Solution to the advance of the perihelion

of Mercury in Newtonian theory

Christian Corda
June 15, 2020

International Institute for Applicable Mathematics and Information

Sciences, B. M. Birla Science Centre, Adarshnagar, Hyderabad
500063 (India). E-mail: [email protected]

Abstract
It is shown through three dierent approaches that, contrary to a long-
standing conviction, the orbit of Mercury behaves as required by Newton's
equations with a very high precision if one correctly analyzes the situation
in the framework of the two-body problem without neglecting the mass
of Mercury.

1 Introduction

Based on astronomical observations, in the early 1600s Kepler established that

the orbit described by a planet in the solar system is an ellipse, with the Sun
occupying one of its foci. Assuming that a planet is subject only to the gravi-
tational attraction of the Sun, Kepler's result is easily obtained mathematically
in Newton's theory. But the other planets also have a gravitational attraction
on the planet in question. What is the eect of their presence? If one re-
peats the calculation taking into account this complication, one nds that the
attraction exerted by all the other planets of the solar system on the planet
in question induces an advance (a precession), orbit after orbit, of the perihe-
lion (the point of maximum approach to the Sun of the orbit of the planet).
The precession of the Earth's rotation axis also gives rise to the same eect.
For example, Mercury's perihelion moves slightly at the speed of 5,600 arc-
seconds per century, in the same direction in which the planet rotates around
the Sun. However, when the contribution of the Earth's precession is removed
(5,025 arcseconds), that due to the attraction of the other planets, calculated
according to Newtonian physics, is not able to correctly predict what happens
in reality. The balance indeed misses 43 arcseconds. It is a general conviction,
supported by centennial computations, that this deviation of Mercury's orbit
from the observed precession cannot be achieved by Newtonian theory. This is

1
the famous anomalous rate of precession of the perihelion of Mercury's orbit.
It was originally recognized by the French Astronomer Urbain Le Verrier in
1859 as being an important astronomical problem (Le Verrier, 1859). Starting
from 1843 (Le Verrier, 1843), Le Verrier indeed reanalyzed various observations
of the perihelion of Mercury's orbit from 1697 to 1848, by showing that the
rate of the precession of the perihelion was not consistent with the previsions
of Newtonian theory. This discrepancy by 3800 arcseconds per tropical century,
00
which has been corrected to 43 by the Canadian-American astronomer Simon
Newcomb in 1882 (Newcomb, 1882) seemed till now impossible to be accounted
through Newton's theory. Various ad hoc and unsuccessful solutions have been
proposed, but such solutions introduced more problems instead. The most fa-
mous solution by 19th century astronomers was the attempting to explain this
discrepancy through the perturbing eect of a planet, Vulcan, hitherto escaped
observation, smaller than Mercury and closer than this to the Sun. However, the
search for this planet turned out to be unfruitful. The solution of the problem
is due to Albert Einstein through his magnicent general theory of relativity in
1916 (Einstein 1916). Recent analyses due to the MESSENGER data plus the
Cassini mission gave a value of about 42, 9800 to the general relativistic contri-
bution to the precession of perihelion for Mercury per tropical century (Park et.
al., 2017). In general relativity, if one expresses the perihelion shift in radians
per revolution (in this work, polar coordinates will be used), one gets instead
(Dediu, Magdalena, and Martìn-Vide, 2015)

24π 3 a2
4ϕ ' , (1)
T02 c2 (1 − e2 )

where a is the semi-major axis of the orbit, T0 is Mercury's Newtonian orbital

period, c is the speed of light, and e is the orbital eccentricity. Eq. (1) corre-
sponds to a total angle swept per revolution by Mercury

2π 2 a2
 
ϕ ' ϕ0 1+ 2 2 , (2)
T0 c (1 − e2 )

where ϕ0 = 2π is the unperturbed (i.e. in absence of precession) total angle

swept by Mercury during a complete revolution around the Sun. Inserting
the numerical values in Eq. (1), see for example (Williams, 2018), (Williams,
2019 I) and (Williams, 2019 II), one gets the well known general relativistic
value 4ϕ ' 5.02 ∗ 10−7 radians per revolution which corresponds to about 0, 1
arcseconds.
Now, the precession of the perihelion of Mercury's orbit is calculated in the
Newtonian framework. Three dierent approaches will be considered.

2 Approximation of circular orbit

One starts from the case in which Mercury's mass is considered negligible with
respect to the mass of the Sun, i.e. one considers the planet as being a test

2
mass immersed in the Newtonian gravitational eld of the Sun. In addition,
one considers Mercury's orbit as being circular instead of elliptical. Thus, the
case under consideration here is the simplest one. One takes the origin of the
Newtonian inertial frame of reference in the center of the Sun. By using the tra-
ditional Newtonian equations, in order to obtain the orbital period, one merely
equals the gravitational force to the centripetal one as

GM m mv02
2 = , (3)
r0 r0
where G is the gravitational constant, M is the solar mass, m the mass of
Mercury, r0 the orbit's radius and v0 the velocity of rotation of the planet.
Hence, v0 is easily obtained as
 
GM 1
v0 = .2 (4)
r0
Then, the Newtonian orbital period is

3
2πr0 2πr02
T0 = = 1 . (5)
v0 (GM ) 2
The corresponding angular velocity is

2π
ω0 = . (6)
T0
Thus, in radians per revolution the angular distance that Mercury sweeps during
the Newtonian orbital period T0 is

ϕ0 = ω0 T0 = 2π. (7)

Now one asks: what does it happen if one removes the approximation to consider
Mercury's mass negligible with respect to the solar mass? One argues that a
Newtonian observer set in the center of the Sun must replace Eq. (3) with

G (M + m) m mv 2
= , (8)
r02 r0
i.e. one must replace M with M +m in Eq. (3). Let us clarify this point. The
Newtonian law of universal gravitation can be written down in its general form
for Mercury and the Sun as

→
− GM m
FG = ûr , (9)
r2
where r is the distance between the Sun and Mercury and ûr is the unit vector
in the radial direction. Thus, for an external inertial Newtonian observer, the
Newtonian equations of motion for the Sun and Mercury are

GM m Gm
M as ûr = ûr =⇒ as ûr = 2 ûr (10)
r2 r

3
and
GM m GM
mam ûr = − 2
ûr =⇒ am ûr = − 2 ûr , (11)
r r
respectively, where as is the acceleration of the Sun and am is the acceleration
of Mercury. Thus, the relative acceleration of Mercury with respect to the Sun
is  
GM Gm G (M + m)
aûr ≡ am ûr − as ûr = − + 2 ûr = − ûr . (12)
r2 r r2
Then, the total force acting on Mercury as it is seen by a Newtonian observer
set in the center of the Sun is

G (M + m) m
F ûr = − ûr , (13)
r2
which immediately justify Eq. (8) for a circular motion. From Eq. (8) one gets
immediately the perturbed velocity of rotation of the planet as

  12
G (M + m)
v= (14)
r0
corresponding to a period

3
2πr0 2πr02
T = = 1 . (15)
v [G (M + m)] 2
But it is also
− 12 1
 m − 12
(M + m) = M−2 1 + , (16)
M
which, inserted in Eq. (15), gives

3
1
m −2
2πr02 1 + M
 m − 12
T = 1 = T0 1 + . (17)
[G (M )] 2 M

Then, the corresponding perturbed angular velocity is

2π 2π  m  12  m  12
ω= = 1+ = ω0 1 + . (18)
T T0 M M
Hence, the angle that Mercury sweeps during the period T0 is

 m  12 m 
ϕ = ωT0 = 2π 1 + ' 2π 1 + , (19)
M 2M
in radians per revolution, where in the last step the rst-order approximation in
m m 2

1
m
M has been used, that is 1 + M ' 1 + 2M , because it is m  M. Therefore,
in each complete revolution around the Sun, Mercury sweeps an angle larger
than the unperturbed angle (7) and the dierence, in radians per revolution, is

πm
4ϕ = ϕ − ϕ0 ' . (20)
M

4
The NASA ocial data give m ' 3.3 ∗ 1023 Kg (Williams, 2019 I) and M =
1, 99 ∗ 10 Kg 30
(Williams, 2018). Thus, one gets 4ϕ ' 5.21 ∗ 10−7 radians per
revolution which corresponds to about 0, 107 arcseconds. On the other hand,
the Mercury/Earth ratio of the tropical orbit periods is 0.241(Williams, 2019
II). Thus, one gets 44.3900 per tropical century. This is a remarkable result that
shows that, despite the above calculation has been made in the approximation
of circular orbit, the correct value of the contribution of Newtonian theory to
the precession of perihelion for Mercury per tropical century well approximates
the value of about 42, 9800 per tropical century of general relativity (Park et. al.,
2017) and the well known observational value of 4300 per tropical century.
The physical interpretation of this nice result is that it is Mercury's back re-
action, in terms of Newton's third law of motion, see Eqs. from (9) to (13), that
generates the advance of the perihelion of Mercury in Newtonian framework.

3 Mercury's orbit as harmonic oscillator

Following (Price and Rush, 1979), one recalls that each central attractive force
can produce a circular orbit that should not necessarily be closed. It is closed
if the radial oscillation period is a rational multiple of the orbit period. Now,
let Fc (r) be the total central force. Mercury's equation of motion in the radial
direction is given by (Price and Rush, 1979)
 
Fc (r) = m r̈ − θ̇2 r . (21)

The last term in Eq. (21) can be physically interpreted as a force centrifuge.
Since the angular momentum J is a constant of motion, one has that

J = mr2 θ̇. (22)

Solving for θ̇ and substituting in Eq. (21), one gets

J2
 
Fc (r) = m r̈ − 2 3 . (23)
m r
In the case of a circular orbit of radius r0 , r̈ = 0 and Eq. (23) reduces to

J2
Fc (r0 ) = − . (24)
mr03
If Mercury is now slightly perturbed in the plane of its orbit and perpendicularly
to its initial trajectory, it will oscillate around r0 (Price and Rush, 1979). Then,
one introduces x = r − r0 and expresses the radial equation of motion in terms
of x. Therefore (Price and Rush, 1979)

J2
Fc (x + r0 ) = mẍ − m(x+r0 )3
(25)
J2
= mẍ − x

3 .
mr03 1+ r
0

5
 x
Since
r0  1, one can use series expansion for the term in parentheses, con-
x
sidering only the rst order terms in
r0 . Expanding the member on the left in
Taylor series around the point r = r0 one gets (Price and Rush, 1979)

J2
 
3x
Fc (r0 ) + Fc0 (r0 ) = mẍ − 1− . (26)
mr03 r0

Inserting Eq. (24) in Eq. (26) one obtains (Price and Rush, 1979)

 
−1 3Fc (r0 ) 0
ẍ + m − − Fc (r0 ) x = 0 (27)
r0

One notes that this equation describes a simple harmonic oscillator if the term
in parentheses is positive (Price and Rush, 1979). If this term was negative,
there would be an exponential solution and the orbit would not be stable (Price
and Rush, 1979). Thus, for stable orbits, the period of oscillation around r = r0
is (Price and Rush, 1979)

! 21
m
T0 = 2π . (28)
− 3Fcr(r
0
0)
− Fc0 (r0 )
ϕ0
One denes the apse angle
2 as the angle swept by the radial vector between
two consecutive points of the orbit where the radial vector itself takes on an
extremal value (Price and Rush, 1979). The time that Mercury needs to travel
T0
this angle is
2 . Since the orbit can be considered approximately circular and
being therefore constant r and equal to r0 , one solves Eq. (22) for θ̇ and nds
(Price and Rush, 1979)

! 12
ϕ0 T0 m J
= θ̇ = π . (29)
2 2 − 3Fcr(r 0)
− Fc0 (r0 ) mr02
0

Furthermore, observing Eq. (24), one notes that the last term of Eq. (29) can
be rewritten as (Price and Rush, 1979)

  12
J Fc (r0 )
= − . (30)
mr02 mr0

Then, one gets (Price and Rush, 1979)

− 1
Fc0 (r0 ) 2

ϕ0 = 2π 3 + , (31)
Fc (r0 )

and, by setting Fc = FG in Eq. (31), where FG is the Newtonian gravitational

force given by Eq. (9), one nds ϕ0 = 2π , which is exactly Eq. (7).
But, again, in the computation in this Section Mercury's mass has been
considered negligible with respect to the mass of the Sun. A good way to

6
take into account the presence of Mercury's mass is to work in the framework
of the two-body problem. The two-body problem studies the dynamics of a
system consisting of two massive objects (the Sun having mass M and Mercury
having mass m in the present case) subjected to a central force. Central force
is dened as a force that only depends by the modulus of the dierence of the
vectors position of the two objects and which is directed along the junction of
the two bodies. The expression of this kind of force is well known:

→
− →
−rm−→ −rM
F = F (|rm − rM |) →− →
− . (32)
| r m − r M|
One considers, in an inertial reference system, the two objects of mass M and
m subject to the central force of Eq. (32). One introduces the variables relative
position, r , and position of the center of mass, R. In this way, it is always
possible to approach to the general two-body problem with two independent
problems through the following change of variables:

→
− m→
−
r m +M →
−
rM
R = M +m
(33)
→
−
r =→
−
rm−→
−
r M.
With this change of variables the positions of Mercury and the Sun can be
written as:
→
− →
− M →−
rm= R+ M +m r
(34)
→
− →
− m →−
rM = R− M +m r .
Mm
One also denes MT ≡ M + m and µ ≡ M +m as the total mass and the reduced
mass of the system, respectively. It is well known that the problem of the
dynamics of two bodies of masses m and M interacting through one force that
depends only on mutual distance is reduced to the problem of a single body
of reduced mass µ that moves in space under the action of a central eld. In
other words, in order to have a better description of the Sun-Mercury system
one makes the replacement m → µ in Eqs. from (21) to (31). In particular,
Eqs. (29) and (30) now read

! 21
ϕ T µ J
= θ̇ = π . (35)
2 2 − 3Fcr(r 0)
− Fc0 (r0 ) µr02
0

and
 1
J Fc (r0 ) 2
= − , (36)
µr02 µr0
m
respectively. To rst order in
M the reduced mass can be rewritten as

M +m −1 1 1 −1



µ= Mm = m + M
(37)
m −1 m



=m 1+ M 'm 1− M .

7
Thus, Eq. (35) becomes

  12
ϕ T m(1− M
m
) J
2 = 2 θ̇ ' π 3Fc (r0 )
m(1− M
− r0 −Fc0 (r0 )
m
)r02
(38)
  21
m m J


'π 1+ 2M 3Fc (r0 ) mr02
.
− r0 −Fc0 (r0 )

Then, one gets

− 1
Fc0 (r0 ) 2

m 

ϕ = 2π 1 + 3+ , (39)
2M Fc (r0 )
where now it is
GMT µ
Fc = r2
(40)
G(M +m)( M
Mm
+m )
= r2 = FG
and one nds  m 
ϕ = 2π 1 + , (41)
2M
which is the same result of Eq. (19).

4 Weak deviation from third Kepler's law

One starts by replacing m→µ in Eq. (22) obtaining

J = µr2 θ̇ = 2µȦ0 , (42)

where A is the area swept by ~r during the orbital motion. Thus, one obtains

dA dA
J = 2µ and dt = 2µ . (43)
dt J
Then, by integration over a period, one obtains

A
T = 2µ . (44)
J
Recalling that the generic expression for the area of a conic is given by

1
A = πa2 (1 − e) 2 , (45)

where a and e are the semi-major axis and the eccentricity of the ellipse, re-
spectively, one substitutes for (44) and gets

1
a2 (1 − e) 2
T = 2πµ . (46)
J

8
Also remembering that it is

J2
= a (1 − e) , (47)
µk
one obtains
1
J = [µak (1 − e)] 2 . (48)

Then, by inserting Eq. (48) in Eq. (46) and by using a bit of algebra, one gets

 21
a3 µ

T = 2π . (49)
k
As it is k = GM m for the gravitational system of Mercury and the Sun, Eq.
(49) becomes
 12
a3

T = 2π , (50)
GMT
where MT = M + m is the total mass of the system. Hence, from Eq. (50) one
easily obtains
a3 GMT
T2 = 4π 2
(51)
G(M +m) GM m


= 4π 2 = 4π 2 1+ M ,
and one immediately sees that Kepler's third law, that is  the ratio between T2
and a3 is constant for each planet in the solar system, depending only on the
mass of the Sun and not from that of the planet , i.e.

a30 GM
= , (52)
T02 4π 2
is strictly correct only in the approximation m  M, when the mass of the
planet is considered negligible with respect to the solar mass. a0 and and T0 in
Eq. (52) are the unperturbed semi-major axis and the unperturbed period of
revolution of the ellipse, respectively. Therefore, if one considers the mass of the
planet as being not negligible with respect to the solar mass, Eq. (51) shows
that there is a weak deviation from Kepler's third law in Newtonian gravitation.
Combining Eqs. (52) and (51) one obtains

a3 a30 m


T2 = T02
1+ M
(53)
a3 T2 m


=⇒ a30
= T02
1+ M .

On the other hand, if one wants that the variation of the angle merely makes the
ellipse precess (Misner, Thorne and Wheeler, 1973), that means that the shape
and area of the ellipse remain unchanged during the advance of the perihelion,
one must set a = a0 . Then, inserting this in Eq. (53) one immediately gets

T0
T =
12 , (54)
m
1+ M

9
which is exactly the result of Eq. (17) that was obtained in Section 2 in the
approximation of circular orbity. One also easily checks that Eq. (54) is also
consistent with Eq. (38) in Section 3. Thus, the corresponding perturbed
angular velocity is

2π 2π  m  12  m  12
ω= = 1+ = ω0 1 + . (55)
T T0 M M
Hence, the angular distance that Mercury sweeps during the period T0 is

 m  12  m 
ϕ = ωT0 = 2π 1 + ' 2π 1 + , (56)
M 2M
in radians per revolution, where in the last step the rst-order approximation
m
in
M has been used exactly like in previous Sections. The result of Eq. (56)
is the same as that of Eqs. (19) and (41), but the analysis in this Section is
more precise because it has been performed in the framework of the two-body
problem and considering the exact elliptical orbit of Mercury.

5 Conclusion remarks

It has been shown through three dierent approaches that, contrary to a long-
standing conviction longer than 150 years, the orbit of Mercury behaves as re-
quired by Newton's equations with a very high precision if one correctly analyses
the situation in the framework of the two-body problem without neglecting the
mass of Mercury. The results obtained are remarkable. The real value predicted
by Newtonian theory concerning the advance of the perihelion of Mercury is of
44.3900 per tropical century that well approximates the value of about 42, 9800 per
tropical century of general relativity and the well known observational value of
4300 per tropical century. Thus, the real dierence between Einstein's and New-
ton's previsions concerning the advance of the perihelion of Mercury is not of
about 4300 as astronomers and researchers were thinking for more than 150 years.
00
Instead, such dierence is only of 1.41 per tropical century. The physical in-
terpretation of this remarkable result is that it is Mercury's back reaction that
generates the advance of the perihelion of Mercury in Newtonian framework.
General relativity remains more precise than Newtonian theory regarding the
precession of Mercury's perihelion, but the dierence is very little. Another
important point is that general relativity achieves a very precise value for the
advance of the perihelion of Mercury considering the planet as being a test
mass immersed in the general relativistic gravitational eld of the Sun. In-
stead, in order to gain power of predictability, Newtonian theory must consider
Mercury's mass as being not negligible. Thus, surely the results in this paper
are not against the great power of predictability of Einstein's theory. They in-
stead endorse the issue that Newtonian theory is more powerful than researchers
and astronomes were thinking till now! One also recalls Einstein's opinion on
Newton's reseach work (Einstein 1999): Enough of this! Newton, forgive me;
you found the only way which, in your age, was just about possible for a man

10
of highest thought - and creative power. The concepts, which you created, are
even today still guiding our thinking in physics, although we now know that they
will have to be replaced by others farther removed from the sphere of immediate
experience if we aim at a more profound understanding of relationships.
In order to nalize this paper, one argues that it is not too much surprising
that Newtonian theory can give such a very precise value for the advance of the
perihelion of Mercury. In fact, it is a well known issue that general relativis-
tic corrections are strictly necessary in physics only in the cases of relativistic
velocities and/or strong gravitational elds. But, on one hand, the velocity
of Mercury in its revolution around the Sun is never relativistic. On the other
hand, there is no presence of strong gravitational elds, because the distance be-
tween Mercury and the Sun is always much longer than the Sun's gravitational
radius.
Thus, the present paper shows that Newtonian theory still has big surprises
for scientists!

Aknowledgements

The Author thanks Domenico Annunziata for stimulating discussions concerning

the possibility to solve the problem of the perihelion of Mercury's orbit is a
Newtonian framework.

References

• A. H. Dediu, L. Magdalena and C. Martìn-Vide, Theory and Practice

of Natural Computing: Fourth International Conference, TPNC 2015,
Mieres, Spain, December 15-16, 2015. Proceedings (2015).

• A. Einstein, Annalen der Physik 49(7), 769 (1916).

• A. Einstein, Autobiographical Notes, Open Court; Centennial edition (1999).

• U. Le Verrier, Comptes Rendus de l'Académie des sciences 49, 379 (1859,

in French).

• U. Le Verrier, Comptes Rendus de l'Académie des Sciences 16, 1054 (1843,

in French).

• C. W. Misner, K. S. Thorne and J. A. Wheeler, Gravitation (W. H. Feeman

and Co., 1973).

• S. Newcomb, United States. Nautical Almanac Oce. Astronomical pa-

per 1(6), 363 (1882).

• R. S. Park et. al., The Astronomical Journal. 153 (3), 121 (2017).

• M. P. Price and W. F. Rush, American Journal of Physics 47(6), 531

(1979).

11
• D. R. Williams, https://nssdc.gsfc.nasa.gov/planetary/factsheet/ (2019
I).

• D. R. Williams, https://nssdc.gsfc.nasa.gov/planetary/factsheet/sunfact.html
(2018).

• D. R. Williams, https://nssdc.gsfc.nasa.gov/planetary/factsheet/planet_table_ratio.html
(2019 II)

12