aij = ê0i · êj ).

Labor Saving Device 3 (LSD 3): If the body is composed of
several parts, the inertia tensor for the whole body is the sum of the
tensors for the various parts (as long as these are calculated for the
same origin and the same set of axes).
Labor Saving Device 4 (LSD 4): Any plane of symmetry of a
body is perpendicular to a principal axis.
As a simple example of evaluating the inertia tensor, let us calculate
the inertia tensor for a cube of side a and mass M with respect to
coordinates parallel to the sides of the cube and with origin at one
corner, as illustrated in Figure 16.1.

a
y

a
x

Figure 16.1. A cube of mass M and side of length a.

To determine the elements of the inertia tensor I we need to eval-

uate the terms in the following formula
   2 
Ixx Ixy Ixz Z Z Z y + z 2 −xy −xz
I = Iyx Iyy Iyz  = ρ  −yx x2 + z 2 −yz  dτ.
2 2
Izx Izy Izz Vol
−zx −zy x +y
(16.18)
 16.3. PROPERTIES OF THE INERTIA TENSOR 561

For a cube, these integrals are quite simple. I will evaluate Ixx just to
illustrate the process. Note that the density is ρ = M/a3 .
Z Z Z
Ixx = ρdτ (y 2 + z 2 )
Vol
Z a Z a Z a Z a Z a Z a 
M 2 2
= y dxdydz + z dxdydz
a3 x=0 y=0 z=0 x=0 y=0 z=0
2
= M a2 .
3
It takes no imagination to appreciate that Iyy and Izz are also equal to
(2/3)M a2 . It is left as an exercise to show that the products of inertia
are equal to −(1/4)M a2 .
Therefore,  2 
3
− 41 − 14
I =  − 41 3
2
− 14  M a2
1 1 2
−4 −4 3
where it is understood that M a2 multiplies every term in the matrix.

Exercise 16.7. Determine the products of inertia for the cube of

Figure 16.1.

You can appreciate that the evaluation of this particular inertia

tensor was quite simple. If you were asked to determine the inertia
tensor of the cube about a point at the center of the cube you could
use the parallel axis theorem (LSD 1), writing it as
ICM = IO − M (R2 1 − RR).
The vector R from O to CM is
a a a
R = ı̂+ ̂+ k̂,
2 2 2
so √
3a
|R| =
2
and consequently,
3a2
 
4
0 0
3a2
R2 1 =  0 4
0 .
3a2
0 0 4
562 16. ROTATIONAL DYNAMICS

The last term is

a a a  a a a 
RR = ı̂+ ̂+ k̂ ı̂+ ̂+ k̂
2 2 2 2 2 2
a2  
= ı̂ı̂+ı̂̂+ı̂k̂+̂ı̂+̂̂+̂k̂ + k̂ı̂+k̂̂+k̂k̂
4  
2 1 1 1
a 
= 1 1 1 .
4 1 1 1

Therefore
 1

2
− 41 − 14
R2 1 − RR =a2  − 14 2
1
− 14  .
− 14 −4 1 1
2

Finally, the inertia tensor about parallel axes through the center of
mass is
 2   1 
3
− 14 − 14 2
− 14 − 14
ICM = M a2  − 14 2
3
− 14  − M a2  − 14 1
2
− 14 
1 1 2 1 1 1
−4 −4 3
−4 −4 2
 1   
0 0 2 1 0 0
6 Ma 
= M a2  0 16 0  = 0 1 0 .
0 0 1 6 0 0 1
6

You are probably not a bit surprised that the matrix is diagonal. From
the simplicity of the figure (and by LSD 4) you realize that axes parallel
to the cube sides and with origin at the CM will be the principal axes
for the inertia tensor of a cube. Furthermore, you note that the cube
has triply degenerate eigenvalues of M a2 /6.
The principal axes for the cube were found almost accidentally, but
it is not always that simple. To illustrate, let us determine the principal
axes for the object formed when you slice the cube along a diagonal.
As shown in Figure 16.2 the solid is a tetrahedron with three faces that
are right triangles and one face that is an equilateral triangle. We will
first determine the moment of inertia with respect to the (xyz) axes
indicated in the figure, and then determine the principal axes.
The inertia tensor is given by Equation (16.18). I will evaluate a
moment of inertia so you can see how it is done. If the mass of the
tetrahedron is M and its volume is (1/6)a3 , the density is ρ = 6M/a3
and hence,
 16.3. PROPERTIES OF THE INERTIA TENSOR 563

z z

a z+x+y=a a

a a
y y
x'

a a
x
x

Figure 16.2. A tetrahedron formed by slicing a cube

diagonally. In the sketch on the right a plane of sym-
metry is indicated. The x0 -axis is perpendicular to the
plane of symmetry.

6M a
Z Z Z Z Z a−x Z a−x−y
2 2
(y 2 + z 2 )dz


Ixx = ρ y + z dτ = 3 dx dy
a x=0 y=0 z=0
Vol
6M a
Z Z a−x  
2 1 3
= dx dy y (a − x − y) + (a − x − y)
a3 x=0 y=0 3
Z a 5
6M 1 4 Ma
= (a − x) dx = .
a3 x=0 6 a3 5
(The last two integrals are rather messy and it is easiest to evaluate
them with Maple or some other computer algebra system.)
By symmetry, Iyy = Izz = Ixx = 15 M a2 . It is left as a problem to
show that the products of inertia are all equal to −M a2 /20. Therefore,
the inertia tensor for the tetrahedron about the xyz axes is
 1 1 1
  
− − 2 4 −1 −1
5 20 20 Ma 
I = M a2  − 201 1
5
1 
− 20 = −1 4 −1  .
− 1
− 1 1 20 −1 −1 4
20 20 5

Exercise 16.8. Show that the volume of the tetrahedron of Figure

16.2 is a3 /6.

Having obtained the inertia tensor about the xyz axes of Figure
16.2 let us now determine the principal axes. Using LSD 4, note that
the plane of symmetry indicated in the right-hand panel of 16.2 will
564 16. ROTATIONAL DYNAMICS

be perpendicular to a principal axis. Rotating the coordinate system

by 45◦ in a clockwise direction around the z-axis generates a new set
of axes, x0 y 0 z in which the x0 -axis is a principal axis. To carry out the
rotation use LSD 2, that is,
I 0 = AIAT .
The components of the rotation matrix A are given by the dot products
of unit vectors along the old and the new coordinates, and it is easy to
show that  √ √ 
1/√2 −1/√2 0
A =  1/ 2 1/ 2 0  ,
0 0 1
and therefore,
 √ √   √ √ 
1/√2 −1/√2 0 4γ −1γ −1γ 1/√2 1/√2 0
I 0 =  1/ 2 1/ 2 0   −1γ 4γ −1γ   −1/ 2 1/ 2 0  ,
0 0 1 −1γ −1γ 4γ 0 0 1
where γ = M a2 /20. Carrying out the matrix multiplications generates
 
2 5 0 √ 0
Ma 
I0 = 0 √3 − 2 .

20
0 − 2 4
This shows that x0 is indeed a principal axis. (Actually, the reason I
carried out the rotation was not to show that x0 was a principal axis,
but just to show you a practical example of rotating the inertia tensor.)
Although I obtained one principle axis by a simple rotation, it is
not obvious where the other two principal axes are, so let’s go back
to basics and determine the eigenvalues and eigenvectors for I 0 . The
eigenvalue equation is
I 0 R =λR,
or
(I 0 − λ1)R =0,
which, written in matrix form, is
  
5γ − λ 0 0 √ R1
 0 3γ −
√ λ −γ 2
  R2  = 0. (16.19)
0 −γ 2 4γ − λ R3
(There will be one such equation for each eigenvalue.)
Now Equation (16.19) is a system of three coupled homogeneous
algebraic equations and they have a nontrivial solution if and only if
 16.3. PROPERTIES OF THE INERTIA TENSOR 565

the determinant of the coefficients is zero. That is,

 
 5γ − λ 0 0 √ 

 0
 3γ −
√λ −γ 2  = 0,
 0 −γ 2 4γ − λ 
or
√ √
(5γ − λ)(3γ − λ)(4γ − λ) − (5γ − λ)(−γ 2)(−γ 2) = 0,
(5γ − λ) (3γ − λ)(4γ − λ) − 4γ 2 /2 = 0.
 

So,
5γ − λ = 0,
and
λ2 − 7γλ + 10γ 2 = 0.
Therefore, one eigenvalue is
M a2
 
(1)
λ = 5γ = 5 ,
20
and the other two are
M a2
λ(2) = 5γ = 5 ,
20
M a2
λ(3) = 2γ = 2 .
20
Note that the eigenvalue 5γ appears twice, indicating a degeneracy.
The first eigenvector is obtained by inserting λ(1) into Equations (16.19),
yielding
(1) (1) (1)
0R1 + 0R2 + 0R3 = 0,
(1) (1)
√ (1)
0R1 − 2γR2 − 2γR3 = 0,
(1)
√ (1) (1)
0R1 − 2γR2 + γR3 = 0.
(1) (1)
The first of these equations is satisfied by any values of R1 , R2 , and
(1)
R3 , the components of the eigenvector corresponding to λ(1) . How-
ever the second and third of these equations cannot be simultaneously
(1) (1)
satisfied by any nonzero values of R2 and R3 . Since it is convenient
for the eigenvectors to have unit magnitude we select the components
of R(1) to be (1, 0, 0). Note that these are the components of the first
eigenvector in the rotated coordinates, x0 , y 0 , z. Once again we see that
x0 is a principal axis.
566 16. ROTATIONAL DYNAMICS

Let us now consider eigenvalue λ(3) = 2γ. Inserting this value into
Equations (16.19) gives
(2)
3γR1 = 0,
(2)
√ (2)
γR2 − 2γR3 = 0,
√ (2) (2)
2γR2 + 2γR3 = 0.
(2)
The first equation requires R1 = 0, so the x0 component of this eigen-
(2) (2) √
vector is zero. The second equation states that R3 = R2 / 2 and
the third equation yields the same relation.√Thus, the second eigenvec-
tor could be written as R(2) = (0, −1, −1/ 2), but this does not have
magnitude unity, so we set it to R(2) = (0, −0.8165, −0.5774). √ [Note
that this gives a vector in the same direction as (0, −1, −1/ 2) but it
has magnitude unity.]
Finally, recall that for a degenerate eigenvalue you can find one
eigenvector by the usual technique, and the other eigenvector will be
orthogonal to the first one. We found the eigenvector R(1) to be (1, 0, 0)
and we know that there is another eigenvector perpendicular to this
one. It will therefore lie in the shaded plane of Figure 16.2. It will
also be perpendicular to R(2) . It is left as an exercise to verify that the
third eigenvector can be written as R(3) = (0, −0.5774, 0.8165).
And so, finally, the principal axes of the body sketched in Figure
16.2 have been determined. It required a significant amount of labor
but hopefully you appreciate the procedure.

(1) (1)
Exercise 16.9. Show that R2 and R3 must be equal to zero.
Exercise 16.10. Show that R(3) = (0, −0.5774, 0.8165) is a vector
of unit magnitude and is perpendicular to both R(1) and R(2) .

As a final example, let us determine the inertia tensor for an ideal-

ized gyroscope as shown in Figure 16.3. Assume the axis is massless, so
the system is just a disk of radius a and mass M. To make the problem
simpler, assume
RR the disk has negligible thickness and its mass is given
by M = σdA where σ is the mass per unit area. The problem
is to express the inertia tensor relative to the inertial axes x, y, z. As
the gyroscope spins, point P can move in any way consistent with the
bottom end of the gyroscope remaining fixed at O. However, for the
moment, assume point P is fixed and that the disk is not spinning.
 16.3. PROPERTIES OF THE INERTIA TENSOR 567

z P z' z'

x'
d y'
y y' x'
O
x

Figure 16.3. An idealized gyroscope made of a mass-

less axis and a thin disk. The bottom end of the gyro-
scope is fixed at O.

To determine the inertia tensor it is obviously easiest to use body

axes fixed in the disk, such as x0 y 0 z 0 shown on the right in Figure 16.3.
It is natural to select z 0 along the axis of the gyroscope. For convenience
let x0 be parallel to the inertial axis x.
The moments and products of inertia can be evaluated using equa-
tion (16.18). For example, the value of Ix0 0 x0 is
Z Z
0
Ix0 x0 = σ(y 02 + z 02 )dA.

Here z 0 is the distance in the z 0 direction to the mass element, but since
we are assuming a zero thickness disk, it is reasonable to set it equal
to zero. Then
Z Z
0 M
Ix0 x0 = 2 y 02 dA.
πa
Changing to polar coordinates and noting that y 0 = r sin φ
Z a Z 2π
0 M M a2
Ix0 x0 = 2 (r sin φ)2 rdrdφ = .
πa r=0 0 4
2 2
Similarly, Iy0 0 y0 = Ix0 0 x0 = M4a , while Iz0 0 z0 = M2a . (You could have ob-
tained these results by elementary means using the perpendicular axis
theorem.) All of the products of inertia are zero. (You will show this
in an exercise.) The inertia tensor with respect to the body axes with
origin at the center of mass of the disk is
 1 
4
0 0
I 0 = M a2  0 14 0  .
0 0 12
Now translate the origin of the body axes along the z 0 -axis to the
origin of the space axes (x, y, z). The inertia tensor with respect to
568 16. ROTATIONAL DYNAMICS

these parallel translated axes can be determined using LSD 1: IO0 =

ICM + M (R2 1 − RR) where R is the vector from the origin to the
center of mass. The translated axes are still the body coordinates
x0 , y 0 , z 0 , so
R = (d) ẑ0 ,
where d is the distance from O to the center of the disk along the axis
of rotation, and ẑ0 is a unit vector in the z 0 direction, that is, pointing
along the axis of the gyroscope. So R2 = d2 and RR =d2 k̂k̂. Therefore,
 M a2     
4
0 0 M d2 0 0 0 0 0
M a2
IO0 =  0 4
0 + 0 M d2 0 − 0 0 0 
2 2 2
0 0 Ma 0 0 Md 0 0 Md
2
2
M ( a4 − d2 ) 0
 
0
2
=  0 M ( a4 − d2 ) 0 .
M a2
0 0 2
Finally, we have to rotate the inertia tensor to obtain an expression in
terms of the space axes xyz. The body axes can be rotated about the
x-axis (or x0 , as they are the same) and this will line up z 0 and z, and
also y 0 and y. Thus, the transformation to the space axes is simply a
rotation about x0 through θ, where θ is the angle between the z- and
z 0 -axes. So,  
1 0 0
A =  0 cos θ − sin θ  .
0 sin θ cos θ
Consequently, the inertia tensor in terms of the space axes is
I = AI 0 AT
2
M ( a4 +d2 ) 0
   
1 0 0 0 1 0 0
a 2 2
=  0 cos θ − sin θ   0 M ( 4 +d ) 0   0 cos θ sin θ 
0 sin θ cos θ M a2 0 − sin θ cos θ
0 0 2
2
M ( a4 +d2 ) 0
 
0
a2 2 M a2 2
=  0 cos θM ( 4 +d )+ sin θ 2 sin θ cos θM (d2 − a4 )
2 2 
2 2 2
0 sin θ cos θM (d2 − a4 ) sin2 θM ( a4 +d2 )+ cos2 θ M2a

Exercise 16.11. Show that in the body axes, the idealized gyro-
scope of our example has all the products of inertia equal to zero.
 16.4. THE EULER EQUATIONS OF MOTION 569

16.4. The Euler Equations of Motion

We now consider the problem of a rotating body that is subjected
to an external torque. Recall that we have mentioned that the motion
of a rotating rigid body can be resolved into the translational motion
of the center of mass and the rotational motion about the center of
mass. The translational motion of the center of mass can be treated
using F = ma. In this section we analyze the rotational motion about
the center of mass. The results are applicable to the general motion of
a body with one point fixed, in which case the rotation is taken about
the fixed point.
The most straightforward approach to the problem is to start with
Newton’s second law as applied to a rotating body, i.e.,
 
dL
= N.
dt inertial
The time derivative referred to an inertial coordinate system can be
replaced with one referred to a set of axes fixed in the rotating body
by applying the usual prescription,
   
dL dL
= + ω × L.
dt inertial dt body

Replacing dL


dt inertial
by N and dropping the subscripts, this equation
becomes  
dL
+ ω × L = N.
dt
Keep in mind that this equation is valid in the body (rotating) reference
frame. Now L =I · ω and in the rotating frame the inertia tensor I is
constant, so
dω
I· + ω× (I · ω) = N. (16.20)
dt
Equation (16.20) is the equation of motion for a rotating body. An
interesting consequence of this equation is that if a body is to ro-
tate freely at constant angular velocity with no torques applied, then
ω× (I · ω) =0, which means that I · ω is parallel to ω. Therefore, ω
must lie along a principal axis. This may seem a bit far-fetched to you,
but just think about the last time you had your car wheels spin bal-
anced. The purpose of the dynamical balancing is to make sure that a
principal axis of the inertia tensor is aligned with the angular velocity
vector. This ensures that the wheel spins without any torques being
exerted on the axis. (Explain that to your mechanic!)
570 16. ROTATIONAL DYNAMICS

Assuming the body axes are principal axes, Equation (16.20) can
be expressed in component form as
I1 ω̇1 − ω2 ω3 (I2 − I3 ) = N1 ,
I2 ω̇2 − ω3 ω1 (I3 − I1 ) = N2 , (16.21)
I3 ω̇3 − ω1 ω2 (I1 − I2 ) = N3 .
These are called Euler’s equations of motion. They are rotational ana-
logues of Newton’s second law of motion. In the next two sections we
describe applications of Euler’s equations.

Exercise 16.12. Show that for a symmetric body, Euler’s equa-

tions (16.21) reduce to the elementary relation N = Iα (Equation
7.12).

16.5. Torque-Free Motion

The Moon and the Sun exert small but non-zero torques on Earth’s
equatorial bulge, resulting in a very slow precession of the Earth’s ro-
tation vector. (We shall consider a similar mechanism in the next sec-
tion.) But even if we discount this effect and assume that no external
torque acts on the Earth, we find that according to Euler’s equations,
there will be a precession of Earth’s rotation vector. So, for the mo-
ment, let us assume the Earth is a freely rotating oblate spheroid.
If there are no torques acting on a rotating body, Euler’s equations
of motion (Equations 16.21) reduce to
I1 ω̇1 = ω2 ω3 (I2 − I3 ),
I2 ω̇2 = ω3 ω1 (I3 − I1 ),
I3 ω̇3 = ω1 ω2 (I1 − I2 ).
If the body is a spheroid, two of the moments of inertia will be equal.
Let I1 = I2 . Then Euler’s equations further reduce to
I1 ω̇1 = ω2 ω3 (I1 − I3 ),
I1 ω̇2 = ω3 ω1 (I3 − I1 ),
I3 ω̇3 = 0.
 16.5. TORQUE-FREE MOTION 571

The third equation establishes that ω3 is constant. The first two equa-
tions can be written as
I1 − I3
ω̇1 = ω3 ω2 ,
I1
I1 − I3
ω̇2 = −ω3 ω1 .
I1
Denoting the constant ω3 (I1 − I3 ) /I1 by Ω, these become
ω̇1 = −Ωω2 , (16.22)
ω̇2 = +Ωω1 . (16.23)
Taking the derivative of the first of these equations and inserting it into
the second equation yields

ω̈1 = −Ω2 ω.
But this is the equation for simple harmonic motion! A solution is
ω1 = A cos Ωt.
Similarly
ω2 = A sin Ωt.
In conclusion, ω3 is constant but ω1 and ω2 vary sinusoidally out of
phase with each other. This means that the total angular velocity
vector, ω = ω1 ı̂+ω2 ̂+ω3 k̂, precesses about ω3 k̂ with a precession rate
Ω. (See Figure 16.4.)

ω3k
ω

Figure 16.4. Illustrating the torque free motion of a

spherical body. The angular velocity vector precesses at
a rate Ω.

This “free precession” for the Earth should have a period of Ω =

ω3 (I1 − I3 ) /I1 = 306 days or approximately ten months. It may or
may not have been detected. There appears to be a very small devia-
tion of the axis of rotation of the Earth having this period, causing a
572 16. ROTATIONAL DYNAMICS

wandering of the poles with an amplitude of about ten meters. How-

ever, there is a much greater (but still small) precession with a period
of 420 days called the Chandler wobble which might be the free pre-
cession. The reason for the disagreement in period is believed by some
geophysicists to be due to the fact that the Earth is not really a rigid
body.
The free precession should not be confused with the precession
of the equinoxes due to the torques on the Earth’s equatorial bulge.
This is responsible for the gradual precession of the first point in Aries
through the zodiac with a period of 26,000 years.

Exercise 16.13. Show that ω2 = A sin Ωt.

16.6. The Spinning Top

You may recall the elementary analysis of the gyroscope that was
presented in Section 7.6. We now consider the symmetrical9 spinning
top (or gyroscope) using the techniques developed in this chapter.
z
z' θ

y'
x'
y
ϕ
x
ξ

Figure 16.5. A spinning top. The orientation of the

top relative to the inertial axes x, y, z is described by the
Euler angles, θ, φ, ψ.

Figure 16.5 shows the orientation of the top in terms of the Euler
angles, θ, φ, ψ. The spinning motion of the top is given by the rotation
9Theasymmetrical top is a more complicated problem and is usually reserved
for graduate courses in mechanics. If you would like to look at the solution see L.D.
Landau and E. M. Lifshitz, Mechanics, 3rd ed, Pergammon Press, Oxford, 1976,
Section 37.
 16.6. THE SPINNING TOP 573

about the z 0 -axis, and the spin angular velocity is ψ̇ =ψ̇ẑ0 If the top is
precessing, the rotation vector is tracing out a cone around the vertical
or z-axis. As the rotation axis traces out this trajectory, the angle φ is
changing, so the angular velocity associated with precession is φ̇ =φ̇ẑ.
Finally, the top may be nutating, that is, the axis may be nodding up
and down as it precesses. From the figure, it is clear that this type
of motion represents a change in the angle θ. The angular velocity
associated with nutation would be a vector along the line of nodes
(ξ) and can be expressed as θ̇ξ. ˆ Recall that the Euler angles form a
set of three independent coordinates so they are an appropriate set of
generalized coordinates for describing the motion of the system.
Using the Euler angles of Figure 16.5, the angular velocity ω is
written in terms of its components thus

ω = θ̇ξ̂ + φ̇ẑ + ψ̇ẑ0 ,

where the unit vectors are directed along the lines ξ, z, z 0 . Let ê1 , ê2 , and
ê3 be unit vectors directed along the body axes, x0 , y 0 , z 0 . (By symmetry,
these are the principal axes.) In terms of ê1 , ê2 , and ê3 the unit vectors
along ξ, z, z 0 are given by

ξ̂ = ê1 cos ψ − ê2 sin ψ,

ẑ = ê1 sin θ sin ψ + ê2 sin θ cos ψ + ê3 cos θ,
ẑ0 = ê3 .

Consequently, the components of the velocity along the body axes are

ω1 = ω · ê1 = θ̇ cos ψ + φ̇ sin θ sin ψ, (16.24)

ω2 = ω · ê2 = −θ̇ sin ψ + φ̇ sin θ cos ψ,
ω3 = ω · ê3 = φ̇ cos θ + ψ̇.

Note that ω3 is not simply ψ̇; the precession of the axis of rotation also
has a component in the ê3 direction.
Let the top have mass m and assume that the center of mass lies a
distance d along the axis from the point of contact on the floor. Since
the top is symmetrical, the center of mass lies on the z 0 -axis. The
torque acting on the top is N = dẑ0 × mg, a vector pointing along the
line of nodes (ξ) and having magnitude mgd sin θ. This torque is the
source of the precession.
The top is spinning about its axis, but it is also precessing and
nutating. To develop a mathematical description of precession and
nutation, let us begin by writing the Lagrangian for the system.
574 16. ROTATIONAL DYNAMICS

The kinetic energy is given by T = 12 ω · I · ω, so

1
I1 ω12 + I2 ω22 + I3 ω32 .


T = (16.25)
2
This leads to a fairly long and complicated expression. But for a sym-
metrical body with I2 = I1 it reduces to
  2 
1 2 2 2
 
T = I1 θ̇ + φ̇ sin θ + I3 ψ̇ + φ̇ cos θ . (16.26)
2
The potential energy for the spinning top is simply the gravitational
potential energy, V = mgd cos θ. Consequently,
  2 
1 2 2 2
 
L=T −V = I1 θ̇ + φ̇ sin θ + I3 ψ̇ + φ̇ cos θ − mgd cos θ.
2
(16.27)
There is a great deal of information in this Lagrangian, as well as
in what is not in the Lagrangian. In particular, you will note that
the coordinates φ and ψ are not in the Lagrangian at all. They are
ignorable. Consequently the momenta conjugate to these variables are
constant. To be specific, since φ is ignorable,
∂L 2
 
pφ = = I1 φ̇ sin θ + I3 cos θ ψ̇ + φ̇ cos θ = constant. (16.28)
∂ φ̇
Similarly, since ψ is ignorable,
∂L
pψ = = I3 (ψ̇ + φ̇ cos θ) = constant. (16.29)
∂ ψ̇
Furthermore, since time does not appear explicitly in the Lagrangian
or in the transformation equations, the total energy is constant, that
is,
  2 
1 2 2 2
 
E= I1 θ̇ + φ̇ sin θ + I3 ψ̇ + φ̇ cos θ +mgd cos θ = constant.
2
(16.30)
In terms of the constant quantities pφ and pψ , the energy is
1 (pφ − pψ cos θ)2 p2ψ
E = I1 θ̇2 + + + mgd cos θ. (16.31)
2 2I1 sin2 θ 2I3
This expression gives the energy in terms of a single coordinate, θ.
The form of the total energy (Equation 16.31) suggests defining an
effective potential involving all the terms with a θ. For convenience the
constant term p2ψ /2I3 is also included in the effective potential. Then,
1
E = I1 θ̇2 + V 0 (θ),
2
 16.6. THE SPINNING TOP 575

with
0 (pφ − pψ cos θ)2 p2ψ
V (θ) = + mgd cos θ + . (16.32)
2I1 sin2 θ 2I3
Solving for θ̇ yields
r
dθ 2
θ̇ = = (E − V 0 ). (16.33)
dt I1
Given values for the constants E, pφ and pψ this equation can (in
principle) be integrated to obtain θ = θ(t). Inserting θ(t) in Equations
(16.28) and (16.29) will then yield expressions for φ = φ(t) and ψ =
ψ(t). The problem is thus solved. (It may not be easy to do this, but
conceptually it presents no difficulty.)
However, even without actually solving Equation (16.33) we have
obtained a great deal of information about the behavior of the system.
For example, note that the rotation of the top around its axis of rotation
is ω3 . The third of Equations (16.24) and the expression for pψ given
in Equation (16.29) lead to
pψ = I3 ω3 = constant.
Thus the rotational velocity of the top about its axis of symmetry is a
constant. (This assumes, of course, that the top does not slow down
due to friction at the point of contact with the ground or with the air.)
Figure 16.6 is a plot of the effective potential V 0 (θ) as a function
of θ. Note that the effective potential goes to infinity at 0 and π, as is
clear from Equation (16.32). The minimum in V 0 (θ) can be found by
setting the derivative of V 0 with respect to θ equal to zero. Denoting
the angle at which V 0 is minimized by θ0 , this yields

(pφ − pψ cos θ0 )pψ (pφ − pψ cos θ0 )2 cos θ0

0 = −mgd sin θ0 + − . (16.34)
I1 sin θ0 I1 sin3 θ0
After some fairly complicated algebra we can write this in the interest-
ing form
mgdI1 sin4 θ0 − (pφ − pψ cos θ0 )(pψ − pφ cos θ0 ) = 0. (16.35)
Solving gives the value of θ0 for the particular values of the constant
parameters.
If we define β = pφ − pψ cos θ0 we can write Equation 16.34 in the
form
pψ sin2 θ0 mgdI1 sin4 θ0
β2 − β+ = 0, (16.36)
cos θ0 cos θ0
576 16. ROTATIONAL DYNAMICS

that is, a quadratic in β whose solution is

" s #
pψ sin2 θ0 4mgdI1 cos θ0
β= 1± 1− . (16.37)
2 cos θ0 p2ψ
We will return to this equation shortly.
100

90
Effe ctive P ote ntia l (J )

80

70

60

50

40

30

20

10

0
0 0.5 1 1.5 2 2.5
The ta (ra d)

Figure 16.6. The effective potential as a function of θ

for a spinning top.

Exercise 16.14. Obtain Equations (16.24).

Exercise 16.15. Obtain the general expression for the kinetic en-
ergy of a rotating body by evaluating Equation (16.25). Show that for
a symmetrical body this reduces to (16.26).
Exercise 16.16. Obtain the expressions for the generalized mo-
menta pφ and pψ (Equations 16.28 and 16.29).
Exercise 16.17. Carry out the required substitutions to convert
Equation (16.30) into (16.31).

16.6.1. Precession without Nutation. If the total energy E is

equal to V 0 (θ0 ) the top has a constant value of θ equal to θ0 . This is,
of course, the minimum value of E and the top is at the minimum of
the effective potential curve. If, however, the top has somewhat more
 16.6. THE SPINNING TOP 577

energy, the value of θ can vary between two limits (call them θ1 and
θ2 ) and the top nutates.
We shall first consider the case of the minimum energy top with
E = V 0 (θ0 ) and no nutation taking place. The precession rate of the
top is φ̇(θ0 ). To determine the value of this precession we insert θ0 into
Equations (16.28) and (16.29). Thus, from (16.29),
pψ
ψ̇ = − φ̇ cos θ0 ,
I3
so (16.28) reads
 
2 pψ
pφ = I1 φ̇ sin θ0 + I3 cos θ0 − φ̇ cos θ0 + φ̇ cos θ0 .
I3
Consequently,
pφ − pψ cos θ0 β
φ̇ = 2 = . (16.38)
I1 sin θ0 I1 sin2 θ0
Inserting the value of β from Equation 16.37 we obtain
" s #
pψ 4mgdI1 cos θ0
φ̇ = 1± 1− .
2I1 cos θ0 p2ψ
But pψ = I3 ω3 so
" s #
I3 ω3 4mgdI1 cos θ0
φ̇ = 1± 1− (16.39)
2I1 cos θ0 I32 ω32
Since φ̇ must be real, the quantity under the root must be positive,
that is
4mgdI1 cos θ0
1− ≥0
I32 ω32
from which
I1
ω32 ≥ 2 (4mgd cos θ0 ) .
I3
For precession to take place, ω3 must be at least as great as this mini-
mum value. If the minimum is exceeded, there are two possible values
for φ̇, as described in the following worked example.

Worked Example 16.3. :Show that the fast and slow preces-
sion rates are φ̇ ' (I3 ω3 /I1 cos θ0 ) and φ̇ ' mgd/I3 ω3 . (The slow
precession rate is the one most commonly observed.)
578 16. ROTATIONAL DYNAMICS

Solution: From Equation 16.39 we have

 1/2
I3 ω3 I3 ω3 4mgdI1 cos θ0
± 1− .
2I1 cos θ0 2I1 cos θ0 [I3 ω3 ]2
Using the binomial expansion
  
I3 ω3 1 4mgdI1 cos θ0
φ̇ ' 1± 1− .
2I1 cos θ0 2 [I3 ω3 ]2
The positive sign gives
I3 ω3 I3 ω3
φ̇ ' (2) = ,
2I1 cos θ0 I1 cos θ0
and the negative sign gives
I3 ω3 1 4mgdI1 cos θ0 mgd
φ̇ ' 2 = .
2I1 cos θ0 2 [I3 ω3 ] I3 ω3

Exercise 16.18. Show that the slow precession rate agrees with
the result obtained in Exercise 7.15

16.6.2. Nutation. We have assumed that the energy of the top

is E = V 0 (θ0 ), that is, the system is at the minimum of the effective
potential well. Then θ remains constant and equal to θ0 . However, as
mentioned above, if the energy of the top is greater than this minimum
value, θ will oscillate between values θ1 and θ2 . See the sketch in Figure
16.7.
As θ varies between θ1 and θ2 , the top nutates; that is, the axis of
rotation of the top “nods” up and down. (A similar phenomenon was
observed for the spherical pendulum. See Section 12.4.2.) In this case,
the value of the energy at the “turning points” θ1 and θ2 is equal to
the effective potential at those points and θ1 and θ2 are solutions of the
equation
(pφ − pψ cos θ)2 p2ψ
E= + + mgd cos θ. (16.40)
2I1 sin2 θ 2I3
This is the same as Equation (16.31) except that the term involving θ̇
is left out because θ̇ is zero at the turning points. But Equation (16.40)
can be written as
p2ψ
2I1 (E − ) sin2 θ = p2φ − 2pφ pψ cos θ + pψ cos2 θ + 2I1 mgd cos θ sin2 θ.
2I3
 16.6. THE SPINNING TOP 579

Effective Potential V( )
E

0 Angle π
1 2

Figure 16.7. Qualitative sketch of the effective poten-

tial for a symmetrical top. If the energy E is greater
than the minimum value of the effective potential V 0 ,
the value of θ will vary between θ1 and θ2 . (The top nu-
tates.)

Replacing sin2 θ by 1 − cos2 θ,

p2ψ
2I1 (E− )(1−cos2 θ) = p2φ −2pφ pψ cos θ+pψ cos2 θ+2I1 mgd cos θ(1−cos2 θ),
2I3
which is a cubic in cos θ. There are three roots to this equation. Two
of them are θ1 and θ2 . The third root is greater than +1 and does not
represent a physically possible situation.
During nutation, the direction of the axis of rotation (θ) will vary
between θ1 and θ2 . But φ̇ depends on θ so the precession rate varies as
well. We found (see Equation 16.38) that
pφ − pψ cos θ
φ̇ = .
I1 sin2 θ
This indicates that the sign of φ̇ depends on the relative values of pφ
and pψ . Specifically, if cos θ < pφ /pψ , φ̇ is positive, meaning that the
velocity of precession is in the same sense as ω3 , the spin velocity of the
top. Otherwise, φ̇ is negative. To express this in a different way, let
us define an angle Θ = cos−1 pφ /pψ ; then φ̇ is positive for θ > Θ and
negative for θ < Θ. In Problem 16.7 you will show that Θ is smaller
than θ0 , so Θ is smaller than θ2 . The range of values for Θ is 0 < Θ < θ0 .
If Θ < θ1 then φ̇ is always positive and the axis of the nutating top
traces out the curve indicated in Figure 16.8a. On the other hand, if
Θ > θ1 , then as θ varies from θ1 to θ2 , the precession rate φ̇ changes
sign. The path traced out by the axis then has the form shown in
Figure 16.8b. Finally, if a top that is spinning at ω3 is held with its
580 16. ROTATIONAL DYNAMICS

axis initially at some angle θ1 and released, its axis is observed to trace
out the path shown in 16.8c. The details are left to a problem.
θ1

a b c
Figure 16.8. The path traced out by the rotation axis
during nutation on the surface of a sphere centered on
the point of contact of the spinning top with the floor.

An interesting problem that was not solved until fairly recently

involves the motion of a spinning disk. If, for example, you hold a coin
upright and set it spinning with a flick of your finger, you find that
it spins upright for a while, then it leans over, and rotates faster and
faster, then suddenly falls. The classical analysis (as we have done)
would have the coin spinning forever, but actually air resistance rather
quickly stops the motion.10

Exercise 16.19. Show explicitly that if θ < Θ, then φ̇ is negative.

That is, φ̇ has the opposite sign from ω3 .
Exercise 16.20. Show that if Θ < θ1 , then φ̇ is positive.

16.7. Summary
The angular momentum of a body of arbitrary shape that is rotating
about a fixed point is
L =I·ω,
10Aspinning coin will slow down and lean over, so that it will be spinning about
a point on its circumference. The axis of rotation does not pass through the coin.
The point of contact between the coin and the surface is not fixed (it moves around
the circumference) but at any instant it can be considered as the point about which
the coin is rotating. A highly polished disk on a smoothly machined flat plat is sold
as a toy called “Euler’s Disk.” A very interesting analysis of the motion is given in
the paper by H. K. Moffatt, “Euler’s Disk and its Finite-time Singularity” Nature,
404, 833, 2000.
 16.7. SUMMARY 581

where I is the inertia tensor. The elements of the inertia tensor are
Z
Ijk = ρ(r)(r2 δjk − rj rk )dτ,

so
Z
I= ρ(r)(r2 1 − rr)dτ.

The kinetic energy of the rotating body is

1
T = ω · I·ω.
2
The inertia tensor can be represented by a real, hermitian matrix.
This formulation is most useful when the matrix is expressed in diag-
onalized form. Then the axes of the coordinate system lie along the
principal axes of the body.
It is often necessary to determine the inertia tensor in a rotated co-
ordinate system. This is accomplished by the similarity transformation

I 0 = AIAT ,

where A is the appropriate rotation matrix.

To diagonalize I we determine the eigenvalues λ(i) and the eigen-
vectors R(i) . (To assist in this task, four labor saving devices were
described.) The eigenvalues and eigenvectors of the inertia tensor are
obtained from the eigenvalue equation:

(I − λ1)R =0.

A nontrivial solution requires that the determinant of the coefficients

is zero, that is,
|I − λ1| = 0.
This relation yields three equations that can be solved for the three
eigenvalues, λ(i) , i = 1, 2, 3. Plugging the eigenvalues into the eigenvalue
equation leads to expressions for the components of the eigenvectors,
(i)
Rj , i, j = 1, 2, 3 where i specifies the eigenvector and j specifies the
component. (Note that we actually obtain the ratio of components,
(1) (1) (2) (2) (1)
such as R2 /R3 and R2 /R3 . The third component (such as R1 ) is
chosen such that the eigenvector has unit magnitude.)
The Euler equation of motion for a rotating body is
dω
I· + ω × (I · ω) = N,
dt
582 16. ROTATIONAL DYNAMICS

where I is the inertia tensor in the body coordinate system. If the

torque is zero (a freely rotating body) the Euler equation yields
I1 ω̇1 = ω2 ω3 (I2 − I3 ),
I2 ω̇2 = ω3 ω1 (I3 − I1 ),
I3 ω̇3 = ω1 ω2 (I1 − I2 ).
If the torque is not zero, the analysis is much more complicated.
We illustrated the process with the example of a spinning symmetrical
top. Using the Euler angles as coordinates, the Lagrangian is
  2 
1 2 2 2
 
L= I1 θ̇ + φ̇ sin θ + I3 ψ̇ + φ̇ cos θ − mgd cos θ.
2
The two generalized momenta associated with φ and ψ are constants,
that is,
∂L  
pφ = = I1 φ̇ sin2 θ + I3 cos θ ψ̇ + φ̇ cos θ = constant,
∂ φ̇
∂L
pψ = = I3 (ψ̇ + φ̇ cos θ) = constant.
∂ ψ̇
The total energy can then be written
1 2 (pφ − pψ cos θ)2 p2ψ
E = I1 θ̇ + + + mgd cos θ,
2 2I1 sin2 θ 2I3
or
1
E = I1 θ̇2 + V 0 (θ),
2
where the effective potential V 0 (θ) is
(pφ − pψ cos θ)2 p2ψ
V 0 (θ) = + mgd cos θ + .
2I1 sin2 θ 2I3
The precession of the top consists of a variation in φ at a rate
pφ − pψ cos θ0
φ̇ = .
I1 sin2 θ0
Precession can only occur if the top is spinning at a rate
4mgdI1
ψ̇ = ω3 ≥ cos2 θ0 sin2 θ0 .
I32
Nutation is a variation in θ, the angle between the axis of the top and
the vertical. It can vary from θ1 to θ2 , two angles whose values depend
 16.8. PROBLEMS 583

on the total energy. They can be evaluated as two of the roots of the
cubic equation
p2ψ
2I1 (E− )(1−cos2 θ) = p2φ −2pφ pψ cos θ+pψ cos2 θ+2I1 mgd cos θ(1−cos2 θ).
2I3
16.8. Problems
Problem 16.1. Prove that for a symmetrical body, the products
of inertia are zero.
Problem 16.2. For the tetrahedron of Figure 16.2 show that the
products of inertia are −M a2 /20.
Problem 16.3. A system is made up of three particles of mass M
located at (2a, 0, 0), (0, 2a, 0) and (0, 0, a). Find the principal moments
of inertia about the origin. Determine a set of principal axes.
Problem 16.4. Determine the inertia tensor for the ammonia mol-
ecule about its center of mass. Determine the principal axes. The am-
monia molecule (NH3 ) is made up of one nitrogen atom (MW 14) and
three hydrogen atoms (MW1) arranged in a pyramidical shape with the
nitrogen at the apex and the three hydrogens forming an equilateral
triangular base. The distance between the hydrogen atoms and the
nitrogen atom is 1.03 Å and the angles between the bonds are 36.4◦
and 107.2◦ .
Problem 16.5. Let Rn be an antisymmetric matrix whose ele-
ments are given by Rij = ijk xk where the xk are the coordinates of the
kth mass point of a body. Show that the matrix of the inertia tensor
of the body can be written as
I = −mn (Rn )2 .
By definition the Levi-Civita density tensor ijk is zero is any two of
the indices ijk are equal, +1 if ijk = 1, 2, 3 or any even permutation
of 1, 2, 3 and −1 for odd permutations of 1, 2, 3.
Problem 16.6. Prove Labor Saving Device 1, Equation (16.16).
Problem 16.7. Show that Θ = cos−1 (pφ /pψ ) is smaller than θ0 .
Problem 16.8. A uniform right circular cone of height h and base
R has a mass M. It is set on its side and it rolls without slipping in such
a way that the tip of the cone remains fixed. Note that instantaneously
the axis of rotation is along the line of contact between the cone and the
horizontal surface. The angle between this line and a fixed line on the
horizontal plane is θ so the angular velocity of the center of mass of the
584 16. ROTATIONAL DYNAMICS

cone is θ̇. (a) Obtain an expression for the kinetic energy. (b) Obtain
an expression for the angular momentum. (Hint: First determine the
inertia tensor relative to a set of principal axes.)
Problem 16.9. A gyroscope is both rotating about its axis and
precessing about the z-axis and has an angular velocity ω = αẑ + βẑ0 .
Determine the angular momentum about the body axes and the kinetic
energy. Assume I1 = I2 and I3 and θ are known.
Problem 16.10. A gyroscope is hanging from a fixed point (so
θ0 > π/2). Show that there is one positive and one negative value for
φ̇(θ0 ).
Problem 16.11. A spinning top is held with a fixed polar angle
θ = θ1 . Therefore, initially, θ̇ = 0, φ̇ = 0, and ψ̇ = ω3 . It is then
released and allowed to precess and nutate. (a) Write expressions for
pφ , pψ , V 0 , and E. (b) The turning points for the nutation are θ1 and
θ2 , (θ1 ≤ θ ≤ θ2 ). Obtain an expression for cos θ2 . (c) Assume the top
is spinning rapidly so that the quantity α = (2I1 mgd)/(I32 ω32 ) << 1.
Show that for this case, cos θ2 ' cos θ1 − α sin2 θ1 .
Problem 16.12. The spin axis of a rapidly spinning gyroscope
lies in the horizontal plane. That is, θ = π/2. It is precessing about
the vertical axis. (a) Show that elementary considerations lead to a
precession rate of φ̇ = mgd/I3 ω3 . (b) Show that a more sophisticated
analysis based on Equation
D E 16.33 yields the same expression for the
average precession rate φ̇ .

Problem 16.13. A symmetrical top is approximated by a cone of

mass 200 g, height 4 cm, and base diameter 3 cm. It spins on its point
at 400 rad/s. Determine the (slow) precession rate.
Problem 16.14. The disk illustrated in Figure 15.1 is spun up to
an angular velocity ω while the axis on which it is mounted is con-
strained to remain fixed in direction. Suddenly the axis is released and
system (disk plus axis) is free to wobble. (a) Determine the half angle
of the cone in space described by the axis. You may ignore the mass of
the axis. Assume the angle between the axis and a line perpendicular
to the plane of the disk is α. (b) Determine the (precession) period of
the wobble.
Problem 16.15. A set of axes (x, y, z), fixed in a body, are prin-
ciple axes. A line OQ passes through the origin (O) and has direction
cosines α, β, γ relative to the axes x, y, z. Show that the moment of
 16.8. PROBLEMS 585

inertia of the body about OQ is given by

IOQ = α2 Ix + β 2 Iy + γ 2 Iz .
(Hint: It may be helpful to note that the angle between two vectors
with direcion cosines α1 , β1 , γ1 and α2 , β2 , γ2 is given by
cos θ = α1 α2 + β1 β2 + γ1 γ2 .)
Problem 16.16. A set of axes (x, y, z) are fixed in a body. Relative
to these body axes, the moments of inertia are Ix , Iy , Iz and the prod-
ucts of inertia are Pxy , Pxz , Pyz . (a) Show that the moment of inertia
about a line OQ passing through the origin is
I = α2 Ix + β 2 Iy + γ 2 Iz − 2αβPxy − 2βγPyz − 2αγPxz
where α, β, γ are the direction cosines of OQ. (b) Draw a vector R
in the direction of OQ and let R · R be inversely proportional to the
moment of inertia about OQ, that is,
1
R2 = .
I
Show that the tip of R generates the surface of an ellipsoid. (This is
called “Poinsot’s ellipsoid of inertia.”)
Problem 16.17. A top rotates about a vertical axis. Show that the
motion is stable if ω32 > 4mgdI1 /I32 . (Hint: ∂ 2 E/∂θ2 > 0 for stability.)

COMPUTATIONAL PROJECTS
Computational Project 16.1. Write a program to solve for θ0
using Equation (16.35). You may assume m = 1, d = 0.1, pφ = 3, pψ =
2, I1 = 3, and I2 = 6.
Chapter 17
Waves

A wave is an oscillation in a medium, such as a water wave in the

ocean or a wave propagating down a stretched slinky or a vibrating
guitar string. In general, the medium itself does not translate. For
example, a wave in the ocean can be thought of as water molecules
moving vertically up and down while the wave itself moves horizon-
tally.1 A strange and interesting thing about waves is that although
there is no transport of mass, the wave does transport energy and mo-
mentum.
A transverse wave, such as a wave in a string, is one in which the
material particles move perpendicular to the direction of the waveform.
On the other hand, a sound wave or the wave set up in a metal rod when
you hit one end with a hammer, are examples of longitudinal waves.
In these waves the material particles move parallel to the direction of
the waveform.
This introduction to wave motion is limited to an analysis of waves
in strings.2 However, the wave equation and the wave properties de-
rived here are valid for all types of waves. Furthermore, the mathemat-
ical technique of separation of variables used to solve the wave equation
is a very common way to solve partial differential equations. You will
find the material of this chapter to be particularly useful when you
study electromagnetic waves.

1This is not quite true. A molecule of water also oscillates horizontally and
traces out an elliptical path - but you get the general idea.
2An ideal string is perfectly flexible and linearly elastic. This means the tension
is everywhere the same and always directed tangentially. Linear elasticity means
the tension depends linearly on the amount the string is stretched (Hooke’s law).
587
588 17. WAVES

17.1. A Wave in a Stretched String

Consider a long string. If you shake one end of the string up and
down, you can set up a traveling wave. A single shake will generate
a pulse that moves along the string; continuous shaking will produce
a wave train which is often sinusoidal in form. If you have a shorter
string, such as one of the strings in your guitar, and you pluck the
string, you can set up a standing wave. Regardless of how it is set up,
a wave is characterized by a number of physical parameters. The main
ones are:
Amplitude (A): The maximum displacement from equilibrium.
Wavelength (λ): The distance from peak to peak or between
any two corresponding points on the wave.
Period (τ ): The time it takes for a point on the wave to go
through one complete oscillation.
Frequency (f ): The number of oscillations per unit time. Note
that f = 1/τ.
Speed (c): How fast the waveform is displaced in the direction
of its motion, c = λ/τ = λf.
Angular Frequency (ω): Another measure of frequency, defined
by ω = 2πf = 2π/τ.
Wave number (k): A parameter related to wavelength, defined
by k = 2π/λ. Since λ = cτ the wave number can also be expressed
as k = ω/c. Note that the wave number is analogous to the angular
frequency: one of them is an inverse time and the other is an inverse
length.
Consider a string of length L with fixed end points. (An exam-
ple is a string in your guitar.) Suppose you plucked the string and
took a photograph of the wave. In your photograph the string might
look somewhat like the sketch in Figure 17.1. The displacement from
equilibrium of a particle in the string is y. Obviously this displacement
depends on the horizontal position, x, so y = y(x). (There is no time
dependence yet because this is just a description of the instantaneous
photograph of the wave.) The curve repeats after x = λ. Since the
wave appears to be sinusoidal, let us express it mathematically as a
sine curve. The instantaneous vertical displacement is, then,
 
2πx
y = A sin = A sin(kx).
λ
When x is any multiple of λ/2 the quantity in parentheses is a multiple
of π and there is zero displacement at that point. Such points are called
the nodes of the waveform. The wave in Figure 17.1 has a wavelength
 17.1. A WAVE IN A STRETCHED STRING 589

equal to one-half the length of the string, L. The only waves that will
“fit” onto the string are those with wavelengths λ = (2/n)L where
n = 1, 2, 3, · · · .

λ
y
A

Figure 17.1. A wave in a string of length L. The wave-

length and amplitude are indicated on the figure. The
displacement from equilibrium is y = y(x).

Now imagine that instead of an instant photograph you have a video

of the oscillating string. If you look at a particular point on the string
you will see it moving up and down. Thus the displacement y is also a
function of the time t. As you might suspect, the wave can be expressed
mathematically by an expression having a form such as
y = A sin(kx) cos (ωt) .
This expression is just to give you a general idea of what to expect.
To obtain the actual expression for the wave it is necessary to solve the
equation of motion for the oscillating string. Consider a string of length
L under tension F. The two ends of the string are at x = 0 and x = L.
Ignore gravity. If you pluck the string you can set up a “standing wave”
as in a guitar string. If you look at a plucked guitar string you will
notice that the amplitude of the wave is much less than the length of
the string. That is, y is relatively a small quantity. Furthermore, the
slope dy/dx is also small. Also note that the tension in the string is
the same at every point.
Figure 17.2 shows an infinitesimal portion of the pulse in which the
string is displaced slightly in the vertical
The portion of the string extends from x to x + dx. The mass of
this element is dm = ρdx where ρ is the mass per unit length. The two
ends are displaced vertically by y(x, t) and y(x + dx, t). The element of
the string is subjected to a tension F (x) acting to the left and down
at x (at angle θ) and a tension F (x + dx) acting to the right and up at
x + dx (at angle φ). The horizontal components of these two tensions
must cancel because the element of string is not accelerating to the
590 17. WAVES

F(x+dx)
ds

y(x+dx,t)
y(x,t)
F(x)
dx x+dx
x

Figure 17.2. A small portion of a vibrating string.

right or the left. However, there is a vertical downward component of

force bringing the string back to its equilibrium position.
The magnitude of the tension is the same at both ends, that is,
F (x + dx) = F (x) = F.
The vertical acceleration is ay = ∂ 2 y/∂t2 , so applying Newton’s second
law gives
∂ 2 y(x, t)
ρdx = F (x + dx) sin φ − F (x) sin θ,
∂t2
= F (sin φ − sin θ).
Assuming the angles θ and φ are small (because both y and dy/dx are
small)
. dy ∂y(x, t)
sin θ = tan θ = = ,
dx ∂x
. ∂y(x + dx, t)
sin φ = .
∂x
So,
∂ 2 y(x, t)
 
∂y(x + dx, t) ∂y(x, t)
ρdx =F − .
∂t2 ∂x ∂x
Dividing both sides by dx and taking the limit as dx → 0 leads to
∂ 2y
 
∂ ∂y
ρ 2 =F .
∂t ∂x ∂x
You will see in a little
p while that the speed of a traveling wave in a
stretched string is F/ρ. Anticipating this result, I will replace F/ρ
by c2 . Then the equation of motion has the form:
∂ 2y 2
2∂ y
= c . (17.1)
∂t2 ∂x2
 17.2. DIRECT SOLUTION OF THE WAVE EQUATION 591

This very important, fundamental relation is called the wave equation.

It is a second order partial differential equation giving the displacement
of the string as a function of time and position along the string. I
suggest you memorize it.

Exercise 17.1. A string of length 6 meters is fixed at both ends.

Its mass is 0.1 kg and the tension in the string is 50 N. (a) What is the
wavelength of the longest possible standing wave in this string? (b)
What is the frequency of that wave?
Exercise 17.2. If n is the number of half-wavelengths, show that
the frequency of a standing wave in a string is f = nc/2L.

17.2. Direct Solution of the Wave Equation

Let us now solve Equation (17.1), the wave equation. Many second
order partial differential equations (including this one) can be solved
by a technique called separation of variables.
The basic idea behind the separation of variables method is to ex-
press the unknown function (in our case y = y(x, t)) as the product
of two functions, each depending on only one variable. That is, you
assume that you can write y(x, t) in the following form:
y(x, t) = X(x)T (t).
where X is a function only of x and T is a function only of t. Substitute
this expression into the differential equation to get
∂ 2 X(x)T (t) 2
2 ∂ X(x)T (t)
= c ,
∂t2 ∂x2
∂ 2T ∂ 2X
X 2 = c2 T ,
∂t ∂x2
1 ∂ 2T c2 ∂ 2 X
= .
T ∂t2 X ∂x2
Now the left hand side depends only on t and the right hand side
depends only on x. Since x and t can be varied independently, the only
way the equation can be satisfied is if the right hand side and the left
hand side are equal to the same constant. Call it −ω 2 . The partial
592 17. WAVES

differential equation has then been separated into the following two
ordinary differential equations
c2 d 2 X
= −ω 2 ,
X dx2
1 d2 T
= −ω 2 ,
T dt2
or
d2 X ω 2
+ 2 X = 0,
dx2 c
2
dT
+ ω 2 T = 0.
dt2
Both of these equations have the form of simple harmonic oscillators.
You have seen the SHM equation many times, so I will simply write
down the general solutions
ω ω
X(x) = A cos x + B sin x = A cos kx + B sin kx,
c c
T (t) = C cos ωt + D sin ωt.
The solution for y = y(x, t) is the product of these two functions, thus:
y(x, t) = X(x)T (t) = (A cos kx + B sin kx) (C cos ωt + D sin ωt) .
(17.2)
Note that the partial differential equation (17.1) is satisfied by the four
expressions: 

 cos kx cos ωt
cos kx sin ωt

y(x, t) ∝ .

 sin kx cos ωt
 sin kx sin ωt

The general solution (Equation 17.2) is the sum of all of these. The
coefficients, A, B, C, D are determined from the boundary conditions,
as shown in the next section.

Exercise 17.3. Show that an expression of the form

y = A sin kx + B cos kx
can be expressed as
y = C cos(kx + α).
Express C and α in terms of A and
√ B. (The quantity α is called the
2 2 −1
“phase constant.”) Answer: C = A + B , α = tan (−A/B) .
 17.3. STANDING WAVES 593

17.3. Standing Waves

As an example of how to determine the coefficients in the general
solution (Equation 17.2) consider a standing wave, as in a guitar string
or the string illustrated in Figure 17.1. The string is tied down at the
endpoints so
y(x = 0) = y(x = L) = 0.
Then
y(0, t) = 0 = (A cos 0 + B sin 0) (C cos ωt + D sin ωt) ,
or
0 = AC cos ωt + AD sin ωt.
This requires that A = 0. The general equation has now been reduced
to
y(x, t) = B sin(kx) (C cos ωt + D sin ωt) .
The second boundary condition leads to
y(L, t) = 0 = B sin(kL)(C cos ωt + D sin ωt).
This condition is satisfied if B = 0, but in that case y = 0 everywhere
and at all times! Although it satisfies the differential equation, this
is obviously not a satisfactory solution. (It is the “trivial” solution.)
However, there is another way to ensure that this expression is zero; if
sin(kL) = 0,
the condition is satisfied. This means that kL must be an integer
multiple of π because if sin θ = 0, then θ = 0, π, 2π, 3π · · · . That is,
kL = nπ, n = 1, 2, 3, · · · .
Note that n = 0 is not included because that would require k = 0
and again y(x, t) would be zero at all times. If k can take on the
values nπ/L,
p we must also allow ω to take on a range of values because
ω = ck = F/ρk. That is, ω has the values
s
F nπ cnπ
ωn = = ,
ρ L L
and Equation (17.2) reduces to
nπx
y(x, t) = B sin (C cos ωt + D sin ωt).
L
594 17. WAVES

Finally, combining the constants BC and BD and calling them An and

Bn we write
nπx
y(x, t) = (An cos ωt + Bn sin ωt) sin . (17.3)
L
The differential equation and the boundary conditions are satisfied for
any integer value of n. But note that the coefficients An and Bn are
generally different for different values of n. The general solution to the
differential equation is, of course, the sum of all possible solutions, that
is,
∞
X nπx
y(x, t) = (An cos ωn t + Bn sin ωn t) sin , (17.4)
n=1
L
∞ 
X cnπ cnπ  nπx
= An cos t + Bn sin t sin .
n=1
L L L
To determine the coefficients An and Bn we use the initial condi-
tions, as illustrated in the following example.

Worked Example 17.1. A string of length L is pulled up at

its midpoint a distance b and then released. The displacement of
the string at any position (0 ≤ x ≤ L) and time (t ≥ 0) is given
by Equation (17.4). Determine the coefficients An and Bn .
Solution: The waveform at t = 0 is given by the two straight
lines, y = (2b/L)x for 0 ≤ x ≤ L/2 and y = (−2b/L)x + 2b for
L/2 ≤ x ≤ L. Furthermore, at time t = 0 the string is not moving,
so ∂y/∂t = 0. Consider this second initial condition. It says that
"∞ ∞
#
∂ X nπx X nπx
0 = An sin cos ωn t + Bn sin sin ωn t ,
∂t n=1 L n=1
L
t=0
"∞ ∞
#
X nπx X nπx
0 = An sin (−ωn sin ωn t) + Bn sin (ωn cos ωn t) ,
n=1
L n=1
L
t=0
∞
X nπx
0 = Bn sin .
n=1
L
This is zero for all values of x iff Bn = 0 for all n. So we are left
with ∞
X nπx
y(x, t) = An sin cos ωn t,
n=1
L
 17.3. STANDING WAVES 595

∞
X nπx
y(x, 0) = An sin .
n=1
L
But y(x, 0) is given by the function illustrated in Figure 17.3. As
you will show in Problem 17.1, this function can be expressed as
the following Fourier series:
 
8b πx 1 3πx 1 5πx
y(x, 0) = 2 sin − 2 sin + 2 sin · · · . (17.5)
π L 3 L 5 L
That is,
 
8b πx 1 3πx 1 5πx
y(x, 0) = sin − 2 sin + 2 sin ···
π2 L 3 L 5 L
∞
X nπx
= An sin ,
n=1
L
where
8b nπ
An = 2 2
sin n = 1, 3, 5, · · · .
nπ 2
Finally, the waveform of the plucked string is
∞
X 8b nπ nπx
y(x, t) = 2 2
sin cos ωn t sin .
n=odd
nπ 2 L
The factor sin(nπ/2) is included to give the sign of the term.

x=0 x=L

Figure 17.3. A string plucked at its center. The dis-

placement at L/2 is b.

Exercise 17.4. Equation (17.5) is a Fourier series. Plotting the

terms in this series, we find the integer n is equal to the number of
loops in the standing wave when only the yn term is excited. The
various modes, n = 1, 2, 3, · · · are called the first, second, third · · ·
harmonics. Plot the first four harmonics for a string of length L.
596 17. WAVES

17.4. Traveling Waves

The previous section considered a standing wave. Now consider a
traveling wave, such as a pulse traveling down an infinite string (or
an electromagnetic wave propagating through empty space). Imagine
a very long (perhaps infinite) string that is plucked at some point, or
perhaps shaken at one end. Your experience with ropes and slinkies
tells you that a pulse will move down the string, as pictured in Figure
17.4. If there is no loss of energy, the shape will be repeated at a later
time at a further position, as shown in the bottom panel of the figure.
Suppose the top panel gives the waveform at time t = 0, so the plot is
y = y(x, 0). The bottom plot shows the wave at a later time dt. But
y(x, t) = y(x − ct, 0), that is, the displacement at point x at time t
is the same as the displacement at x − ct at the earlier time. Note
that during the time interval t, the wave has traveled a distance ct.
Therefore, c is the speed at which the wave propagates. This is pcalled
the phase speed. Recall that we identified c with the quantity F/ρ.

y
y(x,0) x

ct
y y(x,t)
x

Figure 17.4. A pulse in a string moving to the right.

It does not change shape, so the displacement function
obeys the condition f (x, t) = f (x + dx, t + dt).

The fact that the displacement at time t is the same as the dis-
placement at x − ct at an earlier time suggests that the waveform for
a wave traveling at speed c will have the functional form

y(x, t) = f (x − ct).

In the figure, the lower panel represents a later time, so the wave is
moving toward the right, that is, toward positive x. A wave traveling
in the opposite direction would have the functional form

y(x, t) = g(x + ct).

 17.4. TRAVELING WAVES 597

For example, a sinusoidal traveling wave moving toward the right can
be represented as
y(x, t) = A sin k(x − ct).
Here I introduced k simply as a constant to make the argument of the
sine dimensionless; however, if you think about it, you will realize that
k is actually the wavenumber, so the expression becomes
y(x, t) = A sin(kx − ωt).
You might plot this expression for a few nearby values of time and
verify for yourself that this indeed is a representation of a traveling
sine wave moving in the positive x direction.
Although the expression
y = f (x − ct)
seems a reasonable way to describe a traveling wave, it remains to
be shown that it satisfies the wave equation for any function f that
depends on x and t in the combination x − ct. For example, a number
of different possible forms of f would be
y(x, t) = A cos k(x − ct)
y(x, t) = Aeik(x−ct)
y(x, t) = A sin k 2 (x − ct)2 .
The assumption that any function of (x − ct) satisfies the wave equa-
tion can be proved by plugging f (x − ct) into the equation and noting
that
∂ 2y
   
∂ ∂ ∂ ∂ ∂(x − ct)
= f (x − ct) = f (x − ct)
∂t2 ∂t ∂t ∂t ∂(x − ct) ∂t
∂ 2 f (x − ct)
 
∂ ∂f (x − ct)
= (−c) = +c2 ,
∂t ∂(x − ct) ∂(x − ct)2
and
2
   
2∂
y 2 ∂ ∂ 2 ∂ ∂f (x − ct) ∂(x − ct)
c = c f (x − ct) = c
∂x2 ∂x ∂x ∂x ∂(x − ct) ∂x
2
 
∂ ∂f (x − ct) ∂ f (x − ct)
= c2 = c2 .
∂x ∂(x − ct) ∂(x − ct)2
The right-hand sides of the two expressions are equal, so the wave
equation is satisfied for any function of x − ct.
A particularly useful function of x−ct that is often used to represent
a traveling wave is
y(x, t) = Aeik(x−ct) = Aei(kx−ωt) .
598 17. WAVES

Note that this can be expressed in terms of sines and cosines using the
Euler relation:
y(x, t) = Aei(kx−ωt) = A [cos(kx − ωt) + i sin(kx − ωt)] .
Although this expression satisfies the wave equation, it is not obvious
how to interpret the imaginary part. Therefore, we avoid the problem
by writing
y(x, t) = Re Aei(kx−ωt) ,
 

that is, only taking the real part of the expression. This formulation is
particularly useful when studying electromagnetic waves.

Exercise 17.5. Argue that the constant k introduced above as a

proportionality constant must actually be the wave number.
Exercise 17.6. Plot y = A sin(kx − ωt) for four nearby values of
time to show that this does indeed represent a wave traveling to the
right. (Select any reasonable values for the parameters, but make sure
that the four values of t are less than a quarter of the period.)

17.5. Standing Waves as a Special Case of Traveling Waves

You have just seen that the solution to the wave equation is any
function of (x ± ct). But the standing wave of Figure 17.1 that was
considered earlier was also a solution of the wave equation and it had
the form
y(x, t) = A sin kx cos ωt + B sin kx sin ωt. (17.6)
(This expression is a very slightly modified version of Equation 17.3.) It
does not look at all like a function of (x − ct)! Are the two expressions
equivalent? The answer is yes, and the underlying reason is that a
standing wave can be considered to be the sum of two traveling waves
moving in opposite directions. The general solution for a traveling wave
can be expressed in the form
y(x, t) = f (x − ct) + g(x + ct) (17.7)
where f is a waveform moving to the right and g is a waveform moving
to the left.
Consider a standing wave set up in a string of length L that is fixed
at both ends. For simplicity, assume that the only nodes are at x = 0
and x = L. The displacement of the string y(x, t) is zero at x = 0 and
 17.5. STANDING WAVES AS A SPECIAL CASE OF TRAVELING WAVES 599

x = L at all times. These two boundary conditions, expressed in terms

of the traveling waves are:
y(0, ct) = 0 = f (−ct) + g(+ct)
y(L, ct) = 0 = f (L − ct) + g(L + ct).
Let ζ = x − ct be the phase of the wave moving to the right. Then
the boundary condition at x = 0 can be expressed as
f (ζ) + g(−ζ) = 0,
or
f (ζ) = −g(−ζ).
This relation tells us that f and g have the same functional form.
For example, if f = A sin(kx − ωt), then g = −A sin[−(kx − ωt)]. The
boundary condition at x = 0 is satisfied because sin(−θ) = − sin(θ).
The boundary condition at x = L is a bit more difficult to analyze.
We have

f (L − ct) + g(L + ct) = 0

But from −g(−ζ) = f (ζ) we appreciate that if ζ = L − ct, then at
x = L,
g(L + ct) + f (L − ct) = 0
g(L + ct) = −f (L − ct)
−f (−(L + ct)) = −f (L − ct)
f (−L − ct) = f (L − ct)
f (L − ct − 2L) = f (L − ct)
f (ζ − 2L) = f (ζ)
That is, the function repeats at spatial intervals of 2L. Clearly, 2L = λ.
The time required for the wave to propagate a distance λ is the
period τ, so
2L
τ= ,
c
and the angular frequency is
πc
ω= .
L
This discussion has been a bit abstract, so let’s discuss an explicit
situation in which the waves are sinusoidal. The point is to show that
the sum of two waves traveling in opposite directions will generate a
standing wave. The traveling wave will be a combination of both sine
600 17. WAVES

and cosine terms and it must be allowed to travel in both directions.

Therefore, the wave can be expressed as
y(x, t) = A sin(kx−ωt)+B sin(kx+ωt)+C cos(kx−ωt)+D cos(kx+ωt).
Once again, the boundary conditions require that y = 0 at x = 0 and
at x = L. If x = 0
0 = −A sin ωt + B sin ωt + C cos ωt + D cos ωt,
= (B − A) sin ωt + (C + D) cos ωt.
This condition holds at all times if
B = A,
C = −D.
and the solution reduces to
y = A [sin(kx − ωt) + sin(kx + ωt)] + C [cos(kx − ωt) − cos(kx + ωt)] .
Imposing the second boundary condition (y = 0 at x = L) gives
0 = A [sin(kL − ωt) + sin(kL + ωt)]+C [cos(kL − ωt) − cos(kL + ωt)] .
It is left as an exercise to show that this is satisfied for ω equal to any
of the values
πc
ωn = n , n = 1, 2, 3 · · · .
L
Recall that for the standing wave, kL = nπ, so this condition can be
written ω = kc, or more generally, as
ωn = kn c.
The sum of the two traveling waves (with y = 0 at x = 0, L) is,
therefore,
y(x, t) = A [sin k(x − ct) + sin k(x + ct)]+C [cos k(x − ct) − cos k(x + ct)] .
Using the trigonometric identities for the sums and differences of angles,
y(x, t) = A [sin kx cos kct − cos kx sin kct + sin kx cos kct + cos kx sin kct]
+C [cos kx cos kct + sin kx sin kct − cos kx cos kct + sin kx sin kct] ,
= 2A sin kx cos kct + 2C sin kx sin kct,
= (2A cos kct + 2C sin kct) sin kx = (2A cos ωt + 2C sin ωt) sin kx.
Recall that the standing wave was given by
y(x, t) = A sin(kx) cos ωt + B sin(kx) sin ωt,
so the two expressions are identical except for the notation of the coeffi-
cients. Thus, we have shown that the standing wave can be considered
the superposition of two identical traveling waves moving in opposite
directions.
 17.6. ENERGY 601

Exercise 17.7. Show that y = 0 when x = L as long as ω = n πc

L
where n = 1, 2, 3 · · · .

17.6. Energy
It is quite easy to determine the energy in a wave. As an example,
consider a standing wave in a string of length L.
The kinetic energy of the mass element ρdx as it oscillates up and
down is  2
1 dy
dT = ρdx .
2 dt
To determine the potential energy note that the element of unstretched
length dx has stretched length ds, so the displacement produces a net
stretch in the string of ds − dx. (See Figure 17.2.) The tension in the
string is F, so the work done to stretch the string (which is equal to
the increase in its potential energy) is
dV = F (ds − dx).
But s  2
p dy
ds = dx2 + dy 2 = 1+ dx,
dx
so s 
 2
dy
dV = F  1 + − 1 dx.
dx
Using the binomial expansion,
"  2 #  2
1 dy . 1 dy
dV = F dx 1 − + ··· − 1 = F dx.
2 dx 2 dx
Consequently, the total energy in a vibrating string of length L is
Z L "  2  2 #
1 ∂y 1 ∂y
E =T +V = ρ + F dx.
0 2 ∂t 2 ∂x
For example, if the wave is described by
πx
y(x, t) = A sin cos ωt,
L
602 17. WAVES

then,
∂y πx
= −ωA sin sin ωt,
∂t L
and
∂y π πx
= A cos cos ωt.
∂x L L
Therefore
Z L
1 2 2  2 πx 2  1 π 2 2  2 πx 2

E = ρω A sin sin ωt + F 2 A cos cos ωt dx,
0 2 L 2 L L
 Z L   2 Z L 
1 2 2 2 2 πx 1 2 π 2 2 πx
= A ρω sin ωt sin dx + A F 2 cos ωt cos dx ,
2 0 L 2 L 0 L
π2
 
1 2L 2 2 2
= A ρω sin ωt + F 2 cos ωt .
2 2 L
Now recall that F/ρ = c2 and ω = πc/L so F π 2 /L2 = ρc2 π 2 /(π 2 c2 /ω 2 ) =
ρω 2 and
1 2L 2
E = A ρω (sin2 ωt + cos2 ωt)
2 2
L
= A2 ρω 2 .
4
Note that the energy is proportional to the amplitude squared.

17.6.1. Energy Flow. Consider a wave passing from one medium

into another. An example is a light wave incident on the window of
your room. As you know, some of the incident light will be transmitted
and some will be reflected. Reflection is a characteristic property of
any kind of wave when it is incident upon the interface of two materials
in which the wave travels at different speeds. For the light wave, glass
has a larger index of refraction than air, and the wave travels slower in
glass. In the case of a wave in a string, the speed depends on the mass
per unit length of the string (ρ).
An analogy to a light wave incident on glass would be a traveling
wave in a string of linear density ρ1 incident on a second string of
different linear density ρ2 . This is illustrated in Figure 17.5.
Imagine a wave coming in from the left (from x = −∞). Call this
the incident wave. At the junction of the two strings (at x = 0), the
incident wave is split into two waves, the reflected wave that moves
back toward x = −∞, and the transmitted wave that continues on
toward x = +∞. Let us denote these three waves by yi , yr , and yt .
 17.6. ENERGY 603

ρ ρ
1 2
x
x=0

Figure 17.5. Two strings of different linear densities

are connected at x = 0.

These are all traveling waves, so they can be represented as follows:

yi = Ai sin(k1 x − ωt),
yr = Ar sin(k1 x + ωt), (17.8)
yt = Ai sin(k2 x − ωt).

Note the sign on the ωt term in the expression for yr and the subscript
on k for yt .
The speed of the wave is different in the two mediums. Since c =
ω/k, it is clear that if the speed changes there will be a change in either
ω or k or both. The three waves presented above all have the same
value of ω and different values of k. You may wonder why k changes
but ω stays the same. The reason is that the frequency of the wave
is determined by the physical mechanism that is generating the wave.
If you are standing at one end of the string and shaking it up and
down, you are controlling the frequency of the wave by how quickly (or
slowly) you move the end of the rope. This you can control. However,
you have no control over the wavelength. The wavelength depends on
the properties of the medium. When the properties of the medium
change the wavelength changes, and so the wavenumber k = 2π/λ also
changes.
The energy of a wave is proportional to the square of its amplitude.
An interesting and important question is how much energy is reflected
at an interface and how much is transmitted into the second medium.
The question can be answered by determining the amplitudes of the
reflected and transmitted waves relative to the incident amplitude. In
other words, it is desirable to determine the ratios Ar /Ai and At /Ai .
This can be done by realizing that both y and dy/dx must be continuous
at the junction between the two strings. The value of y is continuous
because the strings are attached to one another. The reason why dy/dx
is the same on both sides of the junction is that the tension is the same
604 17. WAVES

on both sides. These two conditions can be expressed as

[yi + yr ]x=0 = [yt ]x=0 ,
   
∂yi ∂yr ∂yt
+ = .
∂x ∂x x=0 ∂x x=0
Substituting the expressions in Equations (17.8) we obtain
Ai sin(−ωt) + Ar sin(+ωt) = Ai sin(−ωt),
Ai − Ar = At ,
and
k1 (Ai + Ar ) = k2 At ,
or
1 1
(Ai + Ar ) = At .
c1 c2
Therefore,
Ar c1 − c2
= ,
Ai c1 + c2
At 2c2
= .
Ai c1 + c2
To determine the energy flow through the junction consider the rate
at which work is done by a particle on one side of x = 0 on a particle on
the other side. The transverse component of the force at the interface
is
. ∂y
Fy = F sin θ = F tan θ = F .
∂x
The rate at which work is done is the power and is given by P = F · v.
Therefore,
2
      
 dW  ∂y ∂y ∂y 2ω
  = (Fy )
 dt  = F = FA cos2 (kx − ωt) ,
∂t ∂x ∂t c
where we used k = ω/c. The average over the period is just
dW F A2 ω 2
= .
dt 2c
This expression is valid at any point for a wave given by y = A sin(kx−
ωt). At the junction, on the left the wave is the sum of the incident
and the reflected wave, so
F ω2
 
dW  dW  2 2


= = A i + Ar ,
dt x=0− dt i+r 2c
 17.7. MOMENTUM 605

and on the right

F ω2 2
 
dW  dW 
= = A.
dt x=0+ dt t 2c t
These must be equal if there is no energy loss at the junction.

Worked Example 17.2. Obtain an expression for the reflec-

tion coefficient R defined as the ratio of the intensity of the reflected
wave to the intensity of the incident wave. (R = Ir /Ii ).
Solution: The intensity of a wave is defined as the energy
transported across a normal unit area per unit time, that is, power
per unit area. Since energy is proportional to the amplitude squared
2
A2r /(Area × time) A2r

Ir c1 − c2
R = = 2 = 2 =
Ii Ai /(Area × time) Ai c1 + c2
 2
k1 − k2
= .
k1 + k2

Exercise 17.8. Show that if the tension is continuous at the junc-

tion, the slope dy/dx is continuous.
Exercise 17.9. Obtain the appropriate boundary condition for
two
∂y  ∂y 

strings joined by a knot of mass m. Answer: F ∂x 0+ − ∂x 0− =
2 
m ∂∂t2y  .
0

17.7. Momentum
The material particles in a transverse wave move perpendicular to
the motion of the wave. Thus, you might think that a transverse wave
could not transport momentum longitudinally. However, it is a known
fact that waves in strings do transport momentum. This means that
our assumption of purely transverse motion for a particle in a string
must be wrong. There must be some longitudinal motion as well as
transverse motion.3 This can be easily appreciated conceptually by
(mentally) replacing the string with a series of particles of mass m
connected by massless springs of constant k, as shown in Figure 17.6.
As indicated in the figure, if one particle is plucked so that it is displaced
3Electromagnetic waves in a vacuum are transverse. But the momentum density
associated with these waves is 0 E ×B which is perpendicular to the transverse
oscillations of E and B. For a string, however, perfectly transverse waves would not
have a longitudinal component of momentum.
606 17. WAVES

vertically, it causes the particles on either side to be displaced both

vertically and horizontally.

Figure 17.6. A string modelled as a series of particles

connected by springs. The open circles represent the
equilibrium positions of the particles. The filled circles
represent their positions when one particle is displaced
vertically.

When the plucked particle is released, it sets up a (large ampli-

tude) transverse wave as well as a (small amplitude) longitudinal wave.
Rowland and Pask4 carried out numerical studies using this model of
a string and showed the existence of the longitudinal waves. (It is
interesting to note that a particle in the string actually traces out a
figure eight, moving up and down in a large amplitude oscillation while
oscillating horizontally with a small amplitude oscillation of the same
frequency.)
Consider a traveling wave for which
y = f (x − cT t)
where cT is the transverse speed. Now the length of the string with
ends separated by a distance L0 is
Z L0
RL p RL q dy 2


ds = 0 0 dx2 + dy 2 = 0 0 1 + dx dx
0
dy 2
R L0 1


' L0 + 0 2 dx
dx
where we used the binomial expansion. The horizontal “stretch” of the
dy 2
RL
string is, then ∆x = 0 0 12 dx


dx. But
∂y ∂y ∂f ∂y
= = −cT
∂x ∂f ∂x ∂f
4
David R. Rowland and Colin Pask, “The missing wave momentum mystery,”
Am. J. Phys.,67, 378-388 (1999) and David R. Rowland, “Comment on ‘What hap-
pens to energy and momentum when two oppositely-moving wave pulses overlap?’
by N. Gauthier,” Am. J. Phys., 72, 1425-1429 (2004). It is interesting to note
that the theory of a vibrating string was first developed by Lagrange in 1759 but
there are still interesting aspects of the problem to be explored. For example, Eu-
gene Butikov discusses the potential energy associated with the small longitudinal
stretching of a transverse wave in the paper “Peculiarities in the energy transfer by
waves on strained strings” Physica Scripta, 88, 6 (2013).
 17.8. SUMMARY 607

and
∂y ∂y ∂f ∂y
= = (−cT )
∂t ∂f ∂t ∂f
So
∂y 1 ∂y
=− .
∂x cT ∂t
Therefore,
Z L0  2 Z L0   Z L0  
1 dy 1 ∂y 1 ∂y 1 ∂y ∂y
∆x = dx = − dx = − dt.
0 2 dx 0 2 ∂x cT ∂t 0 2 ∂x ∂t
But Z L0 Z L0
∆x = ẋdt = vx dt.
0 0
That is, the horizontal velocity of these waves is given by
1 ∂y ∂y
vx = − .
2 ∂x ∂t
Consider the longitudinal momentum of an element of the string of
length dx. The mass of this element is ρdx. The momentum in the x
direction of this element is
 
1 ∂y ∂y
(ρdx)(vx ) = ρdx − .
2 ∂t ∂x
Thus the momentum per unit length (or momentum density) is
 
1 ∂y ∂y
g = ρvx = ρ − .
2 ∂t ∂x

17.8. Summary
In this chapter you were exposed to an analysis of waves in strings.
The relations obtained, however, are applicable (with slight modifica-
tions) to other kinds of waves.
Considering the motion in an element of a string and applying New-
ton’s second law, it is easy to derive the relation
∂ 2y F ∂ 2y
= .
∂t2 ρ ∂x2
For a traveling wave, the speed of propagation (phase speed) is c =
p
F/ρ and the relation above becomes
∂ 2y 2
2∂ y
= c ,
∂t2 ∂x2
which is called the wave equation.
608 17. WAVES

The wave equation is solved by the technique of separation of vari-

ables, leading to the general solution
y(x, t) = (A cos kx + B sin kx) (C cos ωt + D sin ωt) .
The “arbitrary” constants A, B, C, D are determined from the bound-
ary conditions. For a standing wave in a string tied down at both ends,
the solution can be expressed as
∞ ∞
X nπx X nπx
y(x, t) = An sin cos ωt + Bn sin sin ωt.
n=1
L n=1
L
The coefficients An and Bn can be determined if the waveform is known
at some initial time as described in Worked Example 17.1.
A traveling wave can always be expressed as a function of (x − ct)
and (x + ct), or (kx ± ωt). For example, a wave moving toward positive
x can be expressed as
y(x, t) = Re Aei(kx−ωt) .
 

A standing wave is the sum of two traveling waves moving in op-

posite directions.
The energy in a wave is proportional to the square of the amplitude
of the wave. For example, a standing wave in a stretched string of
length L has energy
L
E = A2 ρω 2 .
4
When a wave is incident on the interface between two mediums it
will, in general, give rise to both a reflected wave and a transmitted
wave. The relative amplitudes of these waves are:

Ar c1 − c2
= ,
Ai c1 + c2
At 2c2
= ,
Ai c1 + c2
where c1 and c2 are the speeds of the waves in medium 1 and medium
2. The average energy transmitted per unit time is:

F ω2 2
 
dW
= A.
dt t 2c2 t

17.9. Problems
Problem 17.1. Obtain expression 17.5 for the shape of the string
in Figure 17.3.
 17.9. PROBLEMS 609

Problem 17.2. A string of linear mass density ρ and length L

is fixed at both ends and subjected to a tension T. Somehow it is
stretched into a parabolic shape described by y = a(L2 /4 − x2 ), where
x = 0 corresponds to the center of the string and a is a constant having
units of inverse length. (a) Plot y = y(x, 0) for −L/2 ≤ x ≤ L/2. (b)
Express y(x, 0) as a Fourier series. (c) The string is released at time
t = 0. Write an expression for y = y(x, t)
Problem 17.3. A string of length L under tension F0 is released
from rest. Assume that initially the shape of the string was given by
y(x, 0) = A sin πx
L
. Determine y(x, t).
Problem 17.4. A string of length 1 m and mass 50 g is fixed at
both ends. It is subjected to a tension of 50 N. The midpoint of the
string is pulled upwards 2 cm and released. Determine the motion.
Problem 17.5. A uniform string of length L and mass density
ρ is subjected to a tension F0 . The end points are fixed. Somehow
it is given an initial shape that is the arc of a circle. The string is
then released from rest. Since the string is not displaced very far from
its horizontal equilibrium position, the radius of the circular arc is
considerably longer than the length of the chord, so terms smaller than
(L/R)2 can be ignored. Determine the motion.
Problem 17.6. Consider the traveling wave represented by y(x, t) =
i(kx−ωt)
Ae . Show that if k is complex and ω is real, this represents a
damped wave.
Problem 17.7. A vibrating string is oscillating in a viscous medium.
Assume that all portions of the string are subjected to a retarding force
proportional to the speed given by −b ∂y ∂t
dx. (a) Write the equation of
motion. (b) Assume y = Σ∞ φ
n=1 n sin nπx/L (the “normal” solution).
Plug into the equation of motion, multiply by sin mπx/L and integrate
to obtain an equation of φm = φm (t). (c) Solve for φm and write the
solution y = y(x, t).
Problem 17.8. Obtain the Lagrangianand Lagrange’s equations
for a vibrating string. Answer: L = 41 ∞ 2 F0 π 2 2 2
P
n=1 ρl Q̇n − l
n Q n where
the “normal” coordinates Qn are given by Qn = An cos ωn t+Bn sin ωn t.)
Problem 17.9. The Lagrangian for the vibrating string is given as
the solution of Problem 17.8. (a) Obtain the equation of motion. (b)
Solve for the “normal” coordinates Qn .
610 17. WAVES

Problem 17.10. A stretched string has one fixed end, but the
other end (at x = L) is tied to a massless ring that slides on a fric-
tionless vertical rod. Determine the boundary conditions and obtain
an expression for a wave in this string.
Problem 17.11. One end of a string of length L is connected to
a mechanism that makes it oscillate according to y(0, t) = A sin ωt.
The other end (x = L) is fixed. Determine the motion. (Answer:
y = A sin ωt[cos kz − cot kL sin kz].)
Problem 17.12. In Worked Example 17.1 we solved the problem of
a plucked string and obtained a standing wave. Now solve the problem
by assuming the resultant standing wave is the sum of two traveling
waves f (x−ct) and g(x+ct). Make sure initial and boundary conditions
are satisfied.
Problem 17.13. A string of length L is fixed at both ends. At
time t = 0 the displacement at any point is given by u(x) and the
vertical speed is given by v(x). Recall that the general solution can be
written as
∞
X nπx
y(x, t) = (An sin ωn t + Bn cos ωn t) sin .
n=1
L
Show that Z L
2 nπx
An = v(x) sin dx,
ωn L 0 L
and Z L
2 nπx
Bn = u(x) sin dx.
L 0 L
Problem 17.14. A normal mode consists of an oscillation at a
single frequency. The general motion of a vibrating string is
∞
X nπx
y(x, t) = (An sin ωn t + Bn cos ωn t) sin .
n=1
L
This can be expressed as a sum of normal modes by defining the normal
coordinate φn as
φn = An sin ωn t + Bn cos ωn t,
leading to
∞
X nπx
y(x, t) = φn sin .
n=1
L
 17.9. PROBLEMS 611

(a) Express the total energy in terms of the normal coordinates φn . (b)
Express the Lagrangian in terms of the normal coordinates and obtain
the equations of motion in terms of φn .
COMPUTATIONAL PROJECTS
Computational Project 17.1. Using the expression for y(x, 0)
obtained in Worked Example 17.1, sum the first five terms and plot.
Then sum the first ten terms and plot.
Computational Project 17.2. Consider the two traveling waves
y1 (x, t) = A sin(kx − ωt) and y2 (x, t) = A sin(kx + ωt). Show that the
sum of these waves is a standing wave. To do so, generate a three panel
(“movie”) plot showing the two traveling waves and the standing wave
as functions of time.
Computational Project 17.3. Using the model of Figure 17.6
as a basis and displacing the leftmost particle vertically, show the de-
velopment of a traveling wave. Compare your results with Figure 5 of
Rowland and Pask (Am. J. Phys.,67,1999, page 383).
Chapter 18
Small Oscillations (Optional)

18.1. Introduction
An important topic in mechanics is the study of small oscillations
of coupled systems of particles about their equilibrium positions. Ide-
alized examples of such systems are the particles connected by massless
springs in Figure 18.1 and the coupled pendulums shown in Figure 18.2.
These systems were treated in a more elementary manner in Chapter
11; you may wish to refer to that chapter as we go along.1

k k k k k

m m m m

Figure 18.1. Coupled masses. The masses are equal

and are connected by springs having equal spring con-
stants.

18.2. Statement of the Problem

Consider a system described by the generalized coordinates q1 , · · · , qn .
Let the system be conservative and subjected to a potential whose value
depends only on the coordinates, thus:
V = V (q1 , · · · , qn ). (18.1)
1A very good discussion of small oscillations is found in the text by A. L. Fetter
and J. D. Walecka, Theoretical Mechanics of Particles and Continua, McGraw Hill,
New York, 1980, Chapter 4.
613
614 18. SMALL OSCILLATIONS (OPTIONAL)

Assume that the equations of constraint are time independent so the

transformation equations can be written in the form

xi = xi (q1 , · · · , qn ). (18.2)

Then the Lagrangian for the system is a function only of the q’s and
the q̇’s, and does not depend on time:

L = L(q1 , · · · , qn ; q̇1 , · · · , q̇n ). (18.3)

The Lagrange equations of motion can be written in the form involving

the generalized forces (see Equation 4.12)
d ∂T ∂T
− = Qi .
dt ∂ q̇i ∂qi
∂V
Using the fact that Qi = − ∂qi
leads to

d ∂T ∂T ∂V
− =− . (18.4)
dt ∂ q̇i ∂qi ∂qi

1 2
k

m m

Figure 18.2. Coupled pendulums. The bobs have

equal masses and the massless strings are of equal length.
The bobs are connected by a massless spring.

18.2.1. Static Equilibrium. When a system is in equilibrium,

by definition, q̈i = 0. Static equilibrium further requires that q̇i = 0.
In this analysis the value of qi in a state of static equilibrium will be
denoted by qi0 .
Since xi = xi (q1 , · · · , qn ), the total time derivative of xi is

n n
d X ∂xi dqj X ∂xi
xi = ẋi = = q̇j .
dt j=1
∂qj dt j=1
∂qj
 18.2. STATEMENT OF THE PROBLEM 615

Consequently, the kinetic energy can be written as follows:

n
! n !
X1 X1 X ∂x i
X ∂xi
T = mi ẋ2i = mi q̇j q̇k
i
2 i
2 j=1
∂qj k=1
∂qk
n n
( )
1 X X X ∂xi ∂xi
= mi q̇j q̇k
2 j=1 k=1 i
∂q j ∂q k
n n
1 XX
= mjk q̇j q̇k (18.5)
2 j=1 k=1

This gives the expression for kinetic energy in terms of generalized

coordinates for the case of time independent transformation equations.2
Note that the expression for the kinetic energy is in quadratic form.
Also note that mjk is not the mass of a particle.

18.2.2. The Mass Matrix. The quantity mjk in Equation (18.5)

is called the mass matrix and is defined by
X  ∂xi   ∂xi 
mjk = mi , (18.6)
i
∂q j q 0 ∂q k q 0

where it is explicitly indicated that the expressions are evaluated at

points of static equilibrium. The mass matrix is a constant, real, sym-
metric matrix. That is, mjk = mkj = m∗kj .
To be consistent in notation and terminology I should have denoted
the mass matrix by M and referred to mjk as the jkth element of the
mass matrix. Indeed, I will use M and matrix notation a bit later, but
for now I will continue to use a somewhat sloppy terminology and refer
to mjk as the mass matrix.
When the system is in static equilibrium, q̇i = 0 and by (18.5),
∂T
T = 0. Furthermore, T is a function only of the q̇i ’s so ∂q i
= 0. Finally,
n
∂T 1X
= mij q̇j = 0.
∂ q̇i 2 j=1

Then by Equation (18.4) ∂V /∂qi = 0 and so Qi = 0. That is, at static

equilibrium, the coordinate is constant, the kinetic energy is zero, and
the generalized force vanishes. The demonstration of this result may
2A more general (and more complicated) expression given in K. Symon, Me-
chanics, 3rd Ed., Addison Wesley, Reading, Mass, 1971, p 357, Equation 9.9. How-
ever, Equation 18.5 above is applicable to most mechanics problems and is sufficient
for our present purposes.
616 18. SMALL OSCILLATIONS (OPTIONAL)

have been a bit abstract, but the result itself merely states an obvious
fact.

18.2.3. The Potential Matrix. Figure 18.3 shows three possible

shapes for the potential energy near an equilibrium point.

Figure 18.3. Three types of equilibrium: stable, un-

stable, and neutral.

From the figure it is obvious that only for case (a) in which the
potential is a minimum at the equilibrium point is the system in stable
equilibrium. Stable equilibrium implies that if the system is displaced
from the equilibrium point by a small amount it will tend to return
to the equilibrium point. For cases (b) and (c) a similar displacement
(even by an infinitesimal amount) does not result in the system return-
ing to equilibrium.
A small displacement from equilibrium can be represented by

qi = qi0 + ηi , i = 1, · · · , n, (18.7)

where ηi is a small quantity. Equation (18.7) implies that q̇i = η̇i , a

result you will be using shortly.
Using Equation (18.7) the potential energy near an equilibrium
point can be written as

V (q1 , · · · , qn ) = V (q10 + η1 , q20 + η2 , · · · qn0 + ηn ). (18.8)

Carrying out a Taylor’s series expansion about qi0 leads to

n   n n  2 
0 0
X ∂V 1 XX ∂ V
V (q1 , · · · , qn ) = V (q1 , · · · , qn )+ ηj + ηj ηk +· · ·
j=1
∂q j q 0 2 j=1 k=1
∂q j ∂q k q 0

  (18.9)
∂V
But ∂qj
= 0 and V (q10 , · · · , qn0 ) is simply a constant that can be set
q0
equal to zero by redefining the zero point of potential energy. Then,
 18.2. STATEMENT OF THE PROBLEM 617

ignoring terms of order higher than η 2 , Equation (18.9) reduces to

n n 
∂ 2V

1 XX
V = ηj ηk,
2 j=1 k=1 ∂qj ∂qk q0
n n
1 XX
= vjk ηj ηk , (18.10)
2 j=1 k=1
where
∂ 2V
 
vjk = . (18.11)
∂qj ∂qk q0
This expression defines the elements of a constant, real, symmetric
matrix called the potential matrix.
18.2.4. The Lagrange Equations. As mentioned above, for a
system displaced slightly from equilibrium, q̇i = η̇i and the kinetic
energy T is
1X
T = mjk η̇j η̇k .
2 jk
By Equation (18.10), the potential energy is
1X
V = vjk ηj ηk .
2 jk
Consequently, the Lagrangian is
1X
L=T −V = (mjk η̇j η̇k − vjk ηj ηk ) . (18.12)
2 jk
Clearly, the appropriate generalized coordinates for this problem are
the η’s. The Lagrange equations of motion are
d ∂L ∂L
− = 0.
dt ∂ η̇k ∂ηk
From Equation (18.12)
∂L 1X
= mjk η̇j ,
∂ η̇k 2 j
and
∂L 1X
= vjk ηj ,
∂ηk 2 j
so the equations of motion can be written
Xn
(mjk η̈j + vjk ηj ) = 0 k = 1, · · · , n. (18.13)
j=1
618 18. SMALL OSCILLATIONS (OPTIONAL)

Note that mkj and vkj are symmetric so the order of the subscripts
makes no difference. Equations (18.13) are a set of n linear, homoge-
neous, coupled, second order, differential equations with real, constant
coefficients. Each equation has 2n terms involving the elements of the
n × n matrices we have named the mass matrix and the potential ma-
trix. If you can find the solutions to Equations (18.13), that is, if
you can find functions ηi (t) that satisfy (18.13), you have solved the
dynamical problem.
Note that Equation (18.13) can be written in matrix form as
Mη̈ + Vη = 0,
where M and V are the n × n mass and potential matrices and η is an
n-component vector giving the displacements of all the particles from
their equilibrium positions.

Worked Example 18.1. Determine the potential matrix and

the mass matrix for the system of two masses connected by springs
as shown in Figure 11.10.
Solution: Note that the spring between the two masses has
spring constant k3 and the springs attached to the walls and to
masses m1 and m2 have spring constants k1 and k2 respectively.
The positions of the masses (relative to equilibrium) are x1 and x2
(measured in opposite directions). Therefore, the potential energy
is
1 1 1
V = k1 x21 + k2 x22 + k3 (x1 + x2 )2 .
2 2 2
We now evaluate  2 
∂ V
vjk = .
∂qj ∂qk q0
The various elements are:
∂ 2V ∂
v11 = = [k1 x1 + k3 (x1 + x2 )] = k1 + k3 ,
∂x1 ∂x1 ∂x1
∂ 2V ∂
v12 = = [k2 x2 + k3 (x1 + x2 )] = k3 ,
∂x1 ∂x2 ∂x1
∂ 2V ∂
v21 = = [k1 x1 + k3 (x1 + x2 )] = k3 ,
∂x2 ∂x1 ∂x2
∂ 2V ∂
v22 = = [k2 x2 + k3 (x1 + x2 )] = k2 + k3 ,
∂x2 ∂x2 ∂x2
 18.3. NORMAL MODES 619

and so the potential matrix is

 
k1 + k3 k3
.
k3 k2 + k3
The mass matrix is given by
X  ∂xi   ∂xi 
mjk = mi ,
i
∂q j q 0 ∂q k q 0

so the elements are

∂x1 ∂x1 ∂x2 ∂x2
m11 = m1 + m2 = m1 ,
∂x1 ∂x1 ∂x1 ∂x1
∂x1 ∂x1 ∂x2 ∂x2
m12 = m1 + m2 = 0,
∂x1 ∂x2 ∂x1 ∂x2
∂x1 ∂x1 ∂x2 ∂x2
m21 = m1 + m2 = 0,
∂x2 ∂x1 ∂x2 ∂x1
∂x1 ∂x1 ∂x2 ∂x2
m22 = m1 + m2 = m2 .
∂x2 ∂x2 ∂x2 ∂x2
The mass matrix is  
m1 0
.
0 m2

Exercise 18.1. Consider a block of mass m connected to a spring

of constant k. The block is allowed to slide on a frictionless horizontal
surface, as illustrated in Figure 4.3. Determine the mass matrix and
the potential matrix for this one-dimensional system.
Exercise 18.2. Determine the mass matrix and the potential ma-
trix for a system of one mass and two springs arranged as in Figure
18.1. (Note that all the springs have the same spring constant and all
the masses have the same mass.)

18.3. Normal Modes

We now obtain solutions to Equations (18.13). This is a rather
formal procedure, but in the course of arriving at the answer, you will
encounter a number of important concepts such as the normal coordi-
nates and the modal matrix. You will find that the determination of
eigenvalues is an important part of the solution, so once again you will
need to diagonalize matrices and use various mathematical techniques
that you studied in previous chapters.
620 18. SMALL OSCILLATIONS (OPTIONAL)

18.3.1. One-Dimensional Motion. Let us begin with a one-

dimensional problem, that is, a system characterized by a single gen-
eralized coordinate q. For the sake of an explicit example, consider a
simple pendulum; in that case, the generalized coordinate is the angle
the string makes with the vertical, i.e., q = θ. From Figure 18.4 you
can see that the transformation equations are:
x = x1 = l sin θ,
y = x2 = l cos θ.

θ l

y m

Figure 18.4. A simple pendulum.

Using Equation (18.6) to evaluate mjk we note that for a single

coordinate, j = 1 and k = 1 and qj = qk = θ. Thus,
2    
X ∂xi ∂xi
m11 = mi ,
i=1
∂q j q 0 ∂q k q 0
 
∂x ∂x ∂y ∂y
= m + ,
∂θ ∂θ ∂θ ∂θ θ=0
= m (l cos θ)2 + (l sin θ)2 ,
 

= ml2 .
Similarly, since V = −mgy = −mgl cos θ, Equation (18.11) leads to
 2   
∂ V ∂ ∂
v11 = =− (mgl cos θ) ,
∂qj ∂qk q0 ∂θ ∂θ θ=0
v11 = mgl cos θ |θ=0 = mgl.
For this problem, the mass matrix and the potential matrix are 1 × 1
matrices having a single element. In general, if the system is charac-
terized by a single coordinate, you will find that mjk = m and vjk = κ
where m and κ are real numbers (but not necessarily equal to the mass
and the force constant) and the equation of motion (18.13) reduces to
mη̈ + κη = 0, (18.14)
or
κ
η̈ = − η.
m
 18.3. NORMAL MODES 621

(I am using κ for the force constant to avoid confusing it with k, the

subscripted index.)
Going back to the pendulum, the mass matrix is ml2 , the potential
matrix is mgl and η = θ. So Equation (18.14) is
ml2 θ̈ + mglθ = 0,
or
g
θ̈ = − θ,
l
as expected.
κ
Let us consider the general one-dimensional case, η̈ = − m η, in more
detail. First of all, note that κ > 0 leads to simple harmonic motion,
as for a mass m on a spring of constant κ. Of course, you know the
solution, but bear with me while I solve the differential equation in a
way that will easily carry over to systems with many particles.
κ
To solve η̈ = − m η it is convenient to introduce a complex variable
z which is assumed to satisfy the differential equation. That is,
κ
z̈ = − z.
m
If you can find a solution (any solution) to this equation, then the
solution to the original problem is
η = Re(z),
because the coefficients in the equation are all real.
A possible solution to the z equation is
z = z 0 eiωt . (18.15)
Plugging this into the differential equation for z you will find that it is
indeed a solution as long as the following condition is met:
κ 0
z = ω2z0. (18.16)
m
This condition can be interpreted as an eigenvalue equation for ω 2 . It is
satisfied if z 0 = 0, but that is the trivial solution and is of no interest.
The other possibility is
p
ω = ± κ/m = ±ω1 , (18.17)
indicating there are two possible frequencies at which the system can os-
cillate. The mathematical solutions corresponding to these two values
of frequency are z+ exp(iω1 t) and z− exp(−iω1 t). Since the differential
622 18. SMALL OSCILLATIONS (OPTIONAL)

equation is linear, the general solution is the linear superposition or

sum of all possible solutions. Therefore, the general solution is
(1) (1)
z(t) = z+ eiω1 t + z− e−iω1 t , (18.18)
where the superscripts (1) on the amplitudes z+ and z− serve to re-
mind us that these amplitudes correspond to the frequencies ±ω1 . The
quantity of interest is η = Re(z), and
1
η = Re(z) = (z + z ∗ )
2
1 nh
(1) (1)∗
i
iω1 t
h
(1)∗ (1)
i
−iω1 t
o
= z+ + z− e + z+ + z− e . (18.19)
2
It is convenient to introduce another complex number in polar form,
defined as follows
∗
ρeiφ = z+ + z− .
Note that ρ and φ are real numbers. In terms of ρ and φ the solution
(18.19) can be written as
1
η = ρ(1) ei(φ+ω1 t) + e−i(φ+ω1 t) = ρ(1) cos(ω1 t + φ).
 
(18.20)
2
This emphasizes the fact that the motion is simple harmonic. The
general solution (18.20) involves two real “arbitrary” constants ρ(1)
and φ, but it is interesting
p to note that it involves only the positive
eigenvalue ω1 = + k/m. (I put the superscript (1) on ρ for future
convenience.)

Exercise 18.3. Show that if z = z 0 eiωt , then condition (18.16)

must be met.
Exercise 18.4. Determine the relationships between ρ, φ of Equa-
tion (18.20) and z+ , z− of Equation (18.18).

18.3.2. Solving the Coupled Equations. Having solved the

one-dimensional problem, let us go on to a set of n coupled equations
having the form
Xn
(mkj z̈j + vkj zj ) = 0, k = 1, · · · , n. (18.21)
j=1

Here the problem is formulated in terms of the complex variable z,

but keep in mind that the physical solution is ηk = Re(zk ). Equations
 18.3. NORMAL MODES 623

(18.21) are the equations of motion for a system of coupled oscillators

such as the masses on springs of Figure 18.1. I will now assume that
all the particles are oscillating with exactly the same frequency. As
you remember from Chapter 11, this type of motion is called a normal
mode. If the assumed frequency is ω, the normal mode solutions are
zj = zj0 eiωt . (18.22)
Plugging Equation (18.22) into the equations of motion (18.21) gives
Xn
(mkj (−ω 2 )zj0 eiωt + vkj zj0 eiωt ) = 0,
j=1
or
n
X
(vkj − ω 2 mkj )zj0 = 0. (18.23)
j=1
In matrix form we can write this equation as
(V − ω 2 M)z 0 = 0,
which is easily recognized to be an eigenvalue equation with eigenvalues
ω 2 and eigenvectors z0 (compare to Equation 15.35).
Equations (18.23) are a set of n linear homogeneous coupled alge-
braic equations. As you have seen previously, such a set of equations
has a nontrivial solution if and only if the determinant of the coeffi-
cients is zero. That is, a nontrivial solution exists iff
 
det vkj − ω 2 mkj  = 0. (18.24)
The values of ω 2 that satisfy this relation will be the eigenvalues.
Evaluating the determinant of Equation (18.24) leads to an nth
order polynomial in ω 2 so there will be n values of ω 2 that satisfy the
condition. Let us denote these n allowed values of ω 2 by ωs2 where
s = 1, 2, · · · , n.
The eigenvalues ωs2 must be real because Equation (18.22) repre-
sents a stable solution only if the ω’s are real. If ωs is zero or complex,
then according to Equation (18.22) the solution zj will be constant or
exponentially increasing or decreasing. These conditions correspond to
the neutral or unstable situations depicted in Figure 18.3.
To prove the ω’s are real, consider a particular eigenvalue, say ωs .
(s)
Let zj be the corresponding solutions to the set of equations (18.23).
 ∗
(s)
Multiply each equation in the set by zk and sum over k to obtain
n
X (s)∗ (s)∗ (s)
(zk vkj − ωs2 zk mkj )zj = 0,
k,j
624 18. SMALL OSCILLATIONS (OPTIONAL)

or
n
X n
X
(s)∗ (s) (s)∗ (s)
zk vkj zj = ωs2 zk mkj zj .
k,j k,j

Consequently
Pn (s)∗ (s)
k,j zk vkj zj
ωs2 = Pn (s)∗ (s)
. (18.25)
z
k,j k mkj z j
If you take the complex conjugate of this expression and use the fact
that the vkj and mkj are real, you obtain
Pn (s) (s)∗
2 ∗

k,j zk vkj zj
ωs = Pn (s) (s)∗
.
z
k,j k mkj z j

But vkj and mkj are elements of symmetrical matrices so

Pn (s) (s)∗ Pn (s)∗ (s)
2 ∗ k,j zk vjk zj k,j zj vjk zk
= ωs2 ,


ωs = Pn (s) (s)∗
= Pn (s)∗ (s)
(18.26)
k,j zk mjk zj k,j zj mjk zk
which proves that ωs2 is real. If ωs2 is real, then ωs is also real as long
as ωs2 ≥ 0. Therefore, the condition for a stable solution is
Pn (s)∗ (s)
k,j zk vkj zj
Pn (s)∗ (s)
≥ 0. (18.27)
k,j zk mkj zj

The elements of the mass matrix (mkj ) are positive (as is evident from
the definition, Equation 18.6) so the denominator in (18.27) is positive.
The numerator will be positive or negative depending on the value of
vkj (which is defined in Equation 18.11):
 2 
∂ V
vkj = vjk = .
∂qj ∂qk q0
If the potential energy V is a minimum at equilibrium point q 0 then its
second derivative is positive, so vkj > 0 and condition (18.27) is met.
Therefore, stable oscillations can only occur around points where the
potential energy is a minimum - a result that is intuitively obvious.
Returning to the problem of determining the motion of the system,
let us pick a particular eigenvalue (ωs2 ) and consider the form of the
solutions of Equations (18.23) which can be written in a more general
way as
Xn
(s)
(vkj − ωs2 mkj )zj = 0. (18.28)
j=1
 18.3. NORMAL MODES 625

As noted previously, these equations have a nontrivial solution iff the

determinant of the coefficients vanishes (Equation 18.24). Recall that if
the determinant is zero, the equations are not all linearly independent.
The nonindependent equation(s) can be discarded. This can be done
as follows: Renumber the equations so that the nth equation is the one
you plan to discard. Your system of equations now looks like this:
a11 z1 + a12 z2 + · · · + a1n zn = 0, (18.29)
a21 z1 + a22 z2 + · · · + a2n zn = 0,
..
.
an1 z1 + an2 z2 + · · · + ann zn = 0.
Divide all the equations by zn and discard the last equation. You will
now have a set of n − 1 equations having the form
a11 z1 /zn + a12 z2 /zn + · · · = −a1n , (18.30)
a21 z1 /zn + a22 z2 /zn + · · · = −a2n ,
..
.
This is a set of n − 1 inhomogeneous equations that can be solved using
Cramer’s rule for the ratios z1 /zn , z2 /zn , · · · .
By Equation (18.28) the coefficients akj in Equations (18.30) are

akj = vkj − ωs2 mkj ,

and they are real. Consequently, the Cramer’s rule solutions zj /zn
are real. But the zj were assumed to be complex! (See Equation
18.22.) How can the ratio of two complex numbers be real? This can
only happen if the complex part of the two numbers is the same. To
appreciate this, write the complex zj ’s in polar form
(s) (s)
zj = ρj eiφs . (18.31)

The value of φs is the same for all the zk ’s. But if eiφs is a common
factor, you can divide all of the equations by it and write Equations
(18.28) in the form
n
X n
X
(s) (s)
vkj ρj = ωs2 mkj ρj , k = 1, 2, · · · , n. (18.32)
j=1 j=1

(s)
Equation (18.32) is the eigenvector equation. Note that the ρj are a
set of n real quantities. They can be ordered into a column matrix,
626 18. SMALL OSCILLATIONS (OPTIONAL)

thus,
 (s) 
ρ1
(s)
(s)
 ρ2 
ρ = . (18.33)
 
..
 . 
(s)
ρn
Here ρ(s) is an eigenvector.
In like manner, the eigenvector equation corresponding to a different
frequency ωt2 is
Xn n
X
(t) 2 (t)
vkj ρk = ωt mkj ρk . (18.34)
k=1 k=1
P (t)
Now operate on Equation (18.32) with k ρk and on (18.34) with
P (s)
j ρj to obtain
X (t) (s)
X (t) (s)
ρk vkj ρj = ωs2 ρk mkj ρj , (18.35)
k,j k,j

and X X
(s) (t) (s) (t)
ρj vkj ρk = ωt2 ρj mkj ρk . (18.36)
k,j k,j

The left-hand sides of Equations (18.35) and (18.36) are identical, so

subtracting the two equations yields
X (t) (s)
0 = (ωs2 − ωt2 ) ρk mkj ρj . (18.37)
k,j

If the eigenvalues are different, that is, if ωs2 6= ωt2 , then

X (t) (s)
ρk mkj ρj = 0, s 6= t. (18.38)
k,j

This is an orthogonality condition on the eigenvectors. You can mul-

tiply ρ(s) and ρ(t) by appropriate constants such that the following
orthonormality condition is met
X (t) (s)
ρk mkj ρj = δts . (18.39)
k,j

The ρ vectors are then said to be normalized.

Considering the form of the differential equation (18.21), it is clear
that you can multiply the solution by an arbitrary constant. Therefore,
the general solution to (18.21) is
(s) (s)
zj = C (s) ρj eiφs , j = 1, · · · , n. (18.40)
 18.4. MATRIX FORMULATION 627

There are n solutions of this form for each of the eigenvalues ωs . Each
(s)
such solution has (in general) a different value of ρj but the factors
C (s) and φs are common to all the solutions corresponding to the same
eigenvalue. For example, for the system illustrated in Figure 18.1, the
displacements of particles 1, 2, 3, · · · , j, · · · , n correspond to the real
(s) (s) (s) (s)
parts of z1 , z2 , · · · , zj , · · · zn where all the particles are assumed to
oscillate at the same (normal) frequency ωs .
The solution of the coupled problem is obtained as a linear super-
position of all the particular solutions, thus,
Xn h i
(s) (s)
zj (t) = z+j eiωs t + z−j e−iωs t , j = 1, · · · , n, (18.41)
s=1
where + −
(s) (s) (s) (s) (s) (s)
z+j = C+ ρj eiφs , z−j = C− ρj eiφs . (18.42)
The general solution is the sum over all the normal modes. It is
clear that if the normal mode solutions satisfy the differential equation,
then a sum of normal mode solutions will also satisfy the differential
equation. This is just the principle of superposition.3
The actual physical motion, ηj , is obtained by taking the real part
of equation (18.41), yielding
n
X (s)
ηj (t) = C (s) ρj cos(ωs t + φs ). (18.43)
s=1

Exercise 18.5. Show that the condition for stable equilibrium is

not met at a maximum of the potential energy.

18.4. Matrix Formulation

In the previous section we solved the problem of coupled oscillators
algebraically, although we did mention that some of the equations can
be formulated in terms of matrices and we did employ the terminology
of matrix analysis, defining eigenvalues and eigenvectors. Nevertheless,
we carried out the analysis without the benefit of matrix notation. As
you will now see, the problem we are considering can be formulated in
3For a differential equation, the general solution is the sum over all the inde-
pendent solutions. Since the normal modes are linearly independent, the general
solution can be expressed as a sum over all normal modes, as in (18.41).
628 18. SMALL OSCILLATIONS (OPTIONAL)

an elegant and very concise way by using matrices. The drawback is

that the analysis tends to be rather abstract.
Recall that if A is an n × n matrix and x is a column matrix having
n elements, then xT is a row matrix, Ax is a column matrix, and xT Ax
is a scalar.
As defined in Sections 18.2.2 and 18.2.3 M is the mass matrix
(with elements mjk ) and V is the potential matrix (with elements vjk ).
In terms of V and M, the eigenvector equation (18.32) is

V − ωs2 M ρ(s) = 0.


(18.44)
The orthonormality condition (18.39) is

T
ρ(s) Mρ(t) = δst ,
and the general solution (18.43) can be expressed as a linear combina-
tion of the normal mode solutions,
X
η(t) = ρ(s) C (s) cos(ωs t + φs ). (18.45)
s

The constants C (s) and φs can be associated with the initial conditions
η(0) and η̇(0) by (18.45) and its derivative:
X
η(0) = ρ(s) C (s) cos φs ,
s

X
η̇(0) = − ρ(s) C (s) ωs sin φs .
s

Usually, you are given η(0) and η̇(0) and asked to find C (s) and φs .

T
These quantities are easily determined by multiplying by ρ(t) M
from the left and using the orthogonality conditions.

T X
T
ρ(t) Mη(0) = ρ(t) Mρ(s) C (s) cos φs = C (t) cos φt ,
s

and

T X
T
ρ(t) Mη̇(0) = − ρ(t) Mρ(s) ωs C (s) sin φs = −ωt C (t) sin φt .
s

18.4.1. The Modal Matrix. Solving the problem of n coupled

oscillators introduced n eigenvectors ρ(s) , each corresponding to a dif-
(s)
ferent frequency ωs . Further, each eigenvector had n elements, ρj , j =
 18.4. MATRIX FORMULATION 629

1, · · · , n. Therefore, you can construct an n × n matrix using the eigen-

vectors as columns, thus:
 (1) (2) (3) (n) 
ρ1 ρ1 ρ1 · · · ρ1
 
ρ(1) ρ(2) ρ(3) · · · ρ(n)
 ρ(1) ρ(2) ρ(3) · · · ρ(n)  
  ↓ ↓ ↓ ↓
A= 2 2 2 2

=

 .
..  ↓ ↓ ↓ ↓


ρ
(1)
ρ
(2)
ρ
(3)
··· ρ
(n) ↓ ↓ ↓ ↓
n n n n
(18.46)
A is called the modal matrix. If the elements of A are denoted aij ,
(j)
then aij = ρi . The modal matrix has some interesting and useful
properties.
Property 1: The matrix product AT MA is the unit matrix:
X X X (i) (j)
AT MA ij = aTik mkl alj =


aki mkl alj = ρk mkl ρl = δij ,
kl kl kl
(18.47)
or
AT MA = 1. (18.48)
T
Consequently, A MA is an n × n diagonal matrix with ones along
the diagonal. This means that the modal matrix diagonalizes the mass
matrix.
Property 2. The matrix product AT VA is a diagonal matrix whose
elements are the normal mode frequencies squared:
( )
X X (i) (j)
X (i)
X (j)
AT VA ij = aTik vkl alj =


ρk vkl ρl = ρk vkl ρl .
kl kl k l

But the eigenvector equation (18.32) allows us to write the summation

(j)
over l as ωj2 l mkl ρl , so
P
( )
X (i) X (j)
X (i) (j)
AT VA ij = ρk ωj2 = ωj2 ρk mkl ρl = ωj2 δij ,


mkl ρl
k l kl
or
ω12 0 0 ···
 
 0 ω22 0 ··· 
AT VA = 
 0 0 ω32
 2
· · ·  ≡ ωD . (18.49)
.. .. .. ..
. . . .
Thus, the modal matrix diagonalizes the potential matrix, generating
a matrix whose only non zero elements are the squares of the normal
mode frequencies.
630 18. SMALL OSCILLATIONS (OPTIONAL)

18.5. Normal Coordinates

I expressed the positions of the oscillating mass points in various
ways: The Cartesian coordinate, xi , is the distance from the origin to
mi . The distance of a mass point from its equilibrium position is denoted
ηi and it is the first of several generalized coordinates. I referred to the
(s)
ηi as the “physical” coordinates. The zi are another set of generalized
coordinates which are allowed to be complex. They are the position
coordinates obtained from the equation of motion assuming a normal
mode solution, i.e., one in which all the particles oscillate at the same
frequency ωs . The physical interpretation of the zi is a bit difficult to
appreciate. Since they are complex, they can be expressed in polar
form as
(s) (s)
zi = ρi eiφs .
(s)
The ρi are elements of the eigenvector ρ(s) which describes the posi-
tion of all the particles in a particular normal mode. The relationship
between the ηi and the eigenvectors for any general motion of the sys-
tem is the real part of a superposition of all the normal modes, thus:
n
X (s)
ηi = C (s) ρi cos(ωs t + φs ).
s=1

Given this multiplicity of ways to express the positions of the particles,

it may seem a bit ludicrous that I should introduce yet another set
of generalized coordinates. However, as you will appreciate shortly,
the normal coordinates (denoted by ζi ) are a very convenient way to
formulate the problem of coupled oscillators. The normal coordinates
are related to the generalized physical coordinates ηi by the following
defining equation
η(t) = Aζ(t). (18.50)
Multiplying by AT M isolates ζ (see Equation 18.48). The result is
ζ(t) = AT Mη(t). (18.51)
Since AT and M are matrices with constant, real elements, the nor-
mal coordinates ζi are linear combinations of the original generalized
coordinates ηi .
Using the normal coordinates, the Lagrange equations of motion
take on a particularly simple form. Equation (18.12) for the Lagrangian
was
1X
L=T −V = (mij η̇i η̇j − vij ηi ηj ) .
2 ij
 18.5. NORMAL COORDINATES 631

This can be expressed as

X X
2L = η̇i mij η̇j − ηi vij ηj ,
ij ij

or
2L = η̇ T Mη̇ − η T Vη. (18.52)
But η(t) = Aζ(t) and η̇(t) = Aζ̇(t), so
 T  
2L = Aζ̇ M Aζ̇ − (Aζ)T V (Aζ) ,
= ζ̇ T AT MA ζ̇ − ζ T AT VA ζ,

= ζ̇ T ζ̇ − ζ T ωD
2
ζ. (18.53)
Or, in component form,
n
1 X 2 
L= ζ̇i − ωi ζi2 . (18.54)
2 i=1

Note that using the modal matrix to define the new generalized coor-
dinates has, in a sense, diagonalized the Lagrangian. Note also that
the ωi ’s in Equation (18.54) are the set of normal frequencies for the
system.
In terms of the normal coordinates, ζi , the Lagrange equations of
motion are
n
! n 
!
d ∂ 1 X 2  ∂ 1 X 
ζ̇ − ωi ζi2 − ζ̇ 2 − ωi ζi2 = 0, (18.55)
dt ∂ ζ̇i 2 i=1 i ∂ζi 2 i=1 i
or
ζ̈i = −ωi2 ζi , (18.56)
which is the equation for simple harmonic motion. Therefore diagonal-
izing the Lagrangian decoupled the equations of motion and reduced
the problem to a set of n independent, decoupled, simple harmonic os-
cillator equations. Each normal coordinate ζi oscillates independently
with angular frequency ωi .
The derivation has been perfectly general and implies that any me-
chanical system undergoing small amplitude oscillations around points
of stable equilibrium can be described in terms of normal coordinates.
The solution to Equation (18.56) can be expressed in component
form as
ζi = C (i) cos(ωi t + φi ), i = 1, · · · , n (18.57)
632 18. SMALL OSCILLATIONS (OPTIONAL)

or in the form of a column vector as

 
C (1) cos(ω1 t + φ1 )
 C (2) cos(ω2 t + φ2 ) 
ζ= . (18.58)
 
..
 . 
(n)
C cos(ωn t + φn )
The physical generalized coordinates are obtained from the definition
η(t) = Aζ(t), which can be written as
  
ρ(1) ρ(2) ρ(3) · · · ρ(n) ζ
 ↓ ↓ ↓ ↓  ↓ 
η=  ↓

 ↓
.
↓ ↓ ↓ 
↓ ↓ ↓ ↓ ↓
Therefore, X (j)
ηi = ρi C (j) cos(ωj t + φj ). (18.59)
j

Comparing Equations (18.59) and (18.57) you can see that the normal
coordinates ζi are the coefficients of the eigenvectors ρi in the expansion
of ηi .

18.6. Coupled Pendulums: An Example

Consider the two pendulums connected by a spring of constant k
illustrated in Figure 18.5. Note that sin θ = η/l. The kinetic energy of
the bobs is
   2  1 
1 2 .
= m η̇12 + η̇22 .

T = m lθ̇1 + lθ̇2 (18.60)
2 2
The potential energy is the gravitational potential energy VG plus the
potential energy in the spring VS . It is easy to obtain expressions for
these quantities in terms of η1 and η2 .
VG = mgl [(1 − cos θ1 ) + (1 − cos θ2 )] ,
h 1   1 i
= mgl 1 − 1 − sin2 θ1 2 + 1 − 1 − sin2 θ2 2 ,
 

. 1
= mgl sin2 θ1 + sin2 θ2 .


2
Therefore, to second order in small quantities,
mg 2
VG = (η + η22 ). (18.61)
2l 1
The potential energy of the spring is 21 k(d−d0 )2 where d is the distance
between the masses and d0 is the unstretched length of the spring.
 18.6. COUPLED PENDULUMS: AN EXAMPLE 633

These quantities are illustrated in Figure 18.5, from which you can see
that
1
d = y 2 + (d0 − η1 + η2 )2 2 ,

! 21
2(η1 + η2 ) (η2 + η1 )2 y 2
= d0 1 + + + 2 + ··· ,
d0 d20 d0
.
= d0 + η2 − η1 .
To second order in small quantities,
1 1
VS = k(d − d0 )2 = k(η2 − η1 )2 . (18.62)
2 2
Having obtained the kinetic and potential energies, the Lagrangian
is
1   mg 2 1
L = T − V = m η̇12 + η̇22 − (η1 + η22 ) − k(η2 − η1 )2 . (18.63)
2 2l 2
The equations of motion are
 mg 
mη̈1 + k + η1 − kη2 = 0, (18.64)
l
 mg 
mη̈2 + k + η2 − kη1 = 0.
l

k m
m

Figure 18.5. Coupled Pendulums.

To obtain a solution, we look for normal modes, i.e., a solution in

which both bobs oscillate at the same frequency ω. That is, we assume
that
ηi = Cρi cos(ωt + φ), i = 1, 2. (18.65)
Substituting into the equation of motion yields the following eigenvalue
equations:
 mg 
k+ − mω 2 ρ1 − kρ2 = 0, (18.66)
l
 mg 
−kρ1 + k + − mω 2 ρ2 = 0.
l
634 18. SMALL OSCILLATIONS (OPTIONAL)

Remember that the eigenvector equation has the form

V − Mω 2 ρ = 0.


(18.67)
For this problem ρ is the two component column vector
 
ρ1
ρ= . (18.68)
ρ2
By comparing Equations (18.66) and (18.67) it is clear that
 
1 0
M=m , (18.69)
0 1
and
k + mg
 
−k
V= l . (18.70)
−k k + mg
l
The eigenvalue equations (18.66) are a set of coupled homogeneous
algebraic equations having a non trivial solution iff the determinant of
the coefficients vanishes. That is,
 k + mg − mω 2
 
l
−k 
mg 2  = 0.
 
 −k k + l − mω
This yields a quadratic equation for ω 2 with solutions
r
g
ω1 = , (18.71)
l
r
g k
ω2 = +2 . (18.72)
l m
To obtain the eigenvectors corresponding to these two eigenvalues, plug
them into Equation (18.66) and obtain
(1) (1)
ρ1 = ρ2 for ω = ω1 , (18.73)
and
(2) (2)
ρ1 = −ρ2 for ω = ω2 . (18.74)
Equation (18.73) represents motion in which the bobs oscillate in phase
with equal amplitudes at the frequency of a free pendulum. Equation
(18.74) represents the bobs oscillating out of phase (amplitudes equal
but opposite) with a frequency larger than that of a free pendulum
(see Equation 18.72). These normal modes are easily demonstrated
with simple apparatus. They are not, however, the most general type
of motion. To describe the general motion go back to the equations of
motion (Equations 18.64). Note that the vector
 
ρ1
ρ=
ρ2
 18.6. COUPLED PENDULUMS: AN EXAMPLE 635

can be multiplied by a constant without affecting anything except the

amplitude. This allows you to normalize the eigenvectors so that the
orthonormality condition

T
ρ(s) Mρ(t) = δst
is √
satisfied. It is easy to show that the normalization constant is
1/ 2m. Thus,
 
(1) 1 1
ρ = √ , (18.75)
2m 1
 
(2) 1 1
ρ = √ .
2m −1
Now construct the modal matrix A using these normalized eigen-
vectors as the columns of the matrix,
 
1 1 1
A= √ .
2m 1 −1
The normal coordinates are obtained from ζ = AT Mη and are given
by the following linear combinations of the η’s:
r
m
ζ1 = (η1 + η2 ), (18.76)
2
r
m
ζ2 = (η1 − η2 ).
2
According to Equation (18.54), the Lagrangian is
2
1X 2
L= (ζ̇ − ωj2 ζj2 ),
2 j=1 j
and the equations of motion are
ζ̈j = −ωj2 ζj , j = 1, 2. (18.77)
The general solution is
ζ1 = C (1) cos (ω1 t + φ1 ) , (18.78)
(2)
ζ2 = C cos (ω2 t + φ2 ) .
The constants C (1) , C (2) , φ1 , and φ2 are determined from the initial
conditions. For example, if bob 1 is displaced pa distance α and released
(1) (2)
from rest, φ1 = φ2 = 0 and C = C = α m/2. Therefore,
r  
m cos ω1 t
ζ=α .
2 cos ω2 t
636 18. SMALL OSCILLATIONS (OPTIONAL)

The original (“physical”) generalized coordinates (obtained from η =

Aζ) are
α
η1 (t) = (cos ω1 t + cos ω2 t) (18.79)
2
α
η2 (t) = (cos ω1 t − cos ω2 t)
2
or
    
1 1
η1 (t) = α cos (ω2 − ω1 )t cos (ω2 + ω1 )t , (18.80)
2 2
    
1 1
η2 (t) = α sin (ω2 − ω1 )t cos (ω2 + ω1 )t .
2 2
The first term (in square brackets) is a slowly varying amplitude and
the second term (involving ω1 + ω2 ) is a rapidly varying term. The
functions η1 and η2 are plotted as functions of time in Figure 18.6.

η1
time

η2

time

Figure 18.6. Oscillations of coupled pendulums.

Exercise 18.6. The system illustrated in Figure 18.5 has bobs of

mass 0.1 kg and a spring whose force constant is 5 N/m. The pendulum
string and the unstretched spring length are both 0.3 m. Determine
the normal mode frequencies. Answers: 5.72/sec and 11.52/sec.

18.7. Many Degrees of Freedom

As a final example, consider a system with many degrees of freedom:
specifically, the transverse oscillations of mass points on a stretched
massless string, as illustrated in Figure 18.7.
 18.7. MANY DEGREES OF FREEDOM 637

Figure 18.7. A massless string with point masses sep-

arated by equal distances.

18.7.1. Statement of the Problem. Assume the N particles

have mass m and are uniformly distributed on the string. Let a be
the horizontal distance between the particles and let yj be the vertical
displacement of particle j from its equilibrium position. Let τ be the
constant, uniform tension in the string. See Figure 18.8.
The equation of motion of the ith particle can be obtained by ele-
mentary means, that is, by applying F = ma. This yields
mÿj = τ sin φ − τ sin θ. (18.81)
But
1
sin φ ' tan φ = (yj+1 − yj ) ,
a
and
1
sin θ ' tan θ = (yj − yj−1 ) .
a
Hence,
2τ τ
mÿj + yj − (yj+1 + yj−1 ) = 0, j = 1, · · · , N. (18.82)
a a
At the two endpoints the string is attached to the walls and not allowed
to move, so
y0 = yN +1 = 0. (18.83)
Normal Modes: The jth particle is a horizontal distance ja from
the origin. It is convenient to define xj = ja and to denote yj by y(xj ).
This is simply a change in notation but it is useful because the system
will oscillate and traveling plane waves can be set up in it. (Recall that
a stationary or standing wave can be considered as two traveling waves
moving in opposite directions.) A traveling plane wave is described
mathematically by
y(xj , t) = Aei(kxj −ωt) . (18.84)
638 18. SMALL OSCILLATIONS (OPTIONAL)

τ
τ

y
φ
θ
a

Figure 18.8. Forces acting on a point mass on a string.

The angles between the string and the horizontal at the
mass point are θ and φ.

Plugging Equation (18.84) into the equations of motion (18.82)

leads to the following expression for ω 2 :
4τ ka
ω2 = sin2 . (18.85)
ma 2
An expression such as this which expresses ω as a function of k, ω =
ω(k) is called a dispersion relation.
In Equation (18.84) and Equation (18.85) both ω and k could be
continuous, and in many applications they are. For the problem at
hand, however, the system is constrained by boundary conditions that
render ω and k discrete.
Equation (18.84) represents a plane wave traveling in the positive
x direction. A plane wave traveling in the opposite direction can be
represented by Be−i(kxj +ωt) and this expression also satisfies the equa-
tions of motion. Therefore, the general solution is the sum of these two
expressions:
y(xj , t) = Aei(kxj −ωt) + Be−i(kxj +ωt) . (18.86)
The boundary condition at xj = 0 is y(0, t) = 0. This can be met by
requiring that B = −A. The other endpoint is also stationary so
y ([N + 1]a, t) = 0.
Therefore,
0 = A eik(N +1)a e−iωt − e−ik(N +1)a e−iωt ,


= Ae−iωt eik(N +1)a − e−ik(N +1)a = Ae−iωt 2i sin (k(N + 1)a) .

 

So,
sin (k(N + 1)a) = 0,
and
k(N + 1)a = nπ, n = 1, 2, · · · , N. (18.87)
 18.7. MANY DEGREES OF FREEDOM 639

Consequently, from Equation (18.85),

 
2 4τ 2 1 nπ
ω = sin , n = 1, 2, · · · , N. (18.88)
ma 2N +1
The solution (18.86) for a particular ωn is
y(xj , t) = An e−iωn t eikxj − e−ikxj ,
 

= An e−iωn t 2i sin(kxj ),
 
nπxj
= 2iAn sin [cos ωn t − i sin ωn t] .
a(N + 1)
Since y(xj , t) is real, we only keep the real part of this solution. As-
suming An is real,
 
nπxj
y(xj , t) = 2An sin sin ωn t, (18.89)
a(N + 1)
where ωn2 is given by Equation (18.88).

Exercise 18.7. Show that ω 2 = (4τ /ma) sin2 (ka/2) follows from
Equations (18.82) and (18.84).

Figure 18.9. Positions of the three point masses.

640 18. SMALL OSCILLATIONS (OPTIONAL)

Worked Example 18.2. Consider a massless string with three

particles on it. Determine the normal mode frequencies (eigenfre-
quencies), the allowed wavelengths, and the maximum frequency.
Solution: For this system, N = 3. The three values of ωn are
obtained from Equation (18.88) and are:
r
τ
ω1 = 0.765 ,
ma
r
τ
ω2 = 1.414 ,
ma
r
τ
ω3 = 1.848 .
ma
These are the normal mode frequencies. According to Equation
(18.89), the amplitudes of the motion for the various normal modes
are functions of position given by 2An sin (nπxj /4a) . Plotting these
amplitudes gives a good qualitative feeling for the motion of the
system. See Figure 18.9.
It is useful to introduce a “characteristic velocity” by
r
τ
c= . (18.90)
m/a
In terms of c, the eigenfrequencies are
2c nπ
ωn = sin . (18.91)
a 2(N + 1)
By definition, the wave number is
2π
k= , (18.92)
λ
where λ is the wavelength. But k = nπ/ [2(N + 1)a] , so
2(N + 1)a
λ= .
n
If the length of the string is l then l = (N + 1)a and
2l
λ= .
n
The largest value of n is N , hence the shortest wavelength is
2l 2(N + 1)a
λmin = = & 2a.
N N
For large N the minimum wavelength is twice the spacing between
the particles, which is an expected conclusion. However, this also
 18.8. TRANSITION TO CONTINUOUS SYSTEMS 641

means there is a maximum frequency, which may not be quite as

intuitively obvious. It is
2c N πa 2c π 2c
ωmax = sin ≈ sin = .
a (N + 1)2a a 2 a

18.8. Transition to Continuous Systems

The previous section treated systems composed of discrete parti-
cles of mass m separated by distances a. If the interparticle spacing
is very small compared to the characteristic scale of the phenomenon,
the system can be considered continuous. In this section, a → 0 and
the system of the previous section more and more closely resembles a
continuous string. We first describe how the analysis presented above
is carried to the continuous limit. Then we consider the motion of a
string, first from an elementary point of view, then from an advanced
point of view using the Lagrangian approach.
The Continuum Limit. To go from the discrete particle case to
the continuous case, let N → ∞ and a → 0, but still require that the
length of the string remain unchanged. Then,
l = (N + 1)a = constant.
Furthermore, assume that the linear mass density, σ, is constant. That
is,
m
σ= = constant. (18.93)
a
The normal mode frequencies can be written as
 2
4c2  nπ 2
 
2 4c 2 1 nπ 2 nπ
 2
ωn = lim sin = = c .
N →∞ a2 2N +1 a2 2N l
(18.94)
(N → ∞ implies that any normal mode under consideration will have
n  N.)
The solutions are, of course, of the form
y(x, t) = Aei(kx−ωt) ,
where k = 2πn/l and ω is any one of the normal frequencies ωn .
A Continuous String. In Chapter 17 the wave equation was
derived from first priniciples. For a string of mass density σ under
tension τ we found (Equation 17.1):
∂ 2y 2
2∂ y
= c , (18.95)
∂t2 ∂x2
642 18. SMALL OSCILLATIONS (OPTIONAL)

where c2 = τ /σ. Note that I am using σ for the mass density of the
string because in this chapter ρ is being used to represent the eigen-
vectors.
Worked Example 18.3. Solve the wave equation (18.95) us-
ing the techniques developed in this chapter. (Recall that in Chap-
ter 17 the wave equation was solved by separation of variables.)
Solution: Assume normal mode solutions of the form
y(x, t) = Cρ(x) cos(ωt + φ). (18.96)
Plugging this expression into the differential equation (18.95) yields
d2 ρ(x)
+ k 2 ρ(x) = 0, (18.97)
dx2
where k = ω/c. Equation (18.97) is an ordinary differential equa-
tion that is also an eigenvalue equation. Solutions exist only for
specified allowed values of k 2 . (The solutions are the eigenfunc-
tions.) Since Equation (18.97) has the form of the SHM equation,
we can write a solution immediately:
r
2
ρ(x) = sin kx, (18.98)
lσ
p
where the coefficient 2/lσ was selected for normalization.
For a string tied at both ends, the boundary conditions are
ρ(0) = 0 and ρ(l) = 0. The solution (18.98) meets the first bound-
ary condition, but the second condition is met iff sin kl = 0, i.e., iff
k is one of the set
nπ
kn = , n = 1, 2, · · · , ∞. (18.99)
l
Since k = ω/c, the normal mode frequencies are
nπc
ωn = kn c = .
l
The allowed wavelengths are
2π 2l
λ= = , n = 1, 2, · · · , ∞. (18.100)
k n
Thus, the string must contain an integer number of half wave-
lengths. The general solution is obtained as a superposition of
normal modes and is given by
X∞
y(x, t) = Cn ρ(n) (x) cos(ωn t + φ), (18.101)
n=1
 18.8. TRANSITION TO CONTINUOUS SYSTEMS 643

or, equivalently,
∞ r
X 2
y(x, t) = sin kn x (an cos ωn t + bn sin ωn t) . (18.102)
n=1
lσ
Here an = Cn cos φn and bn = −Cn sin φn . Each term in this sum
satisfies the wave equation and the boundary conditions. The val-
ues of the constants can be determined from the initial conditions:
∞ r
X 2
y(x, 0) ≡ an sin kn x,
n=1
lσ
and ∞ r
X 2
ẏ(x, 0) ≡ ω n bn sin kn x.
n=1
lσ
The expressions in parentheses are Fourier series so you can use
the orthonormality property of these series to write
r Z l
2σ
an = y(x, 0) sin (kn x) dx, (18.103)
l 0
and r Z l
2σ
ωn bn = ẏ(x, 0) sin (kn x) dx. (18.104)
l 0

Exercise 18.8. Using condition 18.100, plot the first four waves in
a string with fixed endpoints.

18.8.1. The Lagrangian for a Continuous String. Consider

an element of the string, such as that illustrated in Figure 17.2. To
determine its Lagrangian evaluate the kinetic and potential energies of
the element and then integrate over the entire string.
The kinetic energy of an element of length dx is
 2
1 ∂y
dT = (σdx) ,
2 ∂t
and for the entire string,
Z l 2
1 ∂y(x, t)
T = σdx. (18.105)
2 0 ∂t
The potential energy is equal to the work required to stretch the
element from its original length dx to its final length ds. If τ is the
644 18. SMALL OSCILLATIONS (OPTIONAL)

tension in the string, this work is given by

h i 1  ∂y 2
2 2 21
dW = τ (ds − dx) = τ (dx + dy ) − dx = τ dx,
2 ∂x
and for the entire string,
2
1 l
Z 
∂y(x, t)
V = τ dx. (18.106)
2 0 ∂x
The Lagrangian follows immediately from L = T − V. If σ and τ are
constant,
2 2
σ l ∂y(x, t) τ l ∂y(x, t)
Z  Z 
L= dx − dx. (18.107)
2 0 ∂t 2 0 ∂x
Hamilton’s Principle for a Continuous System. Recall that
Hamilton’s principle states that the variation of the action integral is
zero: Z t2
δ Ldt = 0. (18.108)
t1
For a continuous one-dimensional system of length l, the Lagrangian
density L is defined by
Z l
L= Ldx. (18.109)
0
Then Hamilton’s principle is
Z t2 Z l  
∂y ∂y
δ dt dxL y, , ; x, t = 0. (18.110)
t1 0 ∂x ∂t
Now consider a virtual displacement,
y(x, t) → y(x, t) + δy(x, t),
subject to the condition of fixed endpoints in time, i.e.,
δy(x, t1 ) = δy(x, t2 ) = 0, for all x.
This describes a system that has the same configuration at times t1
and t2 , such as a string oscillating with period t2 − t1 .
The string is also subject to the spatial boundary condition of fixed
endpoints, so
δy(x = 0, t) = δy(x = l, t) = 0 for all t.
Equation (18.110) can be written as
Z t2 (Z l "    #)
∂L ∂L ∂y ∂L ∂y
0= dt dx δy + ∂y

δ + ∂y

δ .
t1 0 ∂y ∂ ∂x ∂x ∂ ∂t ∂t
(18.111)
 18.9. SUMMARY 645

We can write  
∂y ∂
δ = δy,
∂x ∂x
and  
∂u ∂
δ = δu.
∂t ∂t
Carrying out integrations by parts as usual, leads to
Z t2 (Z l " # )
∂L ∂ ∂L ∂ ∂L
0= dt dx − ∂y

−
δy .
t1 0 ∂y ∂x ∂ ∂x ∂t ∂ ∂y∂t

Consequently, Lagrange’s equations for a continuous system are

∂ ∂L ∂ ∂L ∂L
∂y

+ ∂y

− = 0. (18.112)
∂t ∂ ∂t ∂x ∂ ∂x ∂y

Worked Example 18.4. Derive the wave equation for a wave

on a string using the Lagrange density and Equation (18.112)
Solution: For a string, from Equation(18.107).
 2  2
σ ∂y τ ∂y
L= − , (18.113)
2 ∂t 2 ∂x
so
∂L ∂y
∂y

=σ ,
∂ ∂t ∂t
and
∂L ∂y
∂y

= −τ .
∂ ∂x ∂x
Lagrange’s equations (18.112) reduce to
   
∂ ∂y ∂ ∂y
σ + −τ − 0 = 0,
∂t ∂t ∂x ∂x
p
or, using c = τ /σ
∂ 2y 2
2∂ y
=c .
∂t2 ∂x2
Thus, we obtain the one-dimensional wave equation.

18.9. Summary
Interacting particles often exhibit small oscillations about equilib-
rium. Two classical examples of such systems are illustrated in Figures
18.1 and 18.2. The problem could be set up in terms of Cartesian
coordinates (xi ), but it is much more convenient to introduce a set
646 18. SMALL OSCILLATIONS (OPTIONAL)

of generalized coordinates qi defined as qi = qi0 + ηi where qi0 is the

equilibrium position of the ith particle and ηi is its instantaneous dis-
placement from equilibrium. We shall refer to the ηi as the “physical”
coordinates.
The kinetic energy can be expressed in terms of the physical coor-
dinates as
n n n n
1 XX 1 XX
T = mjk q̇j q̇k = mjk η̇j η̇k .
2 j=1 k=1 2 j=1 k=1
The potential energy is given by
n n  n n
∂ 2V

1 XX 1 XX
V = ηj ηk = vjk ηj ηk .
2 j=1 k=1 ∂qj ∂qk q0 2 j=1 k=1
The elements mjk and vjk define the mass matrix and the potential
matrix. They are given by
X  ∂xi   ∂xi 
mjk = mi ,
i
∂qj q0 ∂qk q0
and
∂ 2V
 
vjk = .
∂qj ∂qk q0
Knowing the kinetic energy and the potential energy allows one to
determine the Lagrangian and hence the equations of motion. These
are X
(mkj η̈j + vkj ηj ) = 0, k = 1, · · · , n.
j
To solve the equations of motion it is convenient to introduce a set of
complex quantities zj related to the ηj by
ηj = Re (zj ) .
The equations of motion are, then
X
(mkj z̈j + vkj zj ) = 0, k = 1, · · · , n.
j

The solution is obtained by making the usual assumption that all the
particles oscillate at the same frequency ω (normal mode assumption)
so that
zj = zk0 eiωt .
Plugging this into the equations of motion leads to a set of n coupled
algebraic equations that have a nontrivial solution iff
 
det vkj − ω 2 mkj  = 0.
 18.9. SUMMARY 647

The normal mode frequencies ωs can then be determined. Once the fre-
quencies are determined, the complex coordinates zk can be evaluated
using Cramer’s rule. The coefficients in the Cramer’s rule procedure
are
akj = vkj − ωs2 mkj ,
(s)
and the solution gives n values of zk for each frequency ωs . Here the
superscript (s) indicates the particular normal mode. (Actually, we
(s) (s)
obtain n − 1 ratios of the form zk /zn .)
(s)
The complex coordinates zk can be expressed in polar form, thus,
(s) (s)
zk = ρk eiφs .
(s)
There is a different ρk for each value of k, but the φs are all the same
(for a given s). The set of ρ values for a given normal mode (s) is
conveniently expressed as a column vector, thus,
 (s) 
ρ1
 ρ(s) 
ρ(s) =  . 2  .
 
 .. 
(s)
ρn
This is called the eigenvector for the normal mode. The various eigen-
vectors are orthogonal. Once the eigenvectors have been determined,
the physical coordinates can be obtained from
n
X (s)
ηj (t) = C (s) ρj cos(ωs t + φs ),
s=1

where C (s) and φs are obtained from the initial conditions.

The expressions obtained are probably better expressed as matrix
relations. The mass matrix M has elements mkj and the potential
matrix V has elements vkj. Assuming the normal mode frequencies have
been determined, the eigenvectors ρ(s) can be evaluated from
V − ωs2 M ρ(s) = 0.

The physical coordinates can also be expressed as a column vector,

X
η(t) = ρ(s) C (s) cos(ωs t + φs ).
s

The modal matrix A is the matrix generated by using the eigen-

vectors ρ(s) as columns.
648 18. SMALL OSCILLATIONS (OPTIONAL)

It is convenient to introduce a set of coordinates called “normal

coordinates” and denoted by ζ. These are defined by
η(t) = Aζ(t).
Consequently,
ζ(t) = AT Mη(t).
These normal coordinates are useful because they decouple the equa-
tions of motion leading to the set of equations
ζ̈s = −ωs2 ζs , s = 1, · · · , n.
The solutions can be written immediately as
ζs = C (s) cos(ωs t + φs ), s = 1, · · · , n.
Examples of the application of this theory are coupled pendulums
and transverse oscillations of mass points on a massless string. This
last example can be generalized to a continuous string.
The final topic was to determine expressions for the Lagrangian of
a continuous system, such as a string:
Z l 2 2
σ l ∂y(x, t) τ l ∂y(x, t)
Z  Z 
L= Ldx = dx − dx,
0 2 0 ∂t 2 0 ∂x
where σ is the linear mass density, and τ is the tension in the string.

18.10. Problems
Problem 18.1. Determine the mass matrix and the potential ma-
trix for a system of two equal masses and three equal springs arranged
as in Figure 18.1.
Problem 18.2. A particle of mass m finds itself in a region of space
where the potential is given by
V0
V (x, y, z) = 2 4x2 + 5y 2 + 6z 2 − 2ax − 5ay .


a
V0 and a are positive constants. (a) Determine the location and value
of V at its minimum. (b) Evaluate the mass matrix and the potential
matrix. (c) Determine the normal mode oscillation frequencies about
the equilibrium point.
Problem 18.3. A massless spring of constant k is fixed to the
ceiling and is hanging vertically. A particle of mass m is attached to
the bottom of the spring. Then another identical spring is attached
to the particle and finally a second particle (also having mass m) is
attached to the second spring. The system only moves in the vertical
 18.10. PROBLEMS 649

direction. Determine the normal frequencies and the eigenvectors ρ(1)

and ρ(2) .
Problem 18.4. For the problem of two coupled pendulums de-
scribed in Section 18.6, show that AT MA = 1 and AT VA = ωD 2
. Also
show that the Lagrangian L = 12 2j=1 (ζ̇j2 − ωj2 ζj2 ) reduces to the form
P

of Equation (18.12).
Problem 18.5. Three equal masses m are connected by four equal
springs of force constant k and length a, as in the system of Figure
18.1. Somehow the masses are constrained to oscillate transversely,
that is, perpendicular to the line of the springs. (Perhaps the masses
are mounted on thin rods.) The oscillations all take place in a single
plane. Find the normal mode frequencies.
Problem 18.6. Three equal springs are placed in a frictionless
circular trough of radius a. At the junctions of the springs are masses
m, m, and 2m. Determine the normal coordinates and frequencies of
oscillation.
Problem 18.7. Assume the system of Figure 18.1 consists of two
identical masses and three identical springs. (a) Find the Lagrangian
and Lagrange’s equations. (b) Determine the normal mode frequencies
and the eigenvectors. (c) Construct the modalmatrix. Answers: (b)
1 1 1
ω12 = k/m, ω32 = 3k/m. (c) A = √2m ..
1 −1
Problem 18.8. Consider the system of Figure 18.1 as described in
the previous problem Let one mass be displaced from equilibrium by a
small distance a in the direction of the other mass. Obtain the motion.
Problem 18.9. A system like the one in Figure 18.1 consists of
two equal masses and three equal springs. The system is initially at
rest. Then the mass on the right is subjected to an additional force
F = A sin ωt. Determine the motion.
Problem 18.10. Consider a linearly symmetrical triatomic mole-
cule. In equilibrium two atoms of mass m are located on either side
of an atom of mass M. All three atoms lie along a straight line and
in equilibrium the distance between them is b. The interatomic forces
can be assumed to obey Hooke’s law. Assume vibrations take place
along the line of the molecule. Determine the normal mode frequen-
cies. Describe the normal modes. (Note, the zero frequency normal
mode corresponds to a translation of the entire molecule.)
650 18. SMALL OSCILLATIONS (OPTIONAL)

Problem 18.11. Consider again the linearly symmetrical triatomic

molecule described in the previous problem. Show that it can be re-
duced to a problem with only two degrees of freedom by using the
distances between the atoms as coordinates. Write the 2 × 2 mass and
potential matrices and obtain the normal mode frequencies.
Problem 18.12. A system such as that illustrated in Figure 18.1
consists of N identical particles with masses m and N + 1 identical
massless springs with force constant k. In equilibrium each spring is
stretched so that its length (a) is greater than the unstretched length
(a0 ). Note that oscillations in the longitudinal mode and the transverse
mode are allowed, but assume all oscillations take place in a single
plane. (a) Determine the Lagrangian for the system. (b) By evaluating
the equations of motion, show that the longitudinal and transverse
modes are decoupled.
Problem 18.13. A large number of pendulums of length l and
mass m are coupled by springs of constant k. When the system is in
static equilibrium, all the pendulums hang vertically and the distance
between the bobs (a) is equal to the equilibrium length of the springs.
(a) Write the Lagrangian for the system in terms of ηi , the displacement
from equilibrium of the ith pendulum. (b) Obtain the equations of
motion. (c) Assume a traveling wave solution and obtain an expression
for ω.
Problem 18.14. A rod of length l and mass m is suspended from
its ends by two springs of equilibrium length b and constant k. The
springs are hanging from the ceiling and are suspended from two points
a distance L apart. Since L > l, the springs are not vertical, but
form an angle θ with the vertical. Assume the rod oscillates only in
the vertical plane formed by rod and springs. Determine the normal
modes, assuming small oscillations.
Problem 18.15. Use the orthogonality of the Fourier series to
determine the expressions given in Equations (18.103) and (18.104).
COMPUTATIONAL PROJECTS
Computational Project 18.1. Plot Equation (18.80) using rea-
sonable values for the parameters. Study the effect of varying the pa-
rameters.
Computational Project 18.2. Assume the system of Figure
18.1 consists of 10 equal masses (and 11 springs). The masses are all
0.1 kg and the springs all have k = 5 N/m. Write a computer program
to determine the normal mode frequencies.
 Part 5

Special Topics
Chapter 19
The Special Theory of Relativity

This chapter presents some of the basic ideas of the special theory of
relativity. Relativity theory involves serious modifications to classical
mechanics as well as to many of our basic notions about time, space
and causality. For this reason, relativity theory is of as much interest
to philosophers as it is to physicists.

large
ClassicalMechanics RelativityTheory

Size
Relativistic
Quantum
Quantum Mechanics
small Mechanics
slow fast
Speed
Figure 19.1. Different realms of physics. Classical me-
chanics applies to objects that are “large” and “slow.”

Relativity often contradicts “common sense” concepts, as does quan-

tum mechanics, the other great modification of classical physics. It may
help you to think of classical mechanics as the study of objects that
surround us in our familiar world. Balls, and springs and falling ob-
jects all obey classical mechanics. When things are moving very, very
fast, close to the speed of light, then there are modifications to classi-
cal physics. This is the realm of relativity. Also, when things are very,
very small (such as atoms and electrons) then there is a different set
653
654 19. THE SPECIAL THEORY OF RELATIVITY

of modifications to classical physics. This is the realm of quantum me-

chanics. Very small things that are moving very fast fall into the realm
of relativistic quantum mechanics. Figure 19.1 illustrates the realms of
these theories.

19.1. Albert Einstein (Optional Historical Note)

Albert Einstein was born in Ulm, Germany in 1879 and died in the
United States in 1955. When he was a high school student his fam-
ily moved to Italy, but he attended university in Zurich, Switzerland.
After graduation he found a position in the Swiss patent office, exam-
ining patent applications. During this time he developed the theory of
relativity. In 1905 he published a paper describing what is now known
as the “special” theory of relativity. The conclusions were not readily
accepted by the scientific community. During the same year he also
published important papers on the specific heat of solids and a paper
explaining the photoelectric effect. In 1915 he published the first paper
on “general” relativity. Special relativity concerns itself with reference
frames moving at constant velocities and is of particular importance
in electrodynamics. General relativity includes accelerated reference
frames and is a theory of gravitation.1
Einstein thus contributed to both electromagnetic theory and grav-
itation. His lifelong ambition was to develop a unified theory of gravi-
tation and electromagnetism, but this goal eluded him.
Although Einstein was one of the founding fathers of quantum me-
chanics there were many aspects of quantum theory that did not satisfy
him. In particular, he was bothered by the probabilistic interpretation
of quantum mechanics; he once declared, “God does not play dice!”
He also did not like to think that the state of a photon (for example)
could be determined by what happens to another photon a great dis-
tance away. He called this, “Spooky action at a distance.” However,
experimental evidence seems to indicate that Einstein was wrong in his
dismissal of this interpretation of quantum mechanics.
Einstein became a world famous personage, particularly after mea-
surements during an eclipse bore out the prediction of the general the-
ory of relativity that light from a star would be bent in the gravitational
field of the Sun. Newspapers declared that only five people in the world
could understand Einstein’s theory. Of course that was not true, but
it certainly added to his mystique. His opinions were sought out, even
on subjects far removed from physics.
1A
recent readable biography is by Walter Isaacson, Einstein, His Life and
Universe. Simon and Schuster, New York, N.Y., 2007.
 19.2. EXPERIMENTAL BACKGROUND 655

When the Nazi party came into power in Germany and began to
persecute Jews, Einstein left Berlin and emigrated to the United States
where he took a position at the Princeton Institute for Advanced Stud-
ies. He was convinced that German scientists would develop nuclear
weapons and he signed a letter to President Roosevelt urging him to
begin a project to develop the atomic bomb even though this went
against his pacifist philosophy. Eventually he took out American citi-
zenship. His last years were a model of a tranquil academic life. After
he died a medical examiner removed his brain. The story of Einstein’s
brain and its subsequent travels is a fascinating but gruesome tale.2

19.2. Experimental Background

In the latter part of the nineteenth century, physicists were con-
fronted with a paradoxical experimental fact. The work of Michelson
and Morley in the years 1881 to 1887 had proved conclusively that the
speed of light was the same in all reference frames. Whether the source
of the light was approaching or receding from the observer, the speed
of light was always the same, namely c = 3 × 108 m/s (186,000 miles
per second). Furthermore, regardless of whether the observer was ap-
proaching or receding from the source of the light, the speed of light
was still 3 × 108 m/s. For example, as the Earth moves in its orbit, at
one time of the year it will be approaching a distant star at a speed
of about 30 km/s. Six months later, Earth will be moving away from
the star at 30 km/s. Thus, one would expect the speed of light mea-
sured at those two times to differ by 60 km/s, a difference that could
be detected experimentally. However, when the measurement was per-
formed, it was found that there was no difference in the speed of light,
whether the Earth was approaching the star or moving away from it.
This experimental fact caused a great deal of concern among physi-
cists. If you are in a car and someone throws a rock at it, the speed of
the rock relative to the car depends on how fast the car is going. But
the Michelson Morley results showed that the speed of light relative to
the car does not depend at all on the speed of the car!
Understanding the phenomenon of the constancy of the speed of
light and finding an explanation for it, requires an analysis of measure-
ments made in two different reference frames. The reference frames are
quite simple - they are two inertial reference frames in relative motion.
Call the reference frames R and R∗ . Both are moving at constant ve-
locity. To make it easy to visualize the situation, you can imagine that
2Driving Mr. Albert: A Trip Across America with Einstein’s Brain, Michael
Paterniti, Dial Press, New York, 2001.
656 19. THE SPECIAL THEORY OF RELATIVITY

R is at rest with respect to the fixed stars and that R∗ is moving at

velocity v relative to R. Assume that at some initial time (t = 0) the
origins of the two coordinate systems coincide and that the velocity v
is directed along the common x-axes, as shown in Figure 19.2.
Consider the relationship between the primed and the unprimed
coordinates. Since these relationships are different in relativity theory
than in classical theory, physicists use the name “Galilean Transforma-
tion” to denote the classical relations. Note that in reference frame R a
particular event (such as the occurrence of a super nova, the ringing of
a bell, or any such well-localized event) is characterized by its position
and the time it happened. That is, an event is characterized by x, y, z, t.
This involves the concepts of space and time. For centuries, philoso-
phers have debated the meaning of space and time, asking, “What is
space? What is time?” Einstein decided to cut through all the compli-
cations and give simple, operational definitions for these two quantities.
Basically, Einstein said, “Space is what you measure with a ruler. Time
is what you measure with a clock.”

y y*
v
R R*
O x O* x*

Figure 19.2. Two inertial reference frames, R and R∗

The relative velocity v is directed along the common x-
axes.

An event that is located at x, y, z, t in R will be located at x∗ , y ∗ , z ∗ , t∗

in R∗ . What is the relation between these sets of coordinates? The
Galilean transformations between x, y, z, t and x∗ , y ∗ , z ∗ , t∗ are
x∗ = x − vt, (19.1)
y∗ = y,
z∗ = z,
t∗ = t.
where the rulers and clocks in both reference frames are identical in all
respects.
Although the transformations given by Equations (19.1) are very
satisfactory and work perfectly well for most problems in mechanics,
they cannot be correct because they do not give the same speed for
light in both reference frames. If x∗ is the position of an object in R∗
 19.3. THE POSTULATES OF SPECIAL RELATIVITY 657

at some particular time, then the x component of its velocity in R∗

is ẋ∗ , and the x component of its velocity in R is given by the time
derivative of the first of Equations (19.1) as
ẋ = ẋ∗ + v.
If the speed of light is measured in R∗ as c, then the speed of light as
measured in R will be c + v. But this is not the case. The speed of
light is c in both reference frames.
Einstein realized that the problem lay in the Galilean transforma-
tions. He derived a new set of transformations that reduced to the
Galilean transformations for objects moving slowly compared to the
speed of light but which yielded the same value for the speed of light in
both reference frames. These transformation equations are called the
“Lorentz” transformations because they had been derived earlier by
Lorentz. Einstein was not aware of Lorentz’s work and he derived the
equations himself independently. We shall derive the Lorentz transfor-
mation equations shortly, but first it is helpful to state the postulates
of special relativity.

19.3. The Postulates of Special Relativity

The easiest approach to special relativity is to accept Einstein’s two
postulates; that is, to accept as unproved assumptions the following two
statements:
Postulate 1: All inertial reference frames are equivalent.
Postulate 2: The speed of light is the same in all inertial reference
frames.
The first postulate, which is often referred to as the “Principle of
Relativity” implies that the laws of physics are the same in all inertial
reference frames. The first postulate also implies there is no experiment
that can demonstrate that a particular reference frame is at absolute
rest. In fact, the concept of “absolute rest” is meaningless.
The second postulate simply states the results of experiments such
as those of Michelson and Morley.
The two postulates are intimately related. Suppose you accept the
first postulate on the equality of all inertial reference frames. Then
the speed of light must be the same in all inertial reference frames
because otherwise you could distinguish between two reference frames
by simply turning on a lamp and measuring the speed of light. If the
speed of light turns out to be different, then the two reference frames
are not equivalent. Consequently, if the first postulate is true, then the
second postulate must also be true.
658 19. THE SPECIAL THEORY OF RELATIVITY

19.4. The Lorentz Transformations

Our goal is to obtain a set of transformation equations that will
reduce to Equations (19.1) under the conditions of classical mechanics
but will not contradict Postulate 2.
First, however, it is necessary to do away with preconceived notions
of time and space. If you have two identical clocks, you expect them
to run at the same rate. If you put one of the clocks in a rocket
ship moving at a constant velocity, you probably do not expect the
moving clock to run slow. Yet this is actually what you would observe.
For example, if John is in the “stationary” reference frame R and his
friend Mary is in the “moving” reference frame R∗ , John will claim that
Mary’s clock is running slow. Of course, from Mary’s point of view,
she is at rest and John is moving at a velocity −v, and she will say
that his clock is running slower than hers. Since John and Mary will
never meet again, the question as to whose clock is actually running
slow cannot be answered. (Later we will consider the “twin paradox”
in which one of the observers turns around and comes back to compare
clocks. However, if the two reference frames are at all times inertial,
they will never meet again.)
Furthermore, if John uses telescopes and other devices to measure
the length of a meter stick in Mary’s reference frame, he will decide
that her meter stick is shorter than his!
If all of this sounds like utter rot it may be because you believe
that space and time are absolutes and you have not yet accepted that
space and time are quantities that depend on the observer. If you
accept Einstein’s operational definition that time is what you measure
with a clock and you concede the possibility that two clocks in relative
motion might not run at the same rate, then you are close to accepting
relativity theory.3

19.4.1. The Light Clock: A Gedanken Experiment. Ein-

stein was fond of “gedanken” experiments, that is, “thought” experi-
ments, in which he applied the laws of physics to an imaginary situation
and determined the logical outcome.
A well-known gedanken experiment describes the operation of a
“light clock” which is made up of two mirrors. A pulse of light is
emitted at one mirror, bounces off the other mirror, and returns to
the first mirror. Some sort of detector lets us know when the light
3Iwill often refer to a reference frame “at rest” or “moving.” Perhaps each time
I use these words I should add, “with respect to the other reference frame,” but I
will let you supply that phrase mentally.
 19.4. THE LORENTZ TRANSFORMATIONS 659

returns to the first mirror, and the clock ticks. (Nobody said this was
a practical clock, but it certainly is a possible one.) The left panel of
Figure 19.3 illustrates the clock. Assume Mary has the clock in her
rocket ship (reference frame R∗ ). In this reference frame the clock is at
rest. The mirrors are separated by a distance d/2, so the time required
for the light pulse to go from A to B and back to A is d/c where c
is the speed of light. The time between ticks (according to Mary) is
∆tM = d/c.
B vt
B

d/2

A A A
l

Figure 19.3. A light clock at rest in reference frame

R∗ . The sketch on the right is the moving clock as seen
by an observer in reference frame R.

John, the observer in reference frame R, sees the clock as Mary’s

rocket speeds past him. From his point of view, the pulse of light that
goes from A to B to A and makes Mary’s clock tick is now traveling a
longer distance as illustrated in the sketch on the right in Figure 19.3.
During the time the light pulse goes from A to B and back to A, the
clock has moved to the right a distance 2l. The distance traveled by
the light ray, according to John, is 2s where s2 = l2 + (d/2)2 . John says
the time between ticks on Mary’s clock is 2s/c. Let us call this ∆tJ .
Note that s = c (∆tJ /2) . Also l = v (∆tJ /2) . Furthermore, as we have
seen, d = c∆tM . Therefore,
 
d ∆tM
=c .
2 2
Consequently, s2 = l2 + (d/2)2 can be written
 2  2  2
2 ∆tJ 2 ∆tJ 2 ∆tM
c =v +c .
2 2 2
Solving for ∆tJ we find
∆tM
∆tJ = p . (19.2)
1 − v 2 /c2
The denominator in this equation is less than unity so ∆tJ > ∆tM .
John claims that Mary’s clock is ticking more slowly than it would at
660 19. THE SPECIAL THEORY OF RELATIVITY

rest in his reference frame. He says her clock is running slow. This
effect is known as “time dilation.”
Remember that ∆tM is the time between ticks on a clock at rest
with respect to Mary and ∆tJ is the time between ticks on that same
clock as determined by John. Mary’s heart is a kind of clock. She says
her heart is beating at a normal 60 beats per minute. But John says
that Mary’s heart is beating more slowly, say, at 40 beats per minute.
He decides she is aging more slowly than he is.
The time measured on the clock at rest in R∗ is usually denoted t∗ .
Time intervals measured by the clocks in the two reference frames are
related by
p
t∗2 − t∗1 = (t2 − t1 ) 1 − v 2 /c2 . (19.3)
Mary is at rest with respect to the clock in the rocket ship. If she
observes the clock in John’s reference frame, she will say it is his clock
that is running slow. (After all, this is the theory of relativity!)
Having determined that moving clocks run slow (as observed from
a stationary frame), let us consider the length of a moving object as
observed from a stationary frame. Imagine Mary has a table in her
rocket ship and the table is aligned along the common x-axis of the
two systems. To determine the length of the table, she sets up a mirror
at one end and a lamp at the other. She (somehow) measures the time
for a light signal to go from the lamp to the mirror and back to the
lamp. This time is measured on her clock and is denoted ∆tM (or ∆t∗ ).
She concludes that the length of the table is l∗ = (1/2)c∆t∗ . Note that
the time of flight is ∆t∗ = ∆tM = 2l∗ /c.
John observes this experiment. He notes that the light ray that
left the lamp had to travel a distance l + v∆t1 to get to the mirror,
because the rocket moved the mirror a distance v∆t1 while the light
was traveling. The time for the light signal to get to the mirror is
1
∆t1 = (l + v∆t1 ),
c
and solving for ∆t1 we find
l
∆t1 = .
c−v
The light signal now must return to the position of the lamp, but this
time (according to John) it only has to travel a distance l − v∆t2 where
∆t2 is the time of flight. Once again
1
∆t2 = (l − v∆t2 ),
c
 19.4. THE LORENTZ TRANSFORMATIONS 661

and solving for ∆t2 we obtain

l
∆t2 = .
c+v
The total time for the light signal to propagate from lamp to mirror
and back to lamp, according to John, is
l l 2l/c
∆tJ = ∆t1 + ∆t2 = + = .
c−v c+v 1 − v 2 /c2
Finally, recalling that
∆tM
∆tJ = p ,
1 − v 2 /c2
we have
2l/c 2l∗ /c
= p .
1 − v 2 /c2 1 − v 2 /c2
Consequently,
p
l = l∗ 1 − v 2 /c2 . (19.4)
That is, the measured length of a ruler is longer in a reference frame
in which the ruler is at rest than in a frame in which it is moving.
This effect is called “length contraction.” It is sometimes convenient to
think of l∗ as the length of an object in a coordinate frame in which it
is at rest and call it l0 , the “rest length.” Then the length of the object,
as seen by an observer in another reference frame is
l = l0 (1 − v 2 /c2 )1/2 .

Exercise 19.1. The Space Shuttle is 60 meters in length. It flies

past you at 27,000 miles per hour (1.2 × 104 m/s). How much shorter
is it in your reference frame? Answer: 4.8 × 10−8 m.
Exercise 19.2. Mary’s rocket ship is moving at 0.92c relative to
John. Mary’s pulse rate, as measured by her, is 60 beats per minute.
She puts a stethoscope over her heart and transmits her heartbeats
over the radio to John. What is her pulse rate, according to John?
Answer: 23.5/min.
662 19. THE SPECIAL THEORY OF RELATIVITY

19.4.2. The Lorentz Transformation. Having determined the

relativistic properties of clocks and meter sticks, we can now obtain
the relations between positions and times in two coordinate systems in
motion with respect to one another. That is, we will write equations
relating (x∗ , y ∗ , z ∗ , t∗ ) to (x, y, z, t). Before beginning, it is important
to have a very clear idea what these coordinates refer to. Using again
the John and Mary example, imagine that some event (E) occurs in
Mary’s rocket ship. For example, a light bulb burns out with a sudden
pop. This event occurs at a well defined place and time. Mary can
measure the distance from her origin (O*) to the event and she can
determine the time the event took place by looking at her clock. She
characterizes E by the space-time coordinates (x∗ , y ∗ , z ∗ , t∗ ).
John also observes the event, and he characterizes it by specifying
the values of (x, y, z, t).
What is the relationship between the two sets of coordinates for the
event E?
Mary’s rocket ship is moving along the common x, x∗ coordinate
axes. Relativistic effects are due to the relative motion of the two
coordinate systems. Since there is no relative motion along the y or z
directions, it is clear that these components are unchanged. Hence the
coordinates of E in these directions are related by
y ∗ = y,
z ∗ = z.
Suppose that John and Mary’s clocks are at the origins of their coordi-
nate systems and suppose that they synchronized their clocks, setting
them to zero, at the instant the two origins coincided.
According to John, at the time of the event E the origin of Mary’s
coordinate system (O*) was at vt. According to Mary, at the time of
the event, the origin of John’s coordinate system was at −vt∗ .
Mary uses a meter stick to measure the distance from O*, her origin,
to E and she calls it l0 . John sees everything in the rocket ship as
contracted and he says the distance from O* to E is
p
l = l0 1 − v 2 /c2 .
Therefore, according to Mary, the p position of the event is x∗ = l0
and according to John it is x = vt + l0 1 − v 2 /c2 . Replacing l0 by x∗ ,
p
x = vt + x∗ 1 − v 2 /c2 .
Solving for x∗ ,
x − vt
x∗ = p . (19.5)
1 − v 2 /c2
 19.4. THE LORENTZ TRANSFORMATIONS 663

y y*
vt (in R)
-vt* (in R*)
v
t
t*
E
x x*
O*
O

Figure 19.4. Event E is observed from two coordinate

systems. The event is at (x, t) in the stationary system
and at (x∗ , t∗ ) in the moving system.

Our job is half done, as we have determined the relation between x and
x∗ . Now we need to obtain the relation between t and t∗ . This is quite
easily done by noting that Mary will obtain the same sort of transfor-
mation equation between the starred and unstarred x coordinates. She
finds that
x∗ + vt∗
x= p . (19.6)
1 − v 2 /c2
Plugging Equation (19.5) into (19.6) yields
" #
1 x − vt
x= p p + vt∗ .
2
1 − v /c2 2
1 − v /c 2

Solving for t∗ gives

t − (xv/c2 )
∗
t = p . (19.7)
1 − v 2 /c2
This completes the derivation of the Lorentz transformations. Gath-
ering the equations together:
x − vt
x∗ = p , (19.8)
1 − v 2 /c2
y ∗ = y,
z ∗ = z,
t − (xv/c2 )
t∗ = p .
1 − v 2 /c2
In relativity theory one often uses the shorthand expression,
1
γ=p ,
1 − v 2 /c2
664 19. THE SPECIAL THEORY OF RELATIVITY

so the transformation equations have the form

x∗ = γ(x − vt), (19.9)
 xv 
t∗ = γ t − 2 .
c
To obtain the “inverse” Lorentz relations, you can solve Equations
(19.9) for x and t, obtaining
x = γ(x∗ + vt∗ ), (19.10)
∗
 
xv
t = γ t∗ + 2 .
c
The Lorentz transformation equations yield the length contraction
and time dilation relations; this is no surprise, since they were based
on those concepts. However, these equations have to be applied with
care, as it is easy to generate a wrong answer. The following worked
examples may be helpful.

Worked Example 19.1. Mary measures the length of her

rocket ship and finds it to be l0 . (That is, x∗2 − x∗1 = l0 .) John,
who is at rest on the surface of a stationary asteroid, decides to
measure the length of the rocket as it speeds past him. He calls
this length l. Use the transformation equations to determine the
relation between l and l0 .
Solution: John must simultaneously determine the positions
of the two end points of the rocket, x1 and x2 . That is, the determi-
nations of the locations of the end points are two events that take
place at the same time. So t1 = t2 . Then the first of Equations
(19.8) yields
x2 − x 1
x∗2 − x∗1 = p ,
1 − v 2 /c2
or
x2 − x1
l0 = p ,
1 − v 2 /c2
leading to the usual length contraction relation,
p
l = l0 1 − v 2 /c2 .
(Note that the argument depended on our realizing that t1 = t2 .)

Worked Example 19.2. Now suppose that Mary looks at the

flashing red light on the tip of the wing of her rocket. She looks at
the clock in her rocket ship and measures the time interval between
 19.4. THE LORENTZ TRANSFORMATIONS 665

two flashes (or events) and calls it τ0 (= t∗2 − t∗1 ). John also sees the
two events. According to his clock the time between the events is
τ = t2 − t1 . What is the relation between the time measured by
Mary and the time measured by John?
Solution: Note that relative to Mary the two events occur at
the same place, so x∗2 = x∗1 . The time interval is calculated using
the inverse relation, the last equation of (19.10). The result is
t∗ − t∗1
t2 − t1 = p 2 .
1 − v 2 /c2
Consequently,
τ0
τ=p .
1 − v 2 /c2
Thus τ > τ0 . The time interval in the rest frame (τ0 ) is shorter
than the time interval in the reference frame that is moving with
respect to the events. Although John is at rest on the asteroid, in
this example, his is the moving clock and moving clocks run slow.
It is interesting to consider the time read on a distant clock by
a moving observer. Suppose that John has set out three synchro-
nized clocks at positions x = −l, 0, +l. Mary is passing by in her
rocket ship and at the instant she passes the origin of John’s refer-
ence frame, she checks the times on the clocks. (Actually, it might
take her a long time to make the determination, but she can al-
ways figure out later what the clocks read at the instant she passed
John’s origin.) In Problem 19.3 you are asked to show that accord-
ing to Mary, the three clocks will read times of +lv/c2 , 0, −lv/c2 .
Thus, the clock she is approaching reads an earlier time than the
clock at rest by −lv/c2 .

Exercise 19.3. Obtain Equation (19.6). Note the change in the

sign in the numerator.

Exercise 19.4. Carry out the steps to obtain Equation (19.7).

Exercise 19.5. Obtain the inverse Lorentz transformation equa-

tions (19.10).

Exercise 19.6. A person in a rocket ship with speed 0.92c relative

to Earth measures the time between two events to be five hours. Both
events happened at the same location on Earth. What is the interval
between these events as measured by an observer on Earth?
666 19. THE SPECIAL THEORY OF RELATIVITY

19.5. The Addition of Velocities

You know that the theory of relativity predicts that no material
object can be accelerated to a velocity greater than the speed of light.
(You will prove this contention in Problem 19.20.)
But if the speed of light is the upper limit - the maximum speed for
any material body - then we will have to drastically revise our idea of
how velocities add. Imagine the following scenario: Professor Einstein
(A) is at rest on the Earth. He is observing Captain Buck Rogers (B)
who is flying away from Earth in a rocket ship at a speed v = 0.75c.
Buck Rogers fires a torpedo (C) in the forward direction at a speed of
0.5c relative to his space ship. See Figure 19.5. The question is, how
fast is the torpedo going with respect to Earth? That is, what will
Professor Einstein say is the speed of the torpedo?

C
B
A

Figure 19.5. Assume A is at rest. B is moving at a

speed v = 0.75c relative to A, and C is moving at a speed
of 0.5c relative to B. What is the speed of C relative to
A?

Your first impulse may be to simply add the velocity of B (0.75c)

and the velocity of C (0.5c) and state that the speed of the torpedo
relative to the Earth is 1.25c. But this is obviously wrong since the
speed of the torpedo must be less than c in any reference frame. So
how should you add the velocities?
It turns out to be quite easy to derive the appropriate transforma-
tion law for velocities. Let the speed of the torpedo relative to the
rest frame be u and its speed relative to the moving frame be u∗ . The
speed of the rocket ship with respect to the rest frame is v, because it
is the speed of the moving frame with respect to the rest frame. For
convenience, assume that A, B, and C all coincide at time t = 0. The
problem consists in determining an expression for u in terms of u∗ .
The position of the torpedo with respect to the moving frame is
x∗ = u∗ t∗ .
 19.5. THE ADDITION OF VELOCITIES 667

Recall the transformation rules

x = γ(x∗ + vt∗ ),
t = γ(t∗ + vx∗ /c2 ).
Replace x∗ by u∗ t∗ and x by ut to obtain
u∗ + v
u= . (19.11)
1 + (u∗ v/c2 )
This is the relativistic “addition of velocities” law.
Worked Example 19.3. Refer to Figure 19.5. Buck Rogers
fires a torpedo at 0.7c at an angle of 60◦ to the direction of motion.
Determine the x and y components of the velocity of the torpedo
as measured by Professor Einstein.
Solution: Let us determine the addition of velocities law when
the object has a velocity component in the y direction. From Equa-
tion (19.11) the velocity in the x direction is
u∗x + v
ux = .
1 + (u∗x v/c2 )
The velocity in the y direction is obtained from the Lorentz trans-
formation equations that tell us that
y = y∗.
Write u∗y t∗ for y ∗ and uy t for y. That is,
uy t = u∗y t∗ ,
uy γ(t∗ + vx∗ /c2 ) = u∗y t∗ ,
u∗y t∗ u∗y t∗
uy = = .
γ(t∗ + vx∗ /c2 ) γ(t∗ + vu∗x t∗ /c2 )
So,
u∗y
uy = .
γ(1 + u∗x v/c2 )
For the problem at hand, v = 0.75c, u∗x = 0.70c cos 60◦ and u∗y =
0.70c sin 60◦ . Therefore,
0.70c cos 60◦ + 0.75c
ux =
1 + (0.70c cos 60◦ ) (0.75c) /c2
1.10c
= = 0.87c
1 + 0.263
668 19. THE SPECIAL THEORY OF RELATIVITY

and
q
u∗y 0.70c sin 60◦ 1 − (0.75c)2 /c2
uy = =
γ(1 + u∗x v/c2 ) (1 + (0.70c cos 60· ) (0.75c) /c2 )
0.10
= = 0.08c
1.263

Exercise 19.7. Carry out the algebra steps to obtain Equation

(19.11).
Exercise 19.8. Show that Equation (19.11) reduces to the ex-
pected relationships in the limits of velocities that are (a) much less
than the speed of light, and (b) equal to the speed of light.
Exercise 19.9. Obtain the inverse relationship u∗ = u∗ (u, v).
Exercise 19.10. A radioactive material is at rest in a physics lab-
oratory. Two electrons are ejected in opposite directions. Each has
a speed of 0.67c as measured in the laboratory frame. Determine the
speed of one electron relative to the other (a) classically and (b) rela-
tivistically. Answer (b): 0.92c.

19.6. Simultaneity and Causality

When two things happen at the same time, they are simultaneous.
But events that are simultaneous for one observer will not, in general,
be simultaneous for another observer.
For example, as shown in Figure 19.6, Mary turns on a light that
is located at the exact center of her rocket ship. The light reaches
the end walls (W1 and W2 ) of her ship simultaneously. But John, in
his stationary reference frame, sees the light reach wall W1 before it
reaches W2 , since the light only travels a distance L − vt to reach W1 ,
but it travels L + vt to reach W2 .
If the light reaching the walls defines events E1 and E2, then Mary
says that E1 and E2 are simultaneous, but John says E1 occured before
E2 . A third stationary observer, toward whom Mary is moving, would
say event E2 occurred before E1 .
If one event causes another event, then it must precede the second
event. In other words, causes must come before effects in any coordi-
nate system. In our example, event E1 can come before or after event
 19.6. SIMULTANEITY AND CAUSALITY 669

L L

v
Mary

John W1 W2

Figure 19.6. Mary turns on a light at the center of her

rocket ship. She says the light reaches the two end walls
simultaneously. But John sees the light reach wall W1
before it reaches wall W2 .

E2 , depending on the observer. Therefore, event E1 cannot in any way

cause E2 .
For E1 to cause E2 there must be some sort of a signal from E1 to
E2 . For example, E1 might be a bell that rings at W1 when the light
reaches it, and E2 might be a bell that is rung at W2 when a person
at W2 hears the bell from W1 . In this sense, you can imagine that E1
causes E2 . As you will see, when two events are thus causally related,
then E1 always precedes E2 in all coordinate systems.
To better appreciate this concept, let us note that according to
Mary, events E1 and E2 were simultaneous. They occurred at positions
x∗1 = −L, x∗2 = +L, and at times t∗1 = L/c and t∗2 = L/c.
According to John, assuming the light was switched on just as the
light bulb flew past him, the positions and times for the two events are
(x1 , t1 ) and (x2 , t2 ). These are related to (x∗1 , t∗1 ) and (x∗2 , t∗2 ) as given in
the following table:

Event = E1 Event = E2
x1 = γ(x∗1 + vt∗1 ) = γ(−L + vt∗1 ) x2 = γ(x∗2 + vt∗2 ) = γ(L + vt∗2 )
t1 = γ(t∗1 + v ∗ L
+ cv2 (−L) t2 = γ(t∗2 + cv2 x∗2 ) = γ Lc + cv2 L



x)
c2 1
=γ c

Note that, although t∗2 − t∗1 = 0,

 
L vL L vL vL
t2 − t1 = γ + 2 − + 2 = 2γ 2 .
c c c c c
Thus, according to John, E2 happened after E1 , as expected.
E1 and E2 cannot be causally related. In Mary’s coordinate system,
E1 and E2 occurred at the same time, so for E1 to cause E2 , a signal
from E1 to E2 would have to be propagated instantaneously, that is, at
infinite speed. From John’s point of view, the time between the events
was not zero, but rather 2γvL/c2 . The spatial separation of the two
670 19. THE SPECIAL THEORY OF RELATIVITY

events was
x2 − x1 = γ(L + vt∗2 ) − γ(−L + vt∗1 )
= γ (2L + v(t∗2 − t∗1 ))
= 2γL,
where the last step uses the fact that t∗2 = t∗1 . Therefore, if a signal were
to propagate from E1 to E2 in John’s reference frame it would have a
speed given by
distance 2γL c
speed = = = c > c.
time 2γLv/c2 v
Consequently, E1 cannot be the cause of E2 in any coordinate system.
A nice geometric way to understand the causality problem is based
on the concept that an event is a point in a four-dimensional space-time
coordinate system. An event is specified by three spatial coordinates
(x, y, z) and one temporal coordinate (t). I cannot draw (or even vi-
sualize) a four-dimensional space, but if, for example, z = 0 at all
times, then an event can be represented graphically as a point in the
three-dimensional projection of the four-dimensional space-time dia-
gram. See Figure 19.7.

t E(x,y,z,t)

y
x

Figure 19.7. The event E represented as a point in

four-dimensional space-time.

Consider two events, call them E0 and E1 . They are separated by a

space-time “distance” S defined by
S = (x1 − x0 )2 + (y1 − y0 )2 + (z1 − z0 )2 − c2 (t1 − t0 )2 . (19.12)
Suppose, for the moment, that event E0 is the turning on of a light
at the origin. The coordinates of E0 are (0, 0, 0, 0). Further, let E1 be
the reception of the light beam at some other point (x, y, z, t). The
distance traversed by the light is ct and this must be equal to the spa-
tial separation between the points. So for these two events, Equation
(19.12) leads to S = 0. In the z = 0 plane, the light from E0 spreads
out in a circle whose radius is (x2 + y 2 )1/2 = ct. The spreading of light
is represented by a family of circles, each one slightly larger than the
 19.7. THE TWIN PARADOX 671

previous one and slightly higher along the time axis. These circles gen-
erate a cone in the three-dimensional slice of space time, as illustrated
in Figure 19.8. Events that are causally related must lie within the
same light cone.

Figure 19.8. The light cone. This is a three dimen-

sional slice of the four-dimensional space time contin-
uum. The z-axis is not shown. Light emitted at time
zero at the origin will lie on the surface of the cone.

19.7. The Twin Paradox

One of the most famous “paradoxes” of special relativity is known
as the Twin Paradox. The paradox is this: One twin stays at home
(at rest) while his brother travels at a high speed to a distant place,
turns around, and comes home. The stay-at-home twin notes that his
brother’s clocks (heartbeat, etc.) have been “ticking” slowly because
he was traveling at a high velocity. The traveling twin will not have
aged as much as the stay-at-home twin, so when they meet again, the
traveling twin will be younger than his brother!
At first glance this may appear to be no paradox at all because spe-
cial relativity is concerned with inertial systems. Two systems moving
at constant relative velocity will only meet once, then separate to in-
finity, never to see each other again. Thus (at first glance) you might
be tempted to exclaim, “The problem is poorly posed! The traveling
twin turned around and therefore underwent accelerations. He was not
in an inertial system. All bets are off!”
However, upon further reflection, you will appreciate that the ac-
celerations of the traveling twin really do not change the nature of the
paradox. Suppose the traveling twin (let’s call him Flash) accelerates
for a short time then travels at a high speed (say 3c/5) for ten Earth
years. He then turns around and travels back to Earth at 3c/5 for
another ten years. Flash will have only aged 16 years (work it out!)
while his stay-at-home brother (called Homer) has aged 20 years.
672 19. THE SPECIAL THEORY OF RELATIVITY

Now suppose that instead of traveling for 10 Earth years at 3c/5,

Flash had traveled for 20 Earth years, then returned. In this case,
Homer aged 40 years but Flash only aged 32. You can appreciate that
the accelerations that Flash underwent were the same in both cases,
but the end result was different. Thus, it is not the accelerations that
causes the age differences, but rather the long term, high speed trip in
the inertial frame.
Let us analyze this twin paradox a bit more carefully. Homer and
Flash are at home and are the same age when Flash jumps onto a rocket
ship that happens to be heading out towards the star Pollux which is
35 light years away. That is, the distance to Pollux from Earth is
35c (if you want to express distances in meters, you have to express c
in meters/year). The outgoing rocket is an inertial system moving at
3c/5. Flash rides the rocket out to the star. He then jumps onto an
incoming rocket which just happens to be heading back to Earth at
3c/5. (Ignore the fact that Flash will be flat as a pancake after leaping
from one rocket to the other.) Homer says that his twin, Flash, was
gone for a total time of 117 years. (2 × 35c)/ [3c/5] . Homer is 117 years
older now. But Flash notes that once he jumped onto the outgoing
rocket, he saw that the distance to Pollux was only (4/5)(35) = 28
light years, and therefore it only took him 46.6 years to get there, and
another 46.6 years to return on the incoming rocket ship, for a total
of 93.2 years. So Flash is only 93 years older, while Homer aged 117
years.4
By the way, to get really dramatic results, you need to let Flash go
much faster, like at 0.95c.

Worked Example 19.4. A 30 year old man has a 10 year old

daughter. The man travels away and back home at (essentially)
constant speed. When he returns, both he and his daughter are 60
years old. How fast did the man travel?
Solution: In the rest frame of Earth the elapsed time is 50
years (t∗ = 50). In the moving reference frame the elapsed time is
30 years (t = 30).
distance l0
t∗ = 50 = = .
velocity v

4Some interesting aspects of the twin paradox are discussed in an article by S.

Wortel, S Malin and M. Semon, “Two examples of circular motion for introductory
courses in relativity,” American Journal of Physics, 75, 1123 (2007).
 19.8. MINKOWSKI SPACE-TIME DIAGRAMS 673

Also
p
distance l l0 1 − v 2 /c2
t = 30 = = =
velocity v v
p
50v 1 − v 2 /c2
30 = .
v
Consequently,
3 p
= 1 − v 2 /c2
5
and
4
v = c.
5

Exercise 19.11. Assume Flash travels to and back from Pollux at

0.98c. Determine how much he and Homer have aged. Answer: 71.42
yr and 14.7 yr.
Exercise 19.12. The “triplet paradox” imagines one triplet trav-
eling to the right at speed v, one traveling to the left at speed v, and
one staying at home. The traveling triplets turn around and come back
home. Will the traveling triplets be the same age or different ages when
they meet again?

19.8. Minkowski Space-Time Diagrams

A geometric way to treat relativistic problems in terms of “space-
time” diagrams was developed by Hermann Minkowski, who had been
one of Einstein’s professors. Many complicated relativistic problems,
especially those concerning simultaneity, become quite easy to analyze
by using these space-time diagrams because they reduce relativistic
situations to simple geometrical constructions.
Suppose observer A is at rest. Draw a two-dimensional rectangular
coordinate system with coordinate axes x and ct, as shown in Figure
19.9. Assume the only spatial coordinate of interest is x. Note that
the time coordinate has been multiplied by c so that it also represents
a distance. During a time interval from 0 to t, a light ray moves a
distance x = ct so light rays are always represented by lines at 45◦ to
both the x and t axes. The vertical line at x = −4 is the “world line”
for an object at rest in A’s frame, so that it is at the same position
at all times. The wiggly line starting at x = −2 is the world line for
674 19. THE SPECIAL THEORY OF RELATIVITY

an object that moves first in a negative direction, then in a positive

direction, and then in a negative direction again. The world line on
the right-hand side represents a ray of light propagating in the +x
direction.

ct

3
2
1

x
-4 -3 -2 -1 1 2

Figure 19.9. The space time diagram for observer A

(assumed at rest). Three “world lines” are shown: a
stationary object, a moving object, and (on the right) a
light ray.

Consider a similar space-time coordinate system for observer B who

is moving to the right with respect to A at a constant speed v. For
simplicity, assume B’s origin coincides with A’s origin when both of
their clocks read zero (t = t∗ = 0). Since B’s origin moves with respect
to A, it also has a world line that can be represented on A’s space-time
diagram. The world line of the origin of B is shown as a sloped line in
Figure 19.10a. Note that this world line has a greater slope than the
world line for the ray of light because, of course, B is moving at less
than the speed of light. The world line for B’s origin makes an angle θ
with the ct-axis, and it is easy to see that tan θ = v/c. You will also note
that this world line is labelled ct∗ because along this line, for any value
of time in B’s system, x∗ = 0. In Figure 19.10b the space-time diagram
for B is completed by adding the x∗ -axis at an angle θ = tan−1 (v/c)
with respect to the x-axis. I have also drawn in a set of oblique lines
to indicate that B’s space-time coordinate system is not rectangular.
I have not, however, placed any numerical values along these oblique
axes because we do not yet know the scale for the axes of the moving
coordinate system. Finally, Figure 19.10c shows the coordinate system
for observer C who is moving to the left with respect to A. However,
the A and B systems are sufficient for the present analysis.
The space-time diagrams (except for the scales for the B system)
are now set up and ready to be used in solving relativistic problems. A
good place to begin is with length contraction, because that not only
illustrates the use of the Minkowski diagrams, it also yields the correct
 19.8. MINKOWSKI SPACE-TIME DIAGRAMS 675

world line of
origin of B ct**
ct* ct ct*
ct ct

3 θ ray of light x*
2
1 x**
x x
1 2 3 (a) x (c)
(b)

Figure 19.10. Sketch (a) shows the world line for the
origin of observer B’s coordinate system. Sketch (b)
shows the oblique coordinate system for observer B and
sketch (c) shows the coordinate system of observer C who
is moving to the left.

way to put scales on the B system. In Figure 19.11 I drew a ruler of

length l0 (which you can consider to be one meter long) at rest in the
A system. The world lines of the two ends of the ruler in the A system
are simply vertical lines. The ruler at rest in A’s system is observed by
B, the moving observer. To determine the ruler’s length, B must see
both ends simultaneously, as shown in the figure. Thus, according to
B the ruler is an object whose world lines are the inclined lines in the
figure. This length is smaller than l0 .

ct*
ct

x*
B

A
x
l0

Figure 19.11. A ruler at rest in the A system as seen

by A. The world lines of the ends of the ruler are verti-
cal lines. These world lines as seen by B in the moving
system are shown as inclined lines.

Now if you were very attentive, you would have noticed something
strange. The ruler, as shown in the figure, is longer in B than in A,
as you can appreciate from the fact that in B it is the hypotenuse
of a triangle and in A it is along the adjacent side. You know that
676 19. THE SPECIAL THEORY OF RELATIVITY

from the moving system, the ruler must appear shorter. The problem
disappears, however, when you realize that we have not decided on the
scale of the various axes. (The ruler must be shorter according to B.
This means that the “tics” along the x∗ -axis must be further apart
than the tics on the x-axis. I’ll get to this point in a moment.) Just
to make things easier to analyze, in Figure 19.12 I moved the origin
of coordinates so that one end of the ruler coincides with the origin in
both systems.

ct*
ct

x*
b
tic*
y
Ob

O x x
f2l0 tic
l0

Figure 19.12. The ruler of length l0 in the A system

appears to have length Ob in the B system. A ruler of
length Ob at rest in the B system, appears to have length
f 2 l0 in the A system.

Figure 19.12 indicates three lengths. The first one is l0 , the length
of the ruler in the system A, which you can take to be one meter. The
next one is Ob, the length of the ruler as measured by B. It is shortened
by a factor f that has not yet been determined. Also we do not (yet)
know the scale for the starred axes. With all these limitations, you
might well ask how you could possibly determine the length Ob. Well,
there is a clever way around the problem, and that involves going back
to the A system that does have a scale. Imagine that B happens to have
a rod that is exactly the length Ob. If A observes this rod, it will appear
to have shrunk by another factor of f. So in the A system, this rod has
a length f 2 l0 . All of this fiddling around with rulers and going from
one inertial system to another, will allow you to determine the amount
f by which moving objects shrink, because now it is a straightforward
geometry problem. In the figure the angle θ = tan−1 (v/c) is indicated
in several places. You can see that
f 2 l0 = l0 − δx
 19.9. 4-VECTORS 677

and δx = δy tan θ. But δy/l0 = tan θ, so δx = l0 tan2 θ, and hence

f 2 = 1 − tan2 θ = 1 − (v 2 /c2 ).

Taking the square root, the length contraction factor is

r
v2
f = 1− 2,
c
as you knew all along. This is the graphical derivation of the Lorentz
contraction factor. Note that I was careful not to use any lengths in
the B system because there was no scale for it. Now, however, you
can determine the scale on the starred coordinate system. That is,
you can determine where to place the “tic marks” along the B system
axes, x∗ and t∗ . Begin by assuming that according to A the distance
l0 corresponds to one meter. Call it one A-meter. If a distance of
one meter (according to B) is marked off along the x∗ axis, it must be
longer than Ob becausep according to B the distance Ob is less than one
B-meter by a factor of 1 − (v/c)2 . So,
Ob
1B-meter = p .
1 − (v/c)2
From Figure 19.12,
Ob 1 p
= = 1 + (v 2 /c2 ).
1A-meter cos θ
Therefore,
p
1 + (v 2 /c2 )
1B-meter = p (1A-meter) .
1 − (v 2 /c2 )

p
Exercise 19.13. Show that cos θ = 1/ 1 + (v 2 /c2 ).

19.9. 4-Vectors
An event is specified by its position and time in a particular co-
ordinate system. It is convenient to introduce a four-dimensional co-
ordinate system in which an event is specified by the four coordinates
678 19. THE SPECIAL THEORY OF RELATIVITY

(x0 , x1 , x2 , x3 ) where
x0 = ct,
x1 = x,
x2 = y,
x3 = z.
(By multiplying the time coordinate by c, all coordinates are distances,
although x0 is obviously the coordinate that gives the time when the
event occurred.)
Just as three spatial coordinates describe an ordinary three-dimensional
vector, so too, the four space-time coordinates (x0 , x1 , x2 , x3 ) describe
a “4-vector” that starts at the origin and ends at some event E. A
4-vector from event E 1 to event E 2 has components (x20 − x10 ), (x21 −
x11 ), (x22 − x12 ), (x23 − x13 ). Note that the superscripts indicate the event.
Unfortunately, this makes for a somewhat confusing notation because
powers are also superscripts. For example, the 4-distance (or “space-
time interval”) between two events is given by
S = (x21 − x11 )2 + (x22 − x12 )2 + (x23 − x13 )2 − (x20 − x10 )2 .
Also note the change in sign on the temporal term. This 4-distance is,
of course, just the space-time distance defined previously by Equation
(19.12).
An important property of S is the fact that its value is unchanged
under a Lorentz transformation that expresses S in some other inertial
coordinate system. A quantity that is invariant under the transforma-
tion of a 3-D coordinate system is called a scalar, as you recall from
the discussion in Section 16.1. Analogously, S is called a four-scalar
(or sometimes, a “world scalar.”)
A more elegant way to write S uses the following “metric” (see
Section 5.3.4):
g0 = −1,
g1 = g2 = g3 = 1,
so that X
S= gµ (x2µ − x1µ )2 .
µ

You may recall from Chapter 16 that a 3-D vector is defined to be

a quantity whose components transform in the same way as the coordi-
nates under the rotation of the coordinate axes. Similarly, a 4-vector is
defined to be a quantity whose components transform according to the
Lorentz transformation (which is to say, the same as the coordinates
 19.9. 4-VECTORS 679

transform). The transformation of the coordinates from the unstarred

to the starred system according to the Lorentz transformation was pre-
sented in Equations (19.8) which I now repeat using a slightly different
notation:
x∗0 = γ(x0 − βx1 ), (19.13)
x∗1 = γ(x1 − βx0 ),
x∗2 = x2 ,
x∗3 = x3 ,
p
where β = v/c and γ = 1/ 1 − β 2 . These equations can be expressed
more succinctly by defining an array aµν as follows
 
γ −βγ 0 0
 −βγ γ 0 0 
aµν =   0
.
0 1 0 
0 0 0 1
Then X
x∗µ = aµν xν . (19.14)
ν
This equation describes the way the four coordinates transform. We
define a 4-vector to be a quantity that transforms according to Equation
(19.14). That is, the components of the 4-vector Aµ transform thus:
X
A∗µ = aµν Aν .
ν

It is traditional in dealing with relativity to use Greek letter subscripts

for 4-vectors. Also, you should note that 4-vectors are not written in
boldface the way 3-vectors are. We usually represent a 4-vector simply
as Aµ .
For example, the 4-vector displacement Dµ is the 4-vector originat-
ing at one event and terminating at another. The components of Dµ
in the unstarred (rest) system are (D0 , D1 , D2 , D3 ) and its components
in the starred system are
D0∗ = γ(D0 − βD1 ),
D1∗ = γ(D1 − βD0 ),
D2∗ = D2 ,
D3∗ = D3 .
The algebra of 4-vectors is straightforward. For example, if s is a 4-
scalar, then
Bµ = sAµ
680 19. THE SPECIAL THEORY OF RELATIVITY

will be a 4-vector if Aµ is a 4-vector. Similarly, if Aµ and Bµ are both

4-vectors, then the sum Aµ + Bµ is also a 4-vector. In many respects,
4-vectors behave like 3-vectors, as well they might, since for v = 0 the
components A1 , A2 , A3 of a 4-vector reduce to the three components of
an ordinary 3-vector.
The scalar product of two 4-vectors is defined as follows:5
X
(Aµ , Bµ ) = gµ Aµ Bµ = A1 B1 + A2 B2 + A3 B3 − A0 B0 . (19.15)
µ

On the other hand, there is no 4-vector analogue to the cross product.

Exercise 19.14. Prove that if Aµ and Bµ are both 4-vectors, then

the sum Aµ + Bµ is also a 4-vector.
Exercise 19.15. Show that (Aµ , Bµ ) is a 4-scalar, that is, its value
is not changed by a Lorentz transformation.

19.9.1. Space-like and Time-like Intervals. Consider two events,

E and E 2 and the related scalar S given by Equation (19.12). If S is
1

positive, the interval between the two events is called space-like. (In
this case each event is within the light cone of the other one.) For such
an interval, it is possible to define the real quantity “proper distance”
between the two events by
σ = S 1/2 .
This quantity is the distance between the two events in a coordinate
system in which the events are simultaneous.
If S is negative, a real “proper time” can be defined by
(−S)1/2
τ= .
c
Physically, this is the time interval between two events in a coordinate
system in which the two events occur at the same place. A simple
example is when a light in my rocket ship flashes twice. In my reference
frame the two events occur at the same location.

5A
purist would point out that the scalar product is the product of a contravari-
ant vector with a covariant vector. See Section 9.8.
 19.9. 4-VECTORS 681

Exercise 19.16. Show that the proper time is invariant under a

Lorentz transformation

19.9.2. The 4-vector Velocity. Once you have defined proper

time, you can define the 4-vector velocity. This is different than the 3-
vector velocity in some important respects. Consider a moving particle
that is at position x at time t and at x+dx at time t + dt. These two
events occur at well defined space-time coordinates xµ and xµ + dxµ
where dxµ is a 4-vector. The proper time between these two events is

dτ = [−(dxµ , dxµ )]1/2 /c.

The 4-velocity is defined by
dxµ
Uµ ≡ . (19.16)
dτ
Note that although the spatial part of the 4-velocity looks like an or-
dinary 3-velocity, it is really quite different. The 3-velocity of a moving
object is determined by measuring the distance it moved and dividing
by the elapsed time, where both distance and time are measured in the
same coordinate frame (usually assumed to be at rest). However, the
4-velocity is a “distance” measured in some reference frame divided by
the proper time, which is a quantity independent of the coordinate sys-
tem. The difference between them is brought out when one writes the
transformation equations for the 3-velocity components and compares
them with the transformation equations for the 4-velocity. These are:
3-velocity transformation equations:
ux − v
u∗x = ,
(1 − vux /c2 )
uy
u∗y = ,
γ(1 − vux /c2 )
uy
u∗z = .
γ(1 − vux /c2 )
4-velocity transformation equations:
U0∗ = γ(U0 − βU1 ),
U1∗ = γ(U1 − βU0 ),
U2∗ = U2 ,
U3∗ = U3 .
682 19. THE SPECIAL THEORY OF RELATIVITY

Worked Example 19.5. Express the components of the 4-

velocity in terms of the 3-velocity
Solution: Consider a clock at rest in R∗ . A time interval read
on this clock is the proper time denoted dτ. The time interval
according to an observer in R is dt. These times are related by
p
dτ = 1 − v 2 /c2 dt.
The 4-velocity is defined by
dxµ
Uµ = ,
dτ
so
dx1 vx dt vx
U1 = =p =p = γvx .
dτ 1 − v 2 /c2 dt 1 − v 2 /c2
Similarly
U2 = γvy ,
U3 = γvz .
Finally,
dx0 cdt c
U0 = =p =p = γc.
dτ 1 − v 2 /c2 dt 1 − v 2 /c2

Worked Example 19.6. A nuclear physicist has marked off

an x-axis and a y-axis on the lab bench and placed a radioactive
source at the origin. The source emits an alpha particle that moves
at a 30◦ angle to the x-axis with a speed of 2c/3. Determine the
four components of the 4-velocity.
Solution: The three ordinary velocity components (relative to
the laboratory) are
√ √
◦ 2c 3 3
u1 = ux = v cos 30 = = c,
3 2 3
2c 1 1
u2 = uy = v sin 30◦ = = c,
3 2 3
u3 = uz = 0.
The 4-velocity components are
Ui = γui , (i = 1, 2, 3)
U0 = γc.
 19.10. RELATIVISTIC DYNAMICS 683
p p √
Note that γ = 1/ 1 − (2c/3)2 /9 = 9/5 = 3/ 5. Therefore,
3
U0 = γc = √ c,
5
√ r
3 3 27
U1 = γu1 = √ c= c,
5 2 20
3 1 c
U2 = γu2 = √ c = √ ,
53 5
U3 = γu3 = 0.

Exercise 19.17. Show explicitly that the components of the 4-

velocity transform like a 4-vector. Start with the definition, Equa-
tion(19.16).

19.10. Relativistic Dynamics

It is interesting to consider two basic dynamical principles, namely
the conservation of momentum and the conservation of energy, using
the methods of special relativity.
Having defined the 4-velocity, it is straightforward to define the
4-momentum of a particle by
pµ = mUµ .
Here m is the mass of the particle measured in a system in which the
particle is at rest (m is the “rest mass” of the particle). Note that the
time and space parts of the momentum are
p0 = γmc,
pi = γmui , (i = 1, 2, 3).
For u  c, the space part of the 4-vector pµ reduces to the classical
momentum 3-vector. For a system of particles, the total 4-vector Pµ
has components X X
Pµ = pjµ = mj Ujµ ,
j j
where mj is the mass of the jth particle.
Now consider the conservation of momentum. Assume that a group
of particles can interact with each other but are isolated from any
other particles (no external forces act on the group of particles of inter-
est). Furthermore, the particles all maintain their identity (no particles
684 19. THE SPECIAL THEORY OF RELATIVITY

break apart nor do particles fuse together). The second condition will
be removed shortly.
Next, assume that the 4-momentum is conserved. (I will not prove
that the 4-momentum is conserved, but I assure you that this conser-
vation is confirmed by a vast body of experimental evidence and it can
also be derived theoretically.) Then the total 4-momentum is the same
at all times, i.e.,
(Pµ )t1 = (Pµ )t2 . (19.17)
What are the consequences of this statement? Equation (19.17) states
that the 3-momentum is conserved, and this is as expected, but it also
means that P0 is constant. That is,
X
γj mj c = constant,
j

where
−1/2
u2j

γj = 1 − 2 .
c
Expanding this expression, using the binomial expansion, yields
−1/2
X X u2j
P0 = γj mj c = c 1− 2 mj ,
j j
c
X 1 u2j 3 u4j

= c 1+ + + · · · mj ,
j
2 c2 8 c4
X 1X1
= mj c + mj u2j + · · ·
j
c 2
or !
X
cP0 = mj c2 + T + O(1/c2 ). (19.18)
j
P1
The last term is negligible. I wrote T for m u2 because that is
2 j j
the kinetic energy of the system of particles. Note that in relativity,
energy and momentum have become enmeshed in the single quantity 4-
momentum. Einstein identified the quantity cP0 with the total energy
E.
E = cP0 .
We can also express the total energy as E = γmc2 .
Writing Equation (19.18) for a single particle,
.
E = mc2 + T (19.19)
 19.10. RELATIVISTIC DYNAMICS 685

which states that the total energy of particle is given by its kinetic
energy plus a quantity mc2 . You might suspect that mc2 is a constant
because it is nearly always acceptable to add a constant to the energy
and only consider energy differences. Thus, you might feel that rel-
ativity theory has only resulted in adding a constant to the energy.
But if the total energy E is constant, this interpretation would mean
that T, the kinetic energy, is also constant, and obviously that is not
necessarily true. In the interaction between two particles (a collision),
the kinetic energy is conserved only if the interactions are perfectly
elastic. Otherwise, kinetic energy is lost, and according to the usual
description, energy has been converted into heat. According to rela-
tivity theory, the term mc2 represents all other forms of energy (heat,
potential energy, chemical energy, etc.). Equation (19.19) states that
the total energy E is equal to the kinetic energy T plus all other forms
of energy. Thus, if you heat an object (at rest), the value of mc2 must
increase. Recall that m is the rest mass. This means that the heated
object is more massive than the same object when cool.
The total energy of an object at rest (so that T = 0) is given by
the well known equation
E = mc2 .
For example, if you heat a gram of water by one degree centigrade,
you supply it with one calorie (4.186 joules) of heat. You will have
increased the mass of the water by 4.18/c2 = 4.64 × 10−17 kg. This is
unmeasurable. In like manner, if a nucleus splits apart (fission) and
the fragments are less massive than the original nucleus, then energy
is emitted. This energy is, as we all know, quite significant.
It is sometimes convenient to write the rest mass of a particle as
m0 . Then the momentum of a particle is
Pi = γm0 ui , (i = 1, 2, 3)
and the “relativistic mass” mrel is defined as
m0
mrel = γm0 = p ,
1 − v 2 /c2
where v is the velocity of the particle. This notation suggests that the
mass of a moving particle increases with its velocity and tends toward
infinity as the speed approaches c. You might consider the concept of
relativistic mass to be a needless variant, and, in fact, modern physi-
cists tend to dismiss it. Nevertheless, it is helpful in solving collision
problems involving relativistic particles, particularly since relativistic
mass is conserved in collisions.
686 19. THE SPECIAL THEORY OF RELATIVITY

An interesting and useful relation is obtained by dotting the four

vector momentum into itself. Now pµ = mUµ so the dot product of the
four velocity Uµ with itself is (according to Equation 19.15)
−c2 + u2
(pµ , pµ ) = −γ 2 m2 c2 + γ 2 m2 (u2x + u2y + u2z ) = m2 = −mc2 .
1 − u2 /c2
But we can also write
(pµ , pµ ) = −(p0 )2 + p · p = −E 2 /c2 + p2
so
E − p2 c2 = m2 c4 .

19.11. Summary
In this brief review of the theory of relativity, I only considered
aspects of the theory that relate to mechanics and did not delve at all
into electromagnetic phenomena. Nevertheless, you should be aware
that it was electromagnetic considerations that led Einstein to develop
his theory.
The most important concepts of the theory are embodied in the
Lorentz transformations, from which time dilation and length contrac-
tion follow. These transformations lead to a number of unusual effects,
including paradoxes such as the relative aging of moving twins. The
Lorentz transformations are
x − vt
x∗ = p ,
1 − v 2 /c2
y ∗ = y,
z ∗ = z,
t − (xv/c2 )
t∗ = p .
1 − v 2 /c2
Time dilation is described by
τ0
τ=p ,
1 − v 2 /c2
and length contraction is given by
p
l = l0 1 − v 2 /c2 .
The relativistic addition of velocities is given by
u∗ + v
u= ∗ .
1 + uc2v
 19.12. PROBLEMS 687

An event is represented by a point in the four dimensional space

time continuum and the separation of two events is
S = (x1 − x0 )2 + (y1 − y0 )2 + (z1 − z0 )2 − c2 (t1 − t0 )2 .
Any two events that are causally related must lie within the same light
cone.
The Minkowski diagrams introduce no new physics, but they help
to analyze relativistic effects. These lead naturally to the mathematical
definition of four-vectors which can be used to determine the famous
expression E = mc2 .
In relativity, velocity has four components and these transform ac-
cording to
U0∗ = γ(U0 − βU1 ),
U1∗ = γ(U1 − βU0 ),
U2∗ = U2 ,
U3∗ = U3 .
The fact that the relativistic expression for the space part of the four-
vector momentum is pi = γmui is often interpreted as the mass increase
of moving objects and one often sees the expression for the mass
m0
m= p .
1 − v 2 /c2
However, from the point of view of the theory, it is not the mass but
the momentum that is affected by the motion of the reference frame.

19.12. Problems
Problem 19.1. An observer in R measures an event at x = 3 m
and t = 7 ns. The R∗ frame is moving along the common x-axis at
v = 0.6c. (a) What are the relativistic space time coordinates measured
by the observer in R∗ ? (b) What space time coordinates would be
measured in R∗ if the Galilean transformation law held?
Problem 19.2. Use the Lorentz transformations to show that an
observer moving at speed v toward a clock at rest in some reference
frame, will find the clock to read a time −lv/c2 compared to the reading
by an observer in the rest frame of the clock. At the instant of the
measurements the two observers are at the same location and at a
(rest frame) distance l from the clock.
Problem 19.3. John, the observer in R, sets out three synchro-
nized clocks at locations x = −l, 0, +l. Mary (in frame R∗ traveling at
speed v along x) observes the times on the clocks at the instant she
688 19. THE SPECIAL THEORY OF RELATIVITY

passes the origin of John’s reference frame. Show that according to

Mary the clocks read +lv/c2 , 0, −lv/c2 .
Problem 19.4. An electron in a linear accelerator is given an en-
ergy of 50 MeV. It is moving through a tube of length 10 meters (ac-
cording to the laboratory reference frame). From the point of view of
the electron, what is the length of the tube?
Problem 19.5. An observer in R notes that an event occurred at
x = 3 × 108 m at time 3 s. Reference frame R∗ has a speed of 0.5c
in the positive x direction. (a) Determine the coordinates of the event
according to an observer in R∗ . (b) Determine the coordinates of the
event according to an observer in a reference frame moving at 0.5c in
the negative x direction. Answer: (b) 8.66 × 108 m, 4.04 s.
Problem 19.6. Consider two events, E1 and E2 . In R, the first
event is a light being turned on at some point along the x-axis. The
second event occurs one microsecond later when another light is turned
on a distance 1000 meters away. (a) There is an inertial reference frame
R∗ moving at speed v relative to R in which these two events are
simultaneous. Explain why this is possible. (b) What is v, the relative
speed of R∗ ? (c) What is the spatial separation of the two events in
R∗ ?
Problem 19.7. A pion is a particle that has a halflife of 1.8 × 10−8
seconds (in its rest frame). Suppose a beam of pions has been produced
by collisions between protons and some appropriate material, and it is
found that they have a speed of 0.99c. Measurements a distance 38
meters away show that half of the pions are present in the beam. (a)
Show that this result is not consistent according to classical physics.
(b) Show that time dilation will account for the measurement.
Problem 19.8. John and Mary both have rocket ships. In their
rest frames the ships have a length of 200 m. They pass one another,
heading in opposite directions. Mary measures the time for John’s ship
to pass her and gets a value of 4 µs. (a) What is the relative speed
of the two ships? (b) Mary has set up two clocks in her ship, one at
either end. Each is turned on when the nose of John’s ship is directly
opposite. What is the time difference between her clocks?
Problem 19.9. Show that the speed of a particle whose kinetic
energy is equal to its rest energy (m0 c2 ) is a constant independent of
its mass. Determine that speed.
 19.12. PROBLEMS 689

Problem 19.10. An electron is accelerated through a potential

difference of 5 × 106 V. Determine its kinetic energy and velocity in the
laboratory frame.
Problem 19.11. The clock in Mary’s rocket ship is moving along
the positive x-axis, which is marked off in meters. As it passes the
origin, the clock reads zero and when it passes the 200 m mark, it
reads 0.5 µs. How fast is the rocket ship moving?
Problem 19.12. Suppose you formulated the Lorentz transforma-
tion equations in terms of a new variable w defined by w ≡ ct. Then
show that the Lorentz equations and their inverses for the x and t
coordinates are given by:
x∗ = γ(x − βw), x = γ(x∗ + βw∗ ),
w∗ = γ(w − βx), w = γ(w∗ + βx∗ ),
where β = v/c. (Note that these transformation equations are more
symmetric than the original set.)
Problem 19.13. The Earth’s speed in its orbit around the Sun is
30 km/s. Determine the shortening of the Earth’s diameter as observed
from a reference frame at rest with respect to the Sun. Is it valid to
assume the Earth is moving in a straight line?
Problem 19.14. Some pions are created high in the Earth’s at-
mosphere by collisions between cosmic rays and the nuclei of nitrogen.
These pions have very high speeds. Assume v = 0.998c. A pion at rest
has a mean life of 26 ns. Assuming the pion moves directly downward,
determine how far toward the surface such a pion can travel before it
decays.
Problem 19.15. Princess Leila is standing at one end of a landing
field waiting for Luke Skywalker to fly over. She sees a light flash from
the far end of the field (it is 2.7 km away) and 10 µs later she sees a
flash from a light that is 200 m away. Luke also observes these flashes
but he sees them at the same location. (a) Use this information to
determine how fast Luke is traveling. (b) What is the time interval
between flashes in Luke’s reference frame? Answer: (a) 0.83c.
Problem 19.16. Show that the spacetime interval S associated
with two events is invariant under a Lorentz transformation. You may
assume the events occur on the common x-axis.
Problem 19.17. The theory of relativity correctly predicts the
aberration of light. (This phenomenon can be described as follows: If
690 19. THE SPECIAL THEORY OF RELATIVITY

the Earth were at rest and a particular star were directly overhead, you
would observe it by pointing your telescope straight up. But since the
Earth is moving, you must compensate by pointing your telescope at
a small angle away from straight up.) The aberration of light means
that a ray of light that makes an angle θ∗ with the x axis in the rest
frame will make an angle θ in the moving frame where
p
sin θ∗ 1 − v 2 /c2
tan θ = .
cos θ∗ + v/c
Consider a light source at rest in R∗ . The fraction of light emitted into
a cone of half angle θ∗ is
1
f = (1 − cos θ∗ ) .
2
In this problem, you will show that in a moving reference frame, the
same amount of light is emitted into a smaller cone of half angle θ. (a)
Derive the relationship f = 0.5 (1 − cos θ∗ ) . (b) Assume that θ∗ = 40◦
and determine f in a coordinate system in which the light source is
at rest. (c) Calculate the half angle of a cone in a frame moving at
v = 0.9c into which the same amount of light will be emitted. (d) Why
do you think this is called the “headlight” effect?
Problem 19.18. Evaluate the work done on an electron when it
is accelerated from rest to 0.98c. Answer: 3.3 × 1013 J.
Problem 19.19. Determine the speed and kinetic energy of a par-
ticle whose momentum is equal to m0 c.
Problem 19.20. Show that a particle that travels at the speed of
light must have zero rest mass.
Problem 19.21. A proton is traveling at 0.995c in the laboratory
frame. What is its kinetic energy? What is its momentum?
Problem 19.22. Two identical particles of rest mass m0 approach
one another at speeds ±0.5c, and undergo a completely inelastic col-
lision. What is the momentum and energy of the final (composite)
particle? Note that the momentum of a particle is given by p = γm0 v,
where γ = γ(v) depends on the speed of that particular particle.)
Problem 19.23. In some particular reference frame (perhaps the
laboratory), a particle traveling at 0.8c collides with and sticks to an
identical particle at rest. (a) Determine the momentum of the resulting
combined particle. (b) What is its speed? (c) What does classical
physics yield for the speed? (You may assume that when the particles
are at rest they have mass m0 .)
 19.12. PROBLEMS 691

Problem 19.24. A particle traveling at 0.8c (in the lab frame)

collides elastically with a particle at rest. The rest mass of the incoming
particle is three times greater than the rest mass of the second particle.
Determine the velocities of the two particles after the collision. (Hint:
Transform to a coordinate system in which the total momentum is
zero. Hint Number 2: The speed of the center of momentum system is
vc = PT c2 /E where PT and E are the total momentum and energy.)
Problem 19.25. A well insulated container maintained at 20◦ C is
placed on an impossibly accurate scale. You put a 50 gram ice cube at
0◦ C in the container and wait a long time until the system comes to
equilibrium at 20◦ C. What is change in reading of the scale?
Problem 19.26. A relativistic train has a speed of 4/5 c relative
to a worker standing outside, next to the tracks. A passenger inside
the train measures its length to be 200 meters. Suddenly, two lightning
bolts strike the train, one at the front end and the other at the rear.
The passenger says they struck the train at exactly the same instant.
When the lightning bolt hit the rear of the train, the rear of the train
was directly opposite the worker. How much time elapses before the
worker sees the lightning flash from the bolt that hit the front of the
train?
Problem 19.27. A wooden wedge in the shape of a right triangle
is on board a√rocket traveling at 4/5 c. The dimensions of the triangle
are 4, 5, and 41 meters in its rest frame. Determine the shape of the
triangle (calculate its dimensions) relative to a reference frame at rest.
What is the fractional change in its area?
Problem 19.28. Determine aµν for the general case in which the
starred axes are not parallel to the unstarred axes. Show
P ∗that a subse-
quent transformation to a third system yields a0λν = aλµ aµν .
Chapter 20
Classical Chaos (Optional)

This book has concentrated on “integrable” problems, that is, prob-

lems whose equations of motion can be integrated yielding closed form
solutions.
Some problems do not yield closed form solutions, nevertheless they
can be integrated numerically and are found to have well behaved solu-
tions. An example of this kind of problem is the motion of the Moon.
As you know, the Moon is affected by the gravitational force of the
Sun and the planets as well as its primary (the Earth). The resulting
equation of motion is very complicated and cannot be integrated by
analytical methods. Numerical solutions, however, have been carried
out. You will be relieved to know that the Moon will continue orbiting
the Earth for many millions of years.
In considering the motion of the Moon, the effects of the Sun and
planets are treated by the techniques of perturbation theory in which
the dominant motion is integrable and the weak interactions are treated
as perturbations. An example of this is found in Section 10.10 where
we considered the effect of a small perturbation on a previously circular
orbit. (The perturbation was a comet colliding with a planet.)
For most mechanical systems a small perturbation, or a slight change
in the initial conditions, will lead to a solution that is not very different
from the original solution. Such cases are called “regular” or “normal.”
However, there are some physical systems for which a very small change
in the initial conditions can lead to a dramatically different motion.
Such systems and their solutions are referred to as “chaotic.” Note,
however, that these systems are perfectly deterministic. If we start
the system again with exactly the same initial conditions, the system
will precisely retrace the same steps. The solutions, however, are not

693
694 20. CLASSICAL CHAOS (OPTIONAL)

predictable because they are highly sensitive to the initial conditions

which are never known with absolute accuracy.1
A completely random system, a system characterized by unpre-
dictable stochastic behavior is not an example of chaos, as considered
here. Classical chaos is a type of motion that has some degree of
regularity. The behavior of a completely random system cannot be
predicted, but a chaotic system is deterministic, and if you know the
state of the system at any given moment, you can predict its future
behavior.
If you are wondering why some systems are chaotic and others are
not, a glib answer is, “Chaos is a consequence of nonlinear equations
of motion.”

20.1. Configuration Space and Phase Space

The Lagrangian for a physical system is a function of the coordi-
nates and the velocities. For example, the Lagrangian for a mass on a
spring can be expressed in terms of x and ẋ as
1 1
L = T − V = mẋ2 − kx2 .
2 2
At any instant of time, the state of the system is described by the
values of x and ẋ at that time and on a plot of ẋ vs x, the state of the
system would be represented by a single point. As time goes on, the
values of x and ẋ will change and the point will trace out a path.
The Lagrangian for the double pendulum of Figure 4.2 could be
expressed as a function of θ1 , θ2 and θ̇1 , θ̇2 and the system of masses
and springs in Figure 11.10 would have a Lagrangian depending on the
coordinates and velocities, x1 , x2 and ẋ1 , ẋ2 .
The values of the positions (such as x1 and x2 ) at any instant of
time are said to describe the “configuration” of the system, and the plot
of x1 vs x2 is said to be a representation of the system in “configuration
space.”
Another way to describe a physical system is to give the Hamilton-
ian. You recall that the Hamiltonian is a function of momentum and
position. H = H(p, q). For example, the Hamiltonian for a mass on a
spring is
p2 k
H= + x2 .
2m 2
1A
nice description of chaotic motion is presented in Chapter 11 of “Classical
Mechanics, 3rd Edition” by H. Goldstein, C. Poole and J. Safko, Addison Wesley,
San Francisco, 2002.
 20.1. CONFIGURATION SPACE AND PHASE SPACE 695

Again, at any instant of time, the state of a system is given by the values
of p and x. For a one-dimensional system, these can be represented by
a point on a p vs x plot. As time goes on the values of p and x will
change and the point will trace out a path in the px plane. This plane
is called “phase space” and the path is called a phase space trajectory.
For most simple systems, the momentum and the velocity are related
by p = mẋ, so plots of ẋ vs x are also called phase space plots (or phase
space “portraits”).
Although chaotic motion is not periodic, it is convenient to begin
our study by considering some aspects of periodic motion. You are
familiar with two periodic systems: simple harmonic oscillators and
the Kepler (two-body) problem. The Kepler problem is interesting
because it involves two different periodicities, namely, the periodicity
in the radial motion and the periodicity in the angular motion.
Recall that the radial position of a particle in an elliptical orbit is
a (1 − e2 )
r= .
1 + e cos θ
The angular velocity is given by
l
θ̇ = .
mr2
Now the unperturbed two body system, as considered in Chapter 10
is degenerate, that is, the period of the radial motion is equal to the
period of the angular motion, and the orbit is repeated over and over
again. A configuration space plot of r vs θ is simply the elliptical orbit
of the particle.
The Hamiltonian for a Keplerian particle is a function of pr , pθ , r, θ.
It is interesting to generate phase space plots for this motion. Note,
again, that one normally plots velocity rather than momentum vs po-
sition, that is, one plots ṙ vs r and θ̇ vs θ. Figure 20.1 gives these plots
for various values of e, using arbitrary units.

Exercise 20.1. Letting k = 1 and m = 1, generate a phase space

plot for a mass on a spring for H = 1, 2, 3.
Exercise 20.2. What would the plots in Figure 20.1 look like for
e = 0?
696 20. CLASSICAL CHAOS (OPTIONAL)

Figure 20.1. Phase space diagram for Keplerian mo-

tion in arbitrary units. Top panel (ṙ vs r), bottom panel
(θ̇ vs θ).

20.2. Periodic Motion

Although the Hamiltonian was introduced in Chapter 4 we did not
really need it to solve the various dynamical problems that have been
considered. However, to study chaos, the Hamiltonian approach is
indispensable.
Recall that the Hamiltonian is a function of the momenta, the po-
sitions, and time, thus:

H = H(p1 , p2 , · · · , pn ; q1 , q2 , · · · qn ; t)

where n is the number of degrees of freedom. We shall only consider

Hamiltonians that do not depend on time.
Recall from Chapter 4 that the equations of motion in the Hamil-
tonian formulation are

. ∂H . ∂H
pi = − and qi = + .
∂qi ∂pi
 20.2. PERIODIC MOTION 697

(See Equations 4.20.) Recall, further, that under most usual conditions,
the Hamiltonian reduces to the total energy and one can write
H = T + V.
Let us consider, as an example of periodic motion, a double os-
cillator in which the motions in mutually perpendicular directions are
independent. The Hamiltonian for such a system can be written as
(p01 )2 1 2 (p0 )2 1 2
H= + m1 ω12 (q10 ) + 2 + m2 ω22 (q20 ) (20.1)
2m1 2 2m2 2
which is just E1 + E2 = total energy. It is convenient to transform to
“normalized” coordinates pi and qi defined by
r
p0i 0 1
pi = √ and qi = qi mi ωi2 .
2mi 2
In terms of these p’s and q’s the Hamiltonian is
H = p21 + q12 + p22 + q22 = E1 + E2 .
√
It is clear that the phase space plots are simply circles of radius E,
as shown in Figure 20.2
p1 p2

q1 q2

Figure 20.2. Phase space diagrams for double oscilla-

tor for the case E2 < E1 .

It is interesting to represent both diagrams on a single plot. Assume

ω2 >> ω1 . First plot p1 vs q1 as a circle in the horizontal plane. Next
plot p2 vs q2 in a vertical plane, selected such that p2 is perpendicular to
the horizontal plane and the q2 axis points towards the origin. Thus,
at a given instant of time, the combined plot will be the two circles
shown in Figure 20.3.

p2
q1
q2 p1

Figure 20.3. A phase space diagram for the double oscillator.

698 20. CLASSICAL CHAOS (OPTIONAL)

At an instant of time, the state of the system is represented by a

single point whose position on the “high frequency” circle gives the
values of p2 and q2 . The instantaneous values of p1 and q1 are given by
the location of the center of the smaller circle. As time goes on the high
freqency circle moves around on the larger circle, and the point traces
out a helix. Eventually all of the locations of this point will generate
a torus. The point representing the state of the system moves on the
surface of the torus. If the two frequencies are commensurable, that is,
if
ω2
= an integer
ω1
then the trajectory of the system point will be closed and the path will
repeat over and over. If the frequencies are not commensurable, the
trajectory will never close and the system point will gradually cover
the surface of the torus, eventually coming arbitrarily close to every
point on the surface.
For three oscillators, the motion takes place on a 3-d surface called a
3-torus in six-dimensional space. (The parameters are p1 , q1 , p2 , q2 , p3 , q3 .)

20.3. Attractors
If a bounded, periodic, dynamical system is subjected to a small
perturbation, its behavior will change somewhat, but you would expect
it to still be bounded. The Russian mathematician Kolmogorov and,
later, Arnold and Moser developed and proved a theorem to that effect.
It has become known as the KAM theorem and it states that a bounded
system subjected to a small perturbation will remain bounded. How-
ever, the theorem does allow for chaotic motion to occur in some sys-
tems if the initial conditions are just right.
Consider a periodic system acted upon by a small perturbing force.
The orbit may decay and the phase space trajectory might end up at
some point in phase space and remain at that point from then on. Or
the system might evolve to a stable orbit (called a “limit cycle”) and
then go around in that orbit from then on. In these cases, the point or
the stable orbit are referred to as “attractors.”
In other words, an attractor is a set of points in phase space to
which the solution evolves, usually after a long interval of time. As an
example, consider a simple pendulum. The motion of an unperturbed
pendulum can be described by a circle in phase space. But if there is
a small perturbation, say a weak drag force, then the amplitude of the
pendulum swings will grow smaller and smaller and the momentum
will gradually decrease, until the pendulum comes to rest. If you were
to plot this on a phase space diagram you would see the trajectory is
 20.3. ATTRACTORS 699

an inward spiral that ends up at the origin (pφ = 0, φ = 0) . The origin

would be the attractor in this case.
Another example is a planet perturbed by an impact with a comet.
As you saw in section 10.10 the original circular orbit became an elip-
tical orbit, but the planet thereafter remained in the elliptical orbit.
This final orbit is the attractor for the perturbed planet. It is a limit
cycle.
If the attractor is a point in phase space (as in the case of the
pendulum) it is called an attractor of dimension zero. If the attractor is
a limit cycle, as in the case of the planet, it is an attractor of dimension
one. If the attractor is the surface of a torus, it has dimension two.
Some attractors associated with chaotic motion have noninteger
dimensions and are called “strange attractors.” For a strange attractor,
the phase space points visit and revisit a region of phase space, filling
it with a series of points. This is often a signature of chaos.
As an example of a system with a one-dimensional attractor, con-
sider the van der Pol equation,
d2 x dx
m 2 − (1 − x2 ) + mω02 x = F cos ωd t. (20.2)
dt dt
This equation actually gives a reasonable description for oscillations of
some mechanical, electrical, and biological systems. It has the form of
the equation for a nonlinearly damped, driven pendulum. Note that
F = 0 and  = 0 lead to simple harmonic motion. If F = 0 and  is
small, the phase space trajectory tends to a circle (a limit cycle), as
indicated in Figure 20.4.

Figure 20.4. Depending on whether x2 is greater or

less than unity, the system spirals in or out to the limit
cycle. For stronger damping, the limit cycle is distorted
in shape.
700 20. CLASSICAL CHAOS (OPTIONAL)

Exercise 20.3. Plot the phase space trajectory for a damped os-
cillator.

20.4. Chaotic Trajectories and Liapunov Exponents

If the phase space trajectory ends up at a point or on a limit cycle,
as described above, the motion is well behaved and the trajectory is
confined to a limited region of phase space. Such motion is obviously
not chaotic.
Sometimes, however, the trajectory will wander about in a seem-
ingly random manner thus filling a region of phase space (which is, of
course, the strange attractor described above). The trajectory never
passes through the same phase space point twice. This is chaotic mo-
tion, and it is characterized by three properties:
Property 1: Mixing
If I1 and I2 , are two small regions in the domain of the motion,
then an orbit or trajectory that passes through I1 will eventually pass
through I2 .
Property 2: Dense, Quasi-Periodic Orbits
Chaotic orbits are called “quasi-periodic” because they repeatedly
pass through the whole range of the domain. However, they are not
closed and there is no periodicity (no regularity) in the time between
visiting and revisiting a particular region. The trajectory visits and
revisits every region of available phase space (which is the “mixing”
property).
Property 3: Sensitivity to Initial Conditions
A very small change in the initial conditions will lead to a very
large change in the final state. For example, in the non-chaotic flow of
a stream of water, two nearby water particles will remain close to one
another, but if the flow is turbulent, they will be separated by an ever
increasing distance.
A measure of the sensitivity to initial conditions is given by the
Liapunov exponent. Consider two nearby point on nearly identical
orbits. If their separation at time t = 0 is s0 , then at a later time their
separation is
s(t) ∼ s0 eλt (20.3)
where λ is called the Liapunov exponent.
If λ > 0, the motion is chaotic and s will grow until it has the size of
the available coordinate space, after which it will vary randomly with
time.
 20.5. POINCARÉ MAPS 701

If λ < 0, the system approaches a regular attractor.

For the planets in our neighborhood, it is believed that λ ≈ 3 ×
−10
10 /year, so the solar system seems to be chaotic, although barely
so. That is, it is a case of “marginal stability.”

20.5. Poincaré Maps

Consider again the double oscillator whose Hamiltonian is given by
Equation (20.1). The motion of the two oscillators were independent,
that is, the motions were uncoupled. The phase space plots were two
circles and showed, as expected, that the two oscillations were indepen-
dent. Nevertheless, it was possible to represent the motion on a single,
albeit somewhat complex, plot.
If, however, the motions are not independent, things get more com-
plicated. For example, perhaps the Hamiltonian contains a term such
as q12 q2 . In this case the phase space plots, in general, do not separate
and we need to use the entire four-dimensional phase space with axes
p1 , p2 , q1 , q2 .
Suppose, however, that the total energy is constant. This gives one
constraint. Recall that each constraint reduces the number of indepen-
dent coordinates by one. Therefore, the motion is confined to a three-
dimensional region of phase space. This is called a three-dimensional
energy surface or an “energy hypersurface.” As an example, consider
Keplerian motion. In Cartesian coordinates the position is specified by
x and y. The Hamiltonian will be a function of px , py , x, y. But if the
energy is constant, then there is a relation between these coordinates,
namely
p2x p2y k
E =T +V = + − 2 . (20.4)
2m 2m (x + y 2 )1/2
The easiest way to visualize the motion is to take a two-dimensional
slice through the energy hypersurface. This 2-D slice is called a “Poincaré
section” or, more commonly, a “surface of section.” The orbit is, of
course, just an ellipse, as shown in Figure 20.5.
At points A and A0 , the x component of momentum is zero. Plotting
the values of x, px at these two points (i.e., when y = 0) gives the
extremely simple plot shown in Figure 20.6.
The surface of section contains only two features, the dots at x = A
and x = A0 . I shaded in the left-hand side of the plot because it is
usually not shown. Then the surface of section has just one point, at
x = A. This just means that each time y = 0, the value of x is A and
the value of px is zero. Obviously the motion is regular.
702 20. CLASSICAL CHAOS (OPTIONAL)

A’ A
x

Figure 20.5. A Keplerian orbit in configuration space.

Points A and A0 are the periapsis and apoapsis.

px

A’ A
x

Figure 20.6. The surface of section for a Keplerian orbit.

Now suppose some sort of perturbation causes the elliptical orbit

to precess. Then the orbit will have the general appearance of Figure
20.7.
y

Figure 20.7. A precessing elliptical orbit.

Note that when y = 0 the planet is not always at the perihelion and
the point going through the x, px plane will gradually move along the
trajectory shown in Figure 20.8. (The point representing the system
configuration gradually moves from A to B to C and so on, generating
the x, px curve shown.)
Thus, a Poincaré section is a 2-D slice through a 3-D energy hy-
persurface in a 4-D phase space. For higher dimension phase spaces
it is possible to draw Poincaré surfaces, but their usefulness is mainly
limited to 4-D phase space.
 20.6. THE HENON HEILES HAMILTONIAN 703

px
C B
A x

Figure 20.8. The surface of section for a precessing ellipse.

20.6. The Henon Heiles Hamiltonian

Henon and Heiles2 suggested that the motion of stars in a galaxy
could be approximated by the following two-dimensional Hamiltonian:
p2x p2y k 1
H= + + (x2 + y 2 ) + λ(x2 y − y 3 ), (20.5)
2m 2m 2 3
where k is related to the gravitational constant. To simplify without
losing any essential information, let k = 1 and m = 1. If λ = 1 the
equations of motion are
ẍ = −x − 2xy, (20.6)
ÿ = −y − x2 + y 2 ,
and the potential energy is
1 1
V = V (x, y) = (x2 + y 2 ) + x2 y − y 3 .
2 3
This expression has an interesting form, which depends on the choice
for the total energy. For very small values of the total energy, the posi-
tion variables are small (x, y << 1) and the cubic terms are negligible,
giving V ≈ 21 (x2 + y 2 ), which is a circle in the xy-plane. Depending on
the value of V, the circles are larger or smaller. For larger values of V,
but requiring that V < 1/6, the equipotential curves have the shapes
indicated in Figure 20.9. At V = 1/6 the equipotential is the bounding
triangle.

V = 1/6
V = 1/8
V = 1/12

V = 1/24

Figure 20.9. Equipotentials for the Henon-Heiles potential.

2M. Henon and C. Heiles, “The applicability of the third integral of motion:
Some numerical experiments,” Astronomical Journal, 69, 73-79, 1964.
704 20. CLASSICAL CHAOS (OPTIONAL)

Obtaining the surface of section for this Hamiltonian is interesting

and instructive. The method is to choose some value for the energy,
then integrate the equations of motion (20.5). Whenever x = 0, the
values of y and ẏ are plotted as a point on the ẏy-plane.
For example, assume the energy is 1/12 and the initial conditions
are x = 0, y = 0.02, and ẏ = 0.08. The initial value of ẋ is determined
from the total energy
1 1 1 1
E = ẋ2 + ẏ 2 + (x2 + y 2 ) + x2 y − y 3 ,
2 2 2 3
and is ẋ = 0.4. New values for x and y can be generated from the
equations of motion. Each time the value x = 0 appears, the values
of y and ẏ are plotted. This gives the plot shown in Figure 20.10. If
you were to watch the figure being generated, at first you would see
a seemingly random set of points being plotted here and there on the
y ẏ-plane. However, after a large number of points have been plotted,
you would see the pattern beginning to form.

Figure 20.10. Surface of section for the Henon-Heiles

potential for initial conditions y = 0.02, ẏ = 0.08, and
E = 1/12.

Using a variety of different initial conditions yields a more complete

surface of section, as shown in Figure 20.11.
Each of the rather complicated curves represents an orbit that re-
peats over and over in a regular manner. The four regions inside the
looping curve that crosses itself three times contain regular orbits, that
is, nested orbits whose circumference depends on the initial conditions.
In the limit they shrink to four points called the “elliptic fixed points.”
A different value of the energy (but using the same initial conditions
as in Figure 20.11) leads to the significantly different surface of section
shown in Figure 20.12.
You can see this still has the four regions with stable orbits but
outside of them there is a region where the trajectories crossing the
 20.7. SUMMARY 705

Figure 20.11. Surface of section for the Henon-Heiles

potential for E = 1/12 and initial conditions (y, ẏ) =
(0.02, −0.08), (.3, .246), (.499, 0), (.366, 0), and (.2, .05).

Figure 20.12. The surface of section obtained with the

same initial conditions as in Figure 20.11 but for a total
energy E = 1/8.

x = 0 plane show no regularity at all. The curves representing orbits

that repeat over and over again now occupy a much smaller region of
the plot and the rest of the plot is taken up by points that show no
apparent pattern. This region of the plot represents chaotic motion,
and is called a strange attractor. Figure 20.13 shows the surface of
section obtained for an energy of E = 1/6. Now nearly all motion is
chaotic, although there are a few small regions with regular orbits.

20.7. Summary
In this very brief introduction to classical chaos, I have attempted
to give you some of the vocabulary used in the theory and to give you a
general idea of the basic ideas of chaos. Note that nonlinear equations
are sometimes very sensitive to initial conditions, in which case they
will lead to chaotic motion. Chaos is characterized by the generation
of random points on a phase portrait. The effect of initial conditions
706 20. CLASSICAL CHAOS (OPTIONAL)

Figure 20.13. The surface of section for E = 1/6.

is quantified by the Liapunov exponent which describes the divergence

of two trajectories that start out near each other. As an example, we
considered in some detail the Henon-Heiles Hamiltonian. The figures
illustrating chaotic motion in the Henon-Heiles potential are instructive
and well worth careful study.
20.8. Problems
Problem 20.1. Write the Henon-Heiles Hamiltonian in polar co-
ordinates.
COMPUTATIONAL PROJECTS
Computational Project 20.1. Assuming V = 1/6, plot the
Henon Heiles potential to show that it is an equilateral triangle. (Hint:
Ignore the complex solutions for the value of r vs θ.)
Computational Project 20.2. Generate a plot like that of Fig-
ure 20.4 for the van der Pol equation for F = 0 and small . Explore
the consequences of changing .
Computational Project 20.3. The van der Pol equation can be
written in the somewhat simpler form
d2 x dx
2
− µ(1 − x2 ) + x = 0.
dt dt
Plot ẋ vs x for µ < 0, µ = 0, and µ > 0.
Computational Project 20.4. Write computer programs to gen-
erate the last three figures in this chapter.
Computational Project 20.5. Consider the following equation
for a nonlinear, damped, driven pendulum:
d2 θ g dθ
2
= − sin θ − α + F0 sin(βt).
dt l dt
 20.8. PROBLEMS 707

Plot θ = θ(t) and ω = dθ

dt
for F0 = 0, 0.4, 1.1. Assume α = 0.5, g/l = 1
and β = 0.6. Let the initial conditions be θ(t = 0) = 0.3 rad and
ω(t = 0) = 0. rad/s.
Computational Project 20.6. Consider two identical damped,
forced pendulums. Both have the same equation of motion
d2 θ g dθ
2
= − sin θ − α + F0 sin(βt)
dt l dt
but differ slightly in the initial value of θ. You are asked to show that
if the F0 is small (say 0.1) the two pendulums stay in phase, but for
larger F0 (say 1.0), the angular distance between the pendulums grows
larger and larger. You may assume α = 0.5, g/l = 1, β = 0.6, and
ω(t = 0) = 0. The initial difference in displacement ∆θ = θ2 − θ1 is
small, say 0.001. Plot ∆θ = ∆θ(t) for the two pendulums on a semilog
graph. For the case of diverging ∆θ, estimate the Liapunov exponent.
Computational Project 20.7. The “logistic map” is a plot of
xn vs n assuming that
xn+1 = µxn (1 − xn ) 0 ≤ x ≤ 1.
This equation can be considered a model for the population growth of
animals. (a) Plot the logistic map for 0 < n < 100 for µ = 2, 3, and
4. Connect the dots with straight lines. You may assume x0 = 0.6. (b)
If the value of xn+1 = xn so that the value of x does not change, the
system has reached a “limiting value.” Show that for µ = 2.0 a limiting
value is reached. Show that for µ = 3.0 the system oscillates between
two limiting values. Show that for µ = 4.0 there is no limiting value
(the system is chaotic).
Computational Project 20.8. The Feigenbaum plot is a plot
of the limiting value of the logistic equation as a function of µ. (See
previous problem.) Let µ range from 1 to 4 and plot xn vs µ for
100 < n < 200.
Computational Project 20.9. The Lorenz equations are
dx
= σ(y − x),
dt
dy
= −xz + rx − y,
dt
dz
= xy − bz.
dt
(a) Plot z vs t for r = 25, σ = 10, and b = 8/3 and initial conditions
x = 1, and y = z = 0. Run the system to a (dimensionless) time t = 50.
(b) Plot z vs x.(Note that the time step must be small.)
708 20. CLASSICAL CHAOS (OPTIONAL)

Computational Project 20.10. The predator-prey system can

be modeled using the equations
ẋ = 0.6x − 0.02xy,
ẏ = −2y + 0.02xy.
Here, x represents the number of prey (say, rabbits) and y represents
the number of predators (say, foxes). Plot x and y as functions of time
on the same plot and interpret in terms of food supply for the predators
and in terms of survival probability for the prey. You may assume that
initially the number of prey is 100 and the number of predators is 5.
Computational Project 20.11. In general the predator-prey
relation is
ẋ = ax − bxy,
ẏ = −cy + dxy,
where a, b, c, d are positive constants. Assume a = 2, b = 0.2, c = 5,
and d = 0.2. Use initial conditions x = 10, 30, 80 and y = 5, 15, 15.
Obtain the configuration plots x vs y and interpret in terms of the
population of predators and prey.
Bibliography

Intermediate Level Texts:

[1] R. A. Becker Introduction to Theoretical Mechanics McGraw Hill, 1954.

[2] R. Kolenkow and D. Kleppner An Introduction to Mechanics Cambridge Uni-
versity Press, 2013.
[3] D. Morin Introduction to Classical Mechanics Cambridge University Press,
2013.
[4] K. R. Symon Mechanics Addison Wesley, 1961.
[5] S. T. Thornton and J. B. Marion Classical Dynamics Brooks Cole, 2003.

Graduate Level Texts:

[6] A. L. Fetter and J. D. Walecka Theoretical Mechanics for Particles and Con-
tinua McGraw Hill, 1980.
[7] H. Goldstein, C. P. Poole and J. Safko Classical Mechanics, 3d Ed. Pearson,
2011.
[8] P. Hamill A Student’s Guide to Lagrangians and Hamiltonians Cambridge
University Press, 2013.
[9] L. D. Landau and E. M. Lifshitz Mechanics, 3d Ed. Butterworth-Heinemann,
1976.

709
Appendix A
Formulas and Constants

Universal Constants
Gravitational constant = G = 6.67 × 10−11 Nm2 /kg2
Speed of light = c = 3 × 108 m/s

Astronomical Constants
Mass of Earth 5.97 × 1024 kg
Mass of Sun 1.99 × 1030 kg
Radius of Earth 6.37 × 106 m
Radius of the orbit of Earth around Sun 1.5 × 1011 m

Definition of Hyperbolic Functions

ez − e−z
sinh z =
2
ez + e−z
cosh z =
2
sinh z
tanh z =
cosh z

711
712 A. FORMULAS AND CONSTANTS

Integrals
Expressions involving terms like a2 ± x2

Z
dx 1 x
= arctan
a2+ x2 a a
Z
dx 1 x
= tanh−1
a − x2
2 a a
Z
dx x
√ = arcsin
a2 − x 2 a
Z
dx √
√ = ln(x + x2 − a2 )
x 2 − a2
Z
dx 2 2cx + b
= √ tan−1 √
a + bx + cx2 4ac − b2 4ac − b2
  
Z
dx  √1 sinh−1 √2cx+b 2 if c > 0
c
√ =  4ac−b 
a + bx + cx2  √1 sin−1 √−2cx−b if c < 0
−c b2 −4ac

Expressions involving trigonometric quantities

Z
tan xdx = − ln cos x
Z
sec xdx = ln(sec x + tan x)
Z
x 1
sin2 xdx =
− sin 2x
2 4
Z
x 1
cos2 xdx = + sin 2x
2 4

Expressions involving hyperbolic functions

Z
sinh xdx = cosh x
Z
cosh xdx = sinh x
Z
tanh xdx = ln cosh x
 A. FORMULAS AND CONSTANTS 713

Expression involving logarithms

Z
ln xdx = x ln x − x

Error Function Z x
2 2
erf(x) = √ e−t dt
π 0
Appendix B
Answers to Selected Problems

1.1 35 mph
1.5 (a) v1 +v
2
2
(b) v2v1 +v
1 v2
2
1.7 0.69 mi
1.9 (c) v = v0 [(sin θ − µ cos θ)/(sin θ + µ cos θ)]1/2
1.11 29 cm
1.13 2.58 × 1016 N
1.15 176 watts
1.17 (b) 2857 rad
1.19 Stable
1.21 (b) 189 sec
2.1 6m/s
2.5 48.5m/s
2.7 23.65◦ and 66.35◦
2.11 5609m h i1/2
2 x2
2.12(b) v = v02 − 2gx tan θo + v2gcos 2θ
gx
, tan θ = tan θ0 − v2 cos2θ .
0
0 0
2.13 15.7m2 2
v cos θ sin θ
2.15 s = 0 2g [ cos 2 θ + log(sec θ + tan θ)]

2.17 v = [(gxt /2yt ) + 2gyt ]1 /2 at θ = arctan(2yt /xt )

2

2.18 v =30m/s
2.19 R=455km
2.20 ac =75m/sp
2.21(a) τ = 2π r/a.
2.22 ω = 10.35π cos πtî + 10.35π sin πtĵ + 39.64π k̂
2.26 -24
2.30 (a) This is the spiral of Archimedes. (b) v = (4 + 100t2 )1/2
m/s.
2.31 v = (cos θ)r̂ + (1 + sin t)e−t θ̂
715
716 B. ANSWERS TO SELECTED PROBLEMS

2.33 θ = θ0 + ln(1 + sin t)

2.37 1.56m/s p
2.39 ∇ · F= 2 + z cos φ/ρ + 12 ρ/z.
2.40(b) ṙ = √v2vtt2 +1 , θ̇ = 1+vv2 t2
2.45(b) v = [b2 + (akt − bkt2 )2 ]1/2
ma
2.46 (b) r = eB ρ̂ + atφ̂ + btk̂
0 √
3.1 a = (3ı̂ + ̂)m/s2 .|a| = 10m/s2
3.2 3.06 m/s
3.3 0.51
3.5 F (x) = −(mαA2 /2) sin 2αx
3.7 (c) 5.63 × 10−15 m/s2
3.8 (b) a = g(m1 − m2 )/(m1 + m2 + I/R2 )
3.11 1020 meters.
3.16 x = (5/6)L.
v0
3.17 (a) F (x) = −mb(v h 0 − bx), (b) x(t) =ix0 + b
(1 − e−bt )
3.18 (a): x(t) = Fκm2 0 − mκ0 t − ln(1 − mκ0 t) .
3.19 0.667 m
3.21 3.58 ft/s2 and -7.16 ft/s2
3.23 3.33 n/m
3.25 7.6 cmh i
A e−αt α
3.27 v = m α2 +β 2 (−α cos βt + β sin βt) + α2 +β 2
q √ 
3.29 v(t) = b tanh mF b t
F

3.31 (b) h = (m/2D) ln [(v02 + vT2 ) /vT2 ]

3.33 (a) 21.8 m2 (b) 0.35 s
3.35 vT = 55.87 m/s
3.39 t = 14.7 s
3.41 1.67
p m/s
3.43 x30 π 2 /8GM
q.
k
3.45 (b) v0 = − m x0 .
3.47 ' 42 minutes.
Index

acceleration BAC-CAB rule, 245

angular, 16 barometric equation, 504
centripetal, 42 baryons, 273
constant, 34 binomial expansion, 357, 424
action, 133, 147 body force, 490
action at a distance, 281, 284 Brahe, Tycho, 319
active interpretation, 523 bulk modulus, 494
adjoint matrix, 525
cable, 487
air resistance, 89
calculus of variations, 128
Angular Momentum
canonical conjugates, 137
relative to center of mass, catenary, 490
242 center of gravity, 497
angular momentum, 233, 323, center of mass, 182, 497
550 angular momentum relative
conservation, 236 to, 243
of a system of particles, 237 center of mass coordinates, 216
tensor notation, 554 center of oscillation, 427
angular velocity, 16 central force, 317, 322
angular velocity vector, 244 centroid, 495
antisymmetric matrix, 544 chaos, 694
Archimedes’ Principle, 502 charge conjugation, 266, 272
areal velocity, 349 closed form solution, 94
Aries coefficient of restitution, 211
first point in Aries, 325 collision, 200
attractor, 698 elastic, 202
Atwood’s machine, 100 glancing, 208
axial vector, 253 head on, 205
azimuthal angle, 55 inelastic, 210
717
718 INDEX

comets, 334 damped harmonic oscillator,

commensurability, 362, 401 383
computational physics, 94, 96 damping coefficient, 385
configuration space, 694 degeneracy, 558
conic section, 335 degree of freedom, 122
conjugate momentum, 125 del, 156
conservation Descartes, Rene, 71
of angular momentum, 236, deterministic system, 694
239 differential equation
of parity, 271 homogeneous, 377
conservation inhomogeneous, 390
of energy, 173 direction cosines, 520
conservation laws, 265 divergence, 157
conservative force, 160 dot product, 45
constant of the motion, 125, drag coefficient, 89
317 dyad, 553
constraint, 122, 505, 518 dyadic, 553
holonomic, 505 dynamics, 76
continuous systems, 641 rotational, 518
coordinate systems, 43
accelerated, 451 eccentricity, 343, 344
cartesian, 47, 51 effective potential, 332, 436
cylindrical, 52 effective potential energy, 317,
inertial, 29, 451 331
plane polar, 47 eigenvalue, 533, 535, 556, 623
rotating, 453 degenerate, 558
spherical, 55 equation, 533
coordinate transformation, 162 eigenvalue
coriolis force, 461 equation, 633
couple, 485 eigenvector, 533, 535, 625
coupled homogeneous equation, 625
equations, 623 eigenvector
coupled pendulums, 632 equation, 634
coupled systems, 613 Eightfold Way, 273
CP violation, 272 Einstein
CPT invariance, 272 summation convention, 521
Cramer’s rule, 403, 625 Einstein, Albert, 76, 654
cross product, 45 elastic collision, 203
curl, 157 electrostatic force, 322
cylindrical coordinates, 165 ellipse, 342
elliptic integral, 420, 422
d’Alembert’s principle, 505 endoergic, 211
 INDEX 719

energy, 149 Galilean transformation, 656

of system of particles, 182 Galileo Galilei, 30, 71
energy diagram, 177 Gauss’s Law, 296
energy integral, 179 Gaussian surface, 297
equation of motion, 74, 110 gedanken experiment, 658
equilibrant, 486 generalized coordinate, 122
equilibrium, 19 generalized momentum, 123,
Euler 268
angles, 518, 528 gradient, 157
equations of motion, 569 gradient
method, 97 geometrical interpretation,
theorem, 532 157
Euler’s formula, 378 Gram-Schmidt procedure, 558
Euler-Cromer algorithm, 98 gravitational
Euler-Lagrange equation, 131, field, 279, 282
133 field equations, 294
exoergic, 211 potential, 290
gravity
Feynman diagrams, 285 center of, 497
Feynman, Richard, 285 gyroscope, 250
fictitious force, 455
field Hamilton’s equations, 135
lines, 294 Hamilton’s principle, 112, 132
theory, 279 Hamiltonian, 134
first integral, 125 harmonic motion, 91, 373
force harmonic oscillator
as function of position, 91 coupled, 402
central, 317, 322 normal modes, 407
coriolis, 453 critically damped, 389
fictitious, 453, 455 damped, 375, 383
function of time, 84 forced, 390
function of velocity, 86 damped, 376, 397
generalized, 506 undamped, 396
gravitational, 280, 322 overdamped, 387
internal, 212 underdamped, 385
Lennard-Jones, 322 Helmholtz theorem, 296
Foucault pendulum, 468 Henon Heiles Hamiltonian, 703
four scalar, 678 hermitian matrix, 525
four vector, 678 hermitian tensor, 556
Fourier series, 595, 643 hodograph, 41
functional, 129 holonomic constraint, 505
720 INDEX

homogeneous algebraic line integral, 151

equations, 403, 472, 534 line of action, 17
Hooke’s Law, 493 line of nodes, 573
hyperbola, 344 linear differential equation, 377
lines of force, 294
ideal gas, 504 Lorentz transformation, 658,
ignorable coordinate, 124, 268 662
impact parameter, 200
impulse, 212 Mach, Ernst, 76, 82
inelastic collision, 210 mass
inertia, 31 gravitational, 76
law of, 71 inertial, 76
moment of, 248, 551 reduced, 215
product of, 551 mass matrix, 615
inertia tensor, 247, 551 mesons, 273
Inertia, moment of, 247 metric, 163, 678
inertial reference frame, 73, 451
Michelson Morley experiment,
655
KAM theorem, 698
Minkowski, 673
Kater’s pendulum, 448
modal matrix, 536, 629
Kepler’s laws, 321, 349
moment
Kepler, Johannes, 318
of force, 17
kinematics, 29, 30, 32
Moment of Inertia, 247
rotational, 518
moment of inertia, 247, 248,
kinetic energy
551
rotating body, 554
momentum
Kolmogorov, 698
Kronecker delta, 520, 551 angular, 550
conservation of, 195
Lagrange’s equations, 117, 118, linear, 195
507 of system of particles, 212
Lagrangian, 110, 112, 330
Lagrangian density, 644 Navier-Stokes equation, 96
Laplace vector, 366 Newton’s first law, 71
Laplace’s equation, 298 Newton’s laws
law of cosines, 65 Newton’s second law, 74
Lee, T. D., 272 Newton’s third law, 78
length contraction, 661 strong form, 78
lever arm, 17 weak form, 78
Levi-Civita density tensor, 583 Newton’s laws of motion, 70
Liapunov exponent, 700 Newton’s Rule, 211
limit cycle, 698, 699 Newton, Isaac, 71
 INDEX 721

law of universal gravitation, Poisson’s equation, 298

280 polar angle, 55
Noether, E., 267 position, 32
normal coordinates, 630 potential
normal frequencies, 627 effective, 317, 331
normal modes, 407, 619, 623, gravitational, 290
634 potential energy, 155, 160
numerical methods, 95 effective potential energy,
numerical solution, 94, 96 331
nutation, 252, 517, 578 potential matrix, 617
power, 14
Oort cloud, 334 precession, 251, 517, 576
orthogonal principal axes, 535, 556
coordinate system, 162 principle of superposition, 76
matrices, 523 principle of virtual work, 506
transformations, 517, 520, priniciple of superposition, 286
536 product of inertia, 551
orthogonality, 45, 52, 59, 520, pseudowork, 183
626 psuedowork, 185
orthogonality condition, 520
orthonormality, 626, 628 Q value, 210
oscillations
small, 613 radius of gyration, 426
reduced mass, 215
Pappus, theorem of, 496 relative coordinate, 215
parabola, 344 relativity, 653, 654
parallel axis theorem, 248 general theory, 654
parity, 271 special theory, 654
Pascal’s law, 501 resonance, 362, 390, 397, 398
passive interpretation, 523 restitution
pendulum coefficient of, 211
compound, 417, 425 resultant, 485
conical, 417, 438 rigid body, 517
physical, 417, 425 RLC circuit, 401
simple rocket, 196
plane, 417 Rotation of a rigid body
spherical, 417, 434, 439 about a fixed axis, 244
perpendicular axis theorem, rotational dynamics, 549
248 rotations
perturbation analysis, 462 infinitesimal, 543
phase space, 695 row vector, 557
Poincare map, 701 Runge-Lenz vector, 366
722 INDEX

Rutherford, 203 transformation equations, 56,

112
scalar product, 45 transformations
scale factor, 163 improper, 253
secular equation, 470, 535 proper, 253
shear modulus, 494 turning point, 178
similarity transformation, 527, twin paradox, 671
536
unit vectors, 44
simple harmonic motion, 91 unitary matrix, 525
simultaneity, 668 universality, 281
small oscillations
frequency of, 359 van der Pol equation, 699
spherical coordinates, 55, 166 Varignon, theorem of, 482
standard model, 273 vector
static equilibrium, 19 four vector, 678
statics, 19, 76 vector product, 45
stochastic behavior, 694 velocity
strain, 493 angular, 16
strange attractor, 699 areal, 349
strangeness, 273 virial theorem, 326
stress, 493 virtual displacement, 506, 644
stress tensor, 494, 495 virtual photon, 285
string, 487 virtual work, 505, 506
successive approximations, 440 principle of, 506
summation convention, 521
surface of section, 701 work, 149
symmetry, 265 virtual, 505, 506
work energy theorem, 150
work-energy theorem, 555
Taylor series, 357, 444 world line, 674
tensor, 551 world scalar, 678
time, 32 Wu, C. S., 272
time dilation, 660
time reversal, 267 Yang, C. N., 272
torque, 17, 18 Young’s modulus, 494