Foundations of Physics, Vol. 1, No.

1, 1970

The Reinterpretation of
Wave Mechanics
Louis de Broglie
Institut de France
Acad6mie des Sciences, Paris, France

Received July 15, 1969

The author begins by recalling how he was led in 1923-24 to the ideas of wave mechanics
in generalizing the ideas of Einstein's theory of light quanta. He made himself at that
time a concrete physical picture of the coexistence of waves and particles and, in 1927,
attempted to give them precise form in his "theory of the double solution." As other ideas
prevailed at the time, he abandoned the development of his conception. But for the past
twenty years, once again convinced, like Einstein, that present-day quantum mechanics
is only a statistical theory and does not give a true picture of physical reality, he has'
again taken up his oM ideas and developed them considerably. He has in particular
introduced an element of randomness into the theory and has thus attained to a "hidden
thermodynamics of particles," the results of which appear to be very interesting.

When I conceived the first basic ideas of wave mechanics in 1923-24, m [ was guided
by the aim to perform a real physical synthesis, valid for all particles, of the coexistence
of the wave and of the corpuscular aspects that Einstein had introduced for photons
in his theory of light quanta in 1905. I did not have any doubts at that time
about the physical reality of the wave and the localization of the particle in the
wave.
At that time, one remark made a deep impression on me. The phase of the
plane monochromatic wave, written as
6 Louis de Broglie

permits the definition of a 4-vector "wave" having components u/c, owl V, fly~ V, ~,v/V;
and this shows that the frequency of the wave is transformed, when we pass from the
proper system where the frequency has a value v0 to a reference system moving with
respect to the proper system with a velocity/3e, by the formula v == v0/(1 -- p~)a/~,
that is, it transforms like an energy. But, and this is remarkable, the frequency of a
clock is transformed according to the different formula v = v0(1 --/3~) 1/2, as results
from the relativistic theory of the retardation of clocks in motion.
I then noticed that it was possible to establish a relation between the 4-vector
defined by the gradient of the phase of a monochromatic wave and the momentum-
energy 4-vector by writing:

r v = h~, p ~ h/1 (1)

(h is Planck's constant), W and p being the energy of the particle and its momentum in
the direction of wave propagation. I was thus induced to imagine that the particle,
localized in one point of the plane monochromatic wave, possessed an energy W and
a momentum p and that it described one of the rectilinear rays of the plane wave.
But I also noticed that, if the particle is considered as containing, at rest, an
internal energy Mo e2 = hvo, it could be compared to a small clock of proper frequency
v0, so that, when it is in motion with a velocity/3e, its frequency is different from that
of the wave, namely, v = v0(1 -- /~2)1/2. I thus easily demonstrated that, during
the motion of the particle in the wave, the internal vibration of the particle was
constantly in phase with that of the wave, and this seemed natural if it is considered as
a local accident incorporated in the wave.
Now, in relativistic thermodynamics, it is generally accepted, since the
classical work of Planck and Lane, that the formula of heat transformation is
Q = Qo(1 -/3~) 1/2. Although this formula has recently been challenged by several
authors, I am convinced today after intensive thought m) that it is exact. We see
therefore that the difference between the relativistic transformation formulas for
energy W = W0/(1 --/3~) 1/2 and for heat Q = Q0(1 -/3~) ~/z is totally analogous
to the difference, that had impressed me so much formerly, between the formula of
transformation of the frequency of a wave v = v0/(1 -- ¢32)~/zand that of the frequency
of a clock v = v0(1 --/32) 1/z. This observation reveals the existence of a very close
link between relativistic thermodynamics and the physical ideas at the origin of the
discovery of wave mechanics.
But the account which I had elaborated in my thesis had the disadvantage of being
only applicable to the particular case of the plane monochromatic wave, which is
never completely realized because of the inevitable existence of a spectral width.
Sometime after my thesis, I was induced to go further and to generalize the ideas
that had guided me in this work, on the one hand, by considering the case of a wave that
is not plane monochromatic and, on the other hand, by distinguishing between the real
physical wave of m y theory and the fictitious wave ~b,arbitrarily normalized, introduced
by SchrSdinger and interpreted by Born as having a purely statistical significance.
That is how I was led to expound, in 1927 in an article C~)entitled "The wave mechanics
and the atomic structure of matter and radiation," a new interpretation of wave
The Reinterpretation of Wave Mechanics 7

mechanics and to generalize, for any given wave, the law of motion of the particle
that I had considered in the particular case of the plane monochromatic wave.
I do not want to develop here in detail the theory of the double solution as it
stands now. I refer those who wish to study it thoroughly to the accounts published
on this subject since I reconsidered, after having forsaken them for a long time, the
ideas originally outlined in my 1927 article. (3-7)
I shall begin by indicating the two main ideas on which this theory rests.
!. The wave, which, in my view, must be a physical wave of very small amplitude
which can evidently not be arbitrarily normalized, has to be distinct from the wave
~b, normalized in accordance with its statistical significance, in the usual formalism
of quantum mechanics. I designate the physical wave by v and I link the wave ~bto the
wave v by the relation ¢ = Cv, where Cis a normalization factor so that J'[ ~b [Sdr = 1.
It is this distinction, an essential one in my opinion, between the two solutions v and ¢
of the wave equation that had caused me to name this theory the "theory of the double
solution?' For a more thorough study of this question, I refer to the publications
mentioned above.
2. For me, the particle, always localized in space in the course of time, constitutes
in the wave v a very small region of high-energy concentration that can be represented
in first approximation as a kind of moving singularity. If the complex solution of the
wave equation that represents the wave v (or, if one wishes, the wave ~b, which amounts
to the same thing because of the relation ¢ = Cv) is written in the form

v = a(x, y, z, t) e (i/n~(~u,~m h = h/27r (2)

where a and ~v are real functions, the introduction of this expression in the wave
equation, followed by separation of real and the imaginary parts, leads us to conclude
that the movement of the particle in its wave should be described by the

W = O~o/~t, p = --grad cp (3)

which, in the case of the plane monochromatic wave, on writing

q~ = hv{t - - [(~x ÷ fly + ),z)/V]}

enables us easily to recover formulas (1). Designating now by M0 the proper mass of
the particle, we write
w = Moc~l(1 --/~)~/~, p = Mov/(1 --/3~)~/~ (4)

for a particle in motion with the velocity v ---- fie. The formulas thus give us

c2p _ c2 grad ~ (5)

v: W 8~lOt

We can call this formula, which determines the motion of the particle at each point
of its trajectory in the wave, "the guidance formula" of the particle by its wave. It
is easily generalized if the particle is subjected to an external field.
8 Louis de Broglie

The equations of the theory then easily show that the proper mass M 0 that
appears in the equations (4) is not equal to the usual mass m0 of the particle. It
is found to be given by Mo =-mo + (qo/cZ), where qo corresponds in the proper
system of the particle to a growth of its mass, which we will soon be led to associate
with the increase of an internal heat hidden in the particle.
In the case where the wave propagation is described by the relativistic Klein-
Gordon equation, one finds
h ~ Da~ 1/2 h
M ° = m ° 2 + -~
c a ] ' h = 2--~- (6)

This permits one to calculate M0 at each point and at each moment. In the Newtonian
approximation represented by the Schr/Sdinger equation, we have
h 2 Aa (7)
q = q0 -- 2m0 a
This is the "quantum potential" in the theory of the nonrelativistic double solution.
It is easy to extend the guidance theory to the case of an electron, which obeys
Dirac's equations, and to the photon, which follows Maxwell's equations augmented
by some very small mass terms/7)
When the particle moves in its wave following the guidance law and the wave
is not plane monochromatic, the proper mass M0 changes constantly in a way that
is measurable if one knows the shape of the wave: Therefore, it obeys the dynamics
of a body of variable proper mass. Now, when one carefully studies relativistic
thermodynamics, one finds that it is intimately linked to this type of dynamics. We can
therefore already surmise that the theory of the double solution should naturally
lead to the introduction of some thermodynamic considerations in wave mechanics.
We will see this idea becoming more precise in what follows.
Before discussing the hidden thermodynamics of particles, I must point out that
there are ways of confirming the exactness of the guidance formula. In one of my
books 13) (pp. 101,287), I have shown that if there exists in a wave a very small region
where the wave amplitude grows very rapidly, this small region must remain confined
inside a very small tube, limited by some guidance trajectories. This seems to justify
the guidance formula. Moreover, in some interesting recent work, Mumm
Thiounn Cs)has shown that all the equations of ordinary wave mechanics (Schr6dinger,
Klein-Gordon, Dirac, Maxwell) admit of solutions of singularities moving in the
course of time in accordance with the guidance law, the singularity giving here a kind
of schematic representation of the particle.
I now turn to the concepts which I have developed after 1960 under the name of
"thermodynamics of the isolated particle" or "hidden thermodynamics of particles."
Here again, I will limit myself to a summary of the main results, referring to my
principal works on this subjecV 9.m for a more detailed exposition.
Let us first return to the idea developed and adopted in my Doctoral thesis,
according to which a particle rest mass Mo can be associated with a small clock having
an internal vibration equal to Moe2/h. According to the relativistic formula of the
retardation of clocks in motion, an observer who sees the particle moving in its wave
The Reinterpretation of W a v e M e c h a n i c s 9

with a velocity fie assumes that it has an internal frequency v = v0(1 --/35)t/5. This
allows us to demonstrate easily that, even in the general case of a wave which is not
plane monochromatic, the internal vibration of the particle remains in phase with
that of the wave which is carrying it. This result, to which we shall return in a moment,
includes as a particular case the one which had been obtained for the plane mono-
chromatic wave, and it can be considered as the essential result of the guidance
formula.
But we have already noticed the analogy between the formulas W = W0/(1 _/32)i/2
and v -----v0/(1 --/32)1/2, where v is the frequency of the wave, on the one hand, and the
formulas Q = Q0(1 - t32)1/2 and v = Vo(1 --/32) 1/2, on the other hand, where v is the
frequency o f a clock. In the same way as shown previously (see especially ref. 11), this
analogy leads one naturally to regard the particle as a very small body containing
a hidden heat equal to Q0 = Mo c2, so that, for an observer who sees the particle
moving at a velocity tic, it contains an internal heat Q = M0d(1 -- fi2)1/2. Now,
when one studies relativistic wave mechanics, one is led to write the relation

Q0 /32)1/2
(1 - - fi2)1/2 -- Q0(1 - - -I- v" p (8)

which asserts that the total energy of the small warm body in motion is equal to the
sum of the internal heat it contains and its total translation energy, equal to v • p.
I have studied this question at length in the article m) to which I have just referred.
I f one now accepts the relation Q0 = M0 c2, the relation (8) can be written as

Moc 2
M0cZ(1 /32)1/2 = (1 --/32) ~/2 v 'p (9)

which can be verified immediately, since

p = Mov/(1 --/35)1/2

We will show that the formula (9) expresses the phase agreement between the
particle and its wave. Indeed, the guidance theory has taught us that, if ~ is the phase
of the wave written as ae i~/n, we have

8q~ 3,1oc2 Mov (lO)

Ot (1 -- ~2)1/2 ' -- grad 5o -- (1 -- fi2)1/2

so that, according to (9), one can write

Moc2(1 _/95)1/2 = 0~o

et + v • grad cp = ~d~p
- (11)

I f the particle conforms to a clock of internal proper frequency Moc2/h, the phase of
this internal vibration written in the form a~e~*m will be equal to ~oi = hvo(1 -/32)1/2t =
Moc2(1 -- fi~)l/2t, and we have, according to (11),

d(~o -- ~ ) = 0 (12)
10 Louis de Broglie

Since in the system where the particle is motionless, we have ~o = % , it is evident

that q~ = ~ continually during the movement of the particle.
The principle of the guidance theory can thus be found contained in the formula
(8) of relativistic thermodynamics, and this is very remarkable.
We know that, for a particle controlled in its wave by the guidance movement,
the proper mass Mo varies in general because of the variation of the wave amplitude
along the trajectory, as shown in formula (7). The dynamics of the particle is thus
a dynamics of variable rest mass, and one can see then that the particle is subjected to
the action of a force due to the variation of its rest mass. This force is the
one represented in the formalism of the double solution by the intervention of a
quantum potential. The coherence of all these conceptions is thus clearly shown.
It is now appropriate to reason as follows. Relativistic dynamics teaches us that
Lagrange's function of a free particle of rest mass Mo in motion with a velocity t% is
L = --Moc2(1 - - ~ 2 ) 1 ] 2 and that the action integral is

f Ldt= --f M0e~(1- fl2)1/2d/= --f M0e2d$ (13)

ds being the proper time of the particle. It is therefore tempting to establish a relation
between the two fundamental relativistic invariants, action and entropy. But in order
to be able to do so, we have to give a well-determined value to the action integral by
choosing properly the time interval over which the integration extends. From our point
of view, it seems natural to choose as integration interval the period T of the
particle of normal proper mass mo in the reference system where it is moving with
a velocity fie. Since 1/T = m0c~(1 --/3~)1/~, we thus define the cyclical action integral as
T
A = --f Mock(1 -- flz)l/z dt 04)
0

Since the period T is very small, it is natural to suppose that Mo and fi remain
essentially constant over the time of the integration, which allows us to write
A ---- --Moc2/moc ~ (15)
and, in order to define the entropy S of the state of the particle, one is led to write
S/lc ~- A/h (16)
where k and h are respectively the Boltzmann and Planck constants. Since Q0 = Mo c2,
we deduce from (16) the formula
3S = --k 3Qo/mo e~ (17)
We have thus succeeded in attributing a given entropy to the movement of the particle
in its wave, and eventually a certain probability P defined by the famous Boltzmann
formula P = e s/~. We shall explain later the origin of the minus sign on the right-hand
side of the relation (17).
I felt one could draw from these considerations two conclusions that seemed
important for the interpretation of quantum physics:
The Reinterpretation of Wave Mechanics 11

1. The principle of least action is but a particular case of the second law of
thermodynamics.
2. The privileged role, whose paradoxical character has been underlined by
Schrrdinger, that present quantum mechanics attributes to plane monochromatic
waves and to stationary states of quantified systems can be explained by the fact that
they correspond to entropy maxima, not because the other states are nonexistent,
but only because they are of a lesser probability.
As far as the second of these conclusions is concerned, I refer to the demonstration
outlined in one of my works, t9~ But, in view of the great interest which attaches to the
identification of the principle of least action with the second law of thermodynamics,
I will summarize the demonstration previously offered.
Hamilton's principle of least action tells us that if a particle, in its natural motion
according to classical dynamics, leaves a point A at the instant t o to arrive at a point B
at fi, then the action integral taken along this motion is a minimum compared with
the same integral taken along all other possible motions that would lead the particle
from the point A at time t0 to point B at instant tl • We are thus led to write

sll[3L]/o dt = O;
to
[g2L]/o dt > 0 (18)

both variations being taken while maintaining the rest mass 3/o constant and equal to
its normal value mo•
I have introduced here a hypothesis which to me seems to have a very interesting
significance The curve A C B (Fig. 1) represents the natural trajectory. But I have
supposed that the alternate trajectories, such as AC'B, do not correspond, as is usually
assumed, to fictitious motions imagined by the theorist, but to movements that
can really occur when the rest mass Mo of the particle undergoes a succession o f
fluctuations between to and q , drawing it away momentarily from its normal value
m0 • Thus, the alternate trajectory A C ' B must, according to Hamilton's principle, be
determined by the equation

3(L 4- 3L) dt = (3L 4- 8~L) dt = 0 (19)

to

But, as the rest mass is not supposed to be constant any longer, one must write

3L = [SL]Mo + 3MoL; 82L = [82L]/o 4- 8~oL (20)

El

Fig. I
12 Louis de Broglie

where 8~oL represents all of the terms in 32L which depend on the variation of M 0 .
Therefore, we have on A C ' B

f~o {[SL]~,-° q- 8~oL + [82LlMo + g~°L} dt = 0 (21)

But the integral of the first term is zero by virtue of Hamilton's principle:, and it is
easy to verify that the fourth is negligible compared to the others. Finally, there
remains:

(22)
to ~o

being the time average of 3~r0L between to and t 1 . The formulas previously accepted
lead us to suppose that --3MoL represents the heat received by the particle, and
formula (22) shows that the temporal mean of the quantity of heat is zero on the
natural trajectory, while it is positive on the "hypothetical" trajectory. Thus, when
the minus sign in (17) is taken into account, the formula (17) shows that the
entropy S diminishes on the average when one goes from A C B to A C ' B . On the
natural trajectory the entropy is therefore maximal relative to the fluctuations
subject to the conditions of Hamiltonian variation. The natural trajectory is
therefore more likely than the other trajectories. Thus, in the framework of our
concepts, there appears to be a very curious link between the principle of least action
and the second law of thermodynamics?
We arrive now at another very important point. The thermodynamic conception
of the particle just outlined leads us to think that even when it seems to us that a
particle is isolated from all macroscopic bodies capable of exchanging heat with the
particle, it is constantly in thermal contact with a kind of thermostat hidden in what
we call the vacuum. When a particle, or a set or particles, is in contact with a thermostat
of temperature T, we know from the work of Boltzmann and Gibbs that the proba-
bility of its energy having value E is Poe -E/kr. In this expression, P0 is often called the
"a priori probability," and we say that it is the probability of the state of the particle
or particles under consideration in the absence of all contact with a macroscopic
thermostat. It seems to me that we must identify this a priori probability with the one
defined above, since, even though it seems isolated, every particle is in contact with a
hidden thermostat.
Any attempt to establish the exact nature of this hidden thermostat seems
premature, but it appears related to the "subquantum level" proposed by Bohm and
Vigier fifteen years ago, (1~) or at least to a part of this subquantum level.
During its guidance movement, the mass M0 of the particle generally varies. We
must interpret this phenomenon by saying that it exchanges heat with the hidden
thermostat. The heat exchanges are linked to the variations of the quantum potential,
that is, to the variations of the wave amplitude at the point where the particle is found;

On the question of kinetic focus, see ref. 12.

T h e Reinterpretation o f W a v e M e c h a n i c s 13

one sees that the wave acts as an intermediary between the particle and the hidden
thermostat.
It is normal to suppose that a particle is a very simple system, and because of this
simplicity it is preferable not to attribute to it a proper temperature and entropy.
The hidden thermostat, on the contrary, whatever its real nature, must be a very
complex system which permits us to attribute an entropy and an apparent temperature
to the particle. The entropy appearing in formula (17), which determines the probability
of the state of the particle, is therefore the entropy of the thermostat, while the quantity
~Qo = ~Mo c2 is (in the proper system of the particle) the quantity of heat it receives
from the thermostat when M0 grows or that which it yields to it when Mo decreases.
With this understanding, one can therefore write in the reference system where the
particle has a velocity fie,

8S = - - ~Qo(1 - fl2)1/2 _ ~Q = - k-6Q° (23)

ToO -- 32) 1/2 T moc 2

provided we recall the relativistic formula for the temperature transition

T =- To(1 -- fi2)1/2

and suppose that To = mod/k. Thus, we find formula (17) again, and the presence of
the minus sign on the right-hand side is now explained.
It might seem strange that the apparent temperature T of the thermostat for the
particle depends on the proper mass m0 of the particle and differs according to the
nature of' the latter. But as remarked above, it is by the intervention of its wave that
the particle is in thermal contact with the hidden thermostat. This remark seems to
give meaning to the fact that, for each particle, in each point of its trajectory, the
apparent temperature of the thermostat could, perhaps by means of some resonance
effect, depend on the local frequency, which is itself a function of the rest mass.
A more detailed description of the hidden thermostat might some day permit further
clarification of this point.
But a very important point has still to be examined. The great progress
accomplished in thermodynamics, when themolecular structure of matter and statistical
mechanics were introduced, suggested that when a body is in a stable thermodynamic
state, it is nonetheless constantly subject to small fluctuations of zero average around
this state. This made it possible to develop the theory of fluctuations and of Brownian
motion. We must expect to encounter some analogous circumstances in the description
of the particle motion in terms of our hidden thermodynamics. Without attempting
to study the question in depth, we shall limit ourselves to two aspects of it.
First, we have seen that, in the regular guidance movement, some heat exchanges
occur between the particles and the hidden thermostat through the quantum potential
that can be defined by the formula q = Moc 2 - - m o c2. But the fluctuation theory
leads us to assume that the wave amplitude also must undergo constant fluctuations,
giving rise to a fluctuating quantum potential ql of zero average. We must then write

Mo c2 = mo c2 + q + qf (24)
14 Louis de Broglie

and, as ~ is zero, we have, on the average,

-/~oc2 ---- moc ~ .-? q (25)

If the potential q is zero, this simply gives us

Mo ---- mo (26)

Therefore, we see that, because of the absence of a nonfluctuating quantum potential,

the normal proper mass rn0 of the particle can then be considered as the mean value of
the fluctuating proper mass.
Here is another important point. I have shown in my previous publications that,
in order to justify the well-established fact that the expression I~(x, y, z, t)t2dr
gives, at least with Schr6dinger's equation, the probability for the presence of the
particle in the element of volume d~- at the instant t, it is necessary that the particle
jump continually from one guidance trajectory to another, as a result of the continual
perturbation coming from the subquantal milieu. The guidance trajectories would
really be followed only if the particle were not undergoing continual perturbations due
to its random heat exchanges with the hidden thermostat. In other words, a Brownian
motion is superposed on the guidance movement. A simple comparison will make this
clearer. A granule placed on the surface of a liquid is caught by the general movement
of the latter. If the granule is heavy enough not to feel the action of individual shocks
received from the invisible molecules of the fluid, it will follow one of the hydro-
dynamic streamlines. If the granule is a particle, the assembly of the molecules of the
fluid is comparable to the hidden thermostat of our theory, and the streamline
described by the particle is its guiding trajectory. But if the granule is sufficiently
light, its movement will be continually perturbed by the individual random impacts
of the molecules of the fluid. Thus, the granule will have, besides a regular movement
along one of the streamlines of the global flow of the fluid, a Brownian movement
which will make it pass from one streamline to another. One can represent Brownian
movement approximately by a diffusion equation of the form ~la/~t = D A o, and
it is interesting to seek, as various authors have done recently, the value of the
coefficient D in the case of the Schr/Sdinger equation corresponding to the Brownian
movement.
I have recently studieda4~ the same question starting from the idea that, even
during the periods of random perturbations, the internal phase of the particle remains
equal to that of the wave. I have found the value D = (27r/3)li/m, which differs only
by a numerical coefficient from the one found by other authors.
This concludes the account of my present ideas on the reinterpretation of wave
mechanics with the help of images which had guided me in my early work. My
collaborators and I are working actively to develop these ideas in various directions.
Today, I am convinced that the conceptions developed in the present article, when
suitably developed and corrected at certain points, may in the future provide a real
physical interpretation of present quantum mechanics.
The Reinterpretation of W a v e Mechanics 15

REFERENCES

1. L. de Broglie, Compt. Rend. Acad. Sci. Paris 177, 506, 548, 630 (1923).
2. L. de Broglie, J. Physique (serie 6) VIII (5), 225 (1927). Th6se de Doctorat (Masson, Paris,
1924).
3. L. de Broglie, Une interprdtation causale et non linOaire de la MOcanique ondulatoire: la thkorie
de la double solution (Gauthier-Villars, Paris, 1956) [English translation: Elsevier, Amsterdam,
1960].
4. L. de Broglie, La th~orie de la Mesure en M~canique ondulatoire (Gauthier-Villars, Paris,
1957).
5. L. de Broglie, J. Physique 20, 963 (1959).
6. L. de Broglie, Etude critique des bases de l'interpr~tation actuelle de la M~canique ondulatoire
(Gauthier-Villars, Paris, 1963) [English translation: Elsevier, Amsterdam, 1964].
7. L. de Broglie, Ondes ~lectromagn~tiques et Photons (Gauthier-Villars, Paris, 1968).
8. Mumm Thiounn, Cahiers Phys. 174, 53 (1965); Compt. Rend. Acad. Sci. Paris B262, 657 (1966);
Port. Phys. 657 (1966).
9. L. de Broglie, La Thermodynamique de la particule isol~e (ou Thermodynamique each~e des
particules) (Gauthier-Villars, Paris, 1964).
10. L. de Broglie, La Thermodynamique ¢ach6e des particules, Ann. Inst. Henri Poincar~ I (1),
1-19 (1964).
l l. L. de Broglie, Thermodynamique relativiste et M6canique ondulatoire, Ann. Inst. Henri
Poinear~ IX (2), 89-108 (1968).
12. L. de Broglie, Compt. Rend. Aead. Sci. Paris 257, 1822 (1963).
13. D. Bohm and J. P. Vigier, Phys. Rer. 96, 208 (1954).
14. L. de Broglie, Compt. Rend. Acad. Sci. Paris 264, 1041 (1967).