The Historians’ Disagreements over the

Meaning of Planck’s Quantum

 O D  *

During the past twenty years, historians have disagreed over the meaning of
the quanta which Max Planck introduced in his black-body theory of 1900.
The source of this confusion is the publication, in 1978, of Thomas Kuhn’s
iconoclastic thesis that Planck did not mean his energy quanta to express a
quantum discontinuity (Kuhn, 1978). The aim of the present essay is a com-
parison of the opinions of various historians on this issue.1
Whether or not Planck introduced a quantum discontinuity is an import-
ant question, for it affects our understanding of the origins of the quantum
theory of Niels Bohr and Arnold Sommerfeld. Yet the focus on this question
could bring with it a few misconceptions, which should be cleared out from
the start. Firstly, one should not infer that the true meaning of the energy
quanta was a central issue for Planck himself around 1900: what Planck
emphasized was the introduction of two universal constants, h and k, and
their power to bridge gas theory and radiation theory.2 Secondly, one should
not assume that historians of Planck’s radiation theory completely disagree
with each other: a consensus exists on many important features of Planck’s
program. Lastly, my focus on the intricacies of Planck’s thoughts should not
be mistaken as an indication that quantum theory was started in an exclus-
ively individual manner. On the contrary, most early quantum concepts
emerged through the confrontation and combination of the views of several
physicists.3

*CNRS: REHSEIS, Paris, France

E-mail: darrigol/paris7.jussieu.fr

C 2001: V. 43: . 219–239

C Munksgaard 2001. Centaurus ISSN 0008-8994. Printed in Denmark. All rights reserved.
220 O. Darrigol

The first section of this paper gives a classification of the various attitudes
historians have taken towards Planck’s quanta. In the second section, a corre-
lation is established between the attitudes and the arguments used to support
them. Next, the validity of these arguments is shown to depend on the his-
torian’s interest and methodology. In particular, it turns out that the his-
torians who see in Planck the discoverer of quantum discontinuity relied not
only on Planck’s publications but also on the ways other physicists perceived
and used his work. A last section is devoted to the consequences of a rigorous
attention to the development of Planck’s program in Planck’s own under-
standing. Although Kuhn’s thesis about the quantum discontinuity is thus
corroborated, his overall account turns out to be seriously flawed. The mean-
ing of the steps taken by Planck in 1900 must instead be based on Allan
Needell’s profound study of Planck’s attitude toward thermodynamic irre-
versibility (Needell, 1980).
As a preliminary, it may help to recall aspects of Planck’s radiation theory
on which most historians would agree. In 1895, Planck started an ambitious
program in which he hoped to provide an explanation of thermodynamic
irreversibility based on electrodynamic processes. He was not satisfied with
Boltzmann’s similar attempt based on gas theory, for it led to a statistical
understanding of irreversibility and thereby contradicted Planck’s belief in
the absolute validity of the entropy law. Besides, Planck knew from Gustav
Kirchhoff that thermodynamic considerations could be applied to radiation
at thermal equilibrium – the so-called black-body radiation – to derive the
universality of its spectrum. Planck hoped that a simple electrodynamic
model of the thermalization of radiation would lead to this fundamental
spectrum, in which Berlin experimenters were increasingly interested for
metrological reasons. The model he selected was a set of electric resonators
(one for each frequency) enclosed in a cavity with perfectly reflecting walls.4
After a humbling exchange with Ludwig Boltzmann, Planck admitted that
he could not obtain an irreversible evolution of his system without the ad hoc
assumption of ‘‘natural radiation,’’ a formal counterpart of the assumption
of molecular chaos which Boltzmann had introduced in order to justify his
own irreversibility theorem (the H-theorem of 1872). Thanks to this assump-
tion, the uncontrollable, irregular aspects of the evolution of the system were
eliminated to yield a deterministic, irreversible evolution. In the case of equi-
librium, Planck established the relation unΩ(8pn2/c3)U between the spectral
density un of thermal radiation and the average energy U of a resonator at
 Planck’s Quantum 221

the frequency n. His proof of irreversibility was based on the introduction of

a monotonous entropy function depending on the observable properties of
the resonators and the surrounding radiation.5
In 1899, Planck believed that the choice of this entropy function was
uniquely determined by his irreversibility theorem. From the entropy S(U)
of a resonator, he could derive its time-averaged energy U(T) through the
thermodynamic relation dS/dUΩ1/T (T being the absolute temperature). The
black-body spectrum then followed from the above relation between un and
U. Planck thus retrieved the exponential law which Wilhelm Wien had al-
ready suggested and which experiments seemed to confirm. A few months
later, however, precise measurements of the infra-red tail of the spectrum
exhibited violations of Wien’s law.6 Planck then recognized that an infinite
number of choices of the function S(U), and therefore an infinite number of
black-body laws were in fact compatible with his irreversibility theorem.
In the Fall of 1900, Planck obtained a new black-body law by interpolating
the second derivative d2S/dU2 – which played a central role in the derivation of
his irreversibility theorem – between the values it takes for Wien’s law (at large
frequencies) and for the measured low-frequency part of the spectrum (Planck,
1900a). The Berlin experimenters immediately confirmed this proposal. On
December 14th, Planck presented a more fundamental proof of his new law at
a meeting of the German Physical Society [Planck, 1900b]. Since the dynamics
of his model had failed to provide the function S(U), he appealed to Boltz-
mann’s relation between entropy and probability: SΩk ln W in Planck’s guise.
He determined W by counting the number of discrete energy distributions over
a set of identical resonators, in analogy with a prescription Boltzmann had
given for a gas of molecules in 1877. Unlike Boltzmann, Planck gave to the
discretizing energy-elements a completely determined value hn, where h is a uni-
versal constant with the dimension of action. Only thanks to this ‘‘most essen-
tial point’’ (ibid. p. 700) could the new black-body law be retrieved.

1. Histories

Statements on the meaning of Planck’s energy quanta fall into three cate-
gories. According to the first category, in 1900 Planck introduced the idea
that some microphysical entities (his resonators) could only have discrete
energy values. According to the second, Planck did not know himself what
222 O. Darrigol

the precise meaning of the energy quanta was. According to the third, Planck
still believed that the energy of his resonators varied continuously, and he
had no intention to revolutionize the laws of dynamics. Table 1 indicates the
position of various historians in this scheme.

Discontinuity Indetermination Continuity

Textbooks Kangro 1970 Kuhn 1978

Klein 1961–7 Needell 1980, 1988
Hund 1967 Galison 1981
Jost 1995

Rosenfeld 1936 Planck 1920, 1943, 1948

Jammer 1966 Darrigol 1988, 1992

Table 1: Statements on Planck’s quantum

The discontinuity category is the one most abundantly represented. The

typical textbook story of Planck’s discovery belongs to it, starting with the
following extract of Sommerfeld’s Atombau und Spektrallinien:
‘‘In several years of persistent efforts, Planck strove to penetrate electrody-
namics with the principles of thermodynamics. In order to thereby remain in
agreement with experiments, he was finally forced to a daring challenge of the
received conceptions of the wave theory, to his hypothesis of energy quanta: he
required that the radiation energy of every frequency n could only be emitted
or absorbed by whole multiples of the elementary energy quantum eΩhn.’’
Accordingly, Sommerfeld made December 14th, 1900 the ‘‘birthday of the
quantum theory’’ (Sommerfeld 1919, p. 44). Among historians, the most
powerful discontinuist is Martin Klein. In papers written in the 1960s, Klein
discussed Planck’s program, his relation to Boltzmann, and his lack of con-
cern for the Rayleigh-Jeans law. Although he did not dwell on the question
of quantum discontinuity, he clearly took the discontinuist interpretation for
granted. In 1962, comparing Einstein’s and Planck’s quantum considerations,
he wrote: ‘‘Planck had quantized only the energy of the material oscillators
and not the radiation’’ (Klein, 1962, p. 476). In 1966, he spoke of a ‘‘strict
quantization of the oscillator’’ (Klein, 1966, p. 28). Most unambiguously, in
his biography of Paul Ehrenfest he stated: ‘‘Planck had achieved his immedi-
ate goal. He had derived a radiation distribution law. But in order to do so
he had taken a fateful step by requiring the energy of one of his oscillators
 Planck’s Quantum 223

always to be a multiple of an energy unit’’ (Klein, 1970, p. 228). Klein admit-

ted that Planck was not fully aware of the revolutionary character of this
step: ‘‘A revolutionary idea,’’ he noted, ‘‘is not always recognized as such, not
even by its propounder’’ (Klein, 1962, p. 476). But the quantum step had
been taken anyway. After the publication of Kuhn’s contrary thesis, Klein
maintained that ‘‘Planck’s position in 1900...required discrete energy levels’’
(Klein, 1979, p. 430).
Physicist-historians have usually agreed with Klein’s view. For example, in
his Geschichte of 1967, Friedrich Hund wrote ‘‘Planck’s hypothesis consisted
in distributing energy quanta on material oscillators. A somewhat different
emphasis [Betonung] gave the statement: harmonic oscillators have only the
discrete energy EΩhnn’’ (Hund, 1967, p. 25). Hund seems to have hesitated
in identifying the distribution of quanta over resonators (for the purpose of
an entropy calculation) with energy quantization. Yet he made Planck a
quantizer by letting the Betonung float in the air. Most recently, in a violent
reaction to Kuhn’s book on quantum discontinuity, Res Jost returned to
Planck’s text of December 1900 and pronounced his ‘‘verdict against Kuhn’’:
‘‘The resonators of the frequency n can only take energy in elements eΩhn’’
(Jost, 1995, p. 72).
Other historians have been more nuanced discontinuists. In his old, in-
sightful history of early quantum theory, Léon Rosenfeld insisted that
Planck’s definition of complexions, being based on finite energy-elements,
contradicted classical statistical mechanics. But he noted that the new notion
of ‘energy element’ entered Planck’s statistical deduction of the black-body
law ‘‘in a still rather obscure manner’’ (Rosenfeld, 1936, p. 166). Max Jamm-
er, in his Conceptual foundations of 1966, similarly nuanced the account of
Planck’s quantum step: ‘‘Nowhere ... did Planck bring into prominence the
fundamental fact that U is an integral multiple of hn. At that time Planck
was apparently not yet quite sure whether his introduction of h was merely a
mathematical device or whether it expressed a fundamental innovation of
profound physical significance’’ (Jammer, 1966, p. 22).7
This statement brings us close to the indeterminist historians, for whom
Planck did not really know what he was doing, so that one should not read
too much in his formal deductions. A fist example of this prudent attitude is
found in Hans Kangro’s Vorgeschichte of 1970: ‘‘At that time [1900], Planck
cannot have been aware of the bearing in the system of physics of the assump-
tion of energy elements – which also exist in Boltzmann’’ (Kangro, 1970, pp.
224 O. Darrigol

225–226). Kangro here alluded to the analogy between Planck’s and Boltz-
mann’s combinatorics: what Boltzmann did not intend to be revolutionary
could not be either in Planck’s tentative transposition of Boltzmann’s reason-
ing. In his dissertation and in later writings, Allan Needell offered other
reasons to leave the meaning of Planck’s energy quanta in the dark: ‘‘Planck
was not concerned with describing details of the motion or behavior of an
oscillator.... The question of whether Planck introduced quantum discontinu-
ities in 1900 is not a major issue; the answer depends on the attitude toward
physical concepts and on the level of commitment one requires to credit
someone with such an introduction’’ (Needell, 1988, p. xix). Therein Needell
made a historical point, that Planck, in his program, deliberately left the
detailed behavior of his resonators undetermined; and a methodological
point, that historians should distinguish different degrees of commitment of
a given actor on a given issue. In a review of the Klein-Kuhn controversy
published in 1981, Peter Galison adopted a similar stance: ‘‘The question of
the continuum vs. discreteness as such, which for us is of such overwhelming
interest, was entirely peripheral to Planck’s other concerns.... It is not always
possible to impose a self-consistent, fully articulated set of beliefs on a scien-
tist’s view of his problem, especially at periods of great upheaval’’ (Galison,
1981, p. 82).8
The third kind of historian of Planck’s quantum is the continuist one, who
makes the energy of Planck’s resonators a continuous variable despite
Planck’s formal use of energy-elements. Thomas Kuhn inaugurated this trend
in 1978 with his book on Black-body theory. A typical statement of Kuhn’s
reads: ‘‘The concept of restricted resonator energy played no role in [Planck’s]
thought until after the Lectures [of 1906]’’ (Kuhn, 1978, p. 126). In 1988, I
have defended the same thesis on a different basis and with a twist: whereas
Kuhn made Planck a persistently classical physicist, I believe with Needell
that Planck left a non-classical behavior of his resonators open (Darrigol,
1988). Moreover, Needell has rightly noted that the word ‘‘classical’’ points
to a post-quantum-theoretical idealization of nineteenth-century physics that
does not apply to the multiple, open, evolving theories of the turn of the
century (Needell, 1988).
Interestingly, Planck’s own retrospective appraisals of his 1900 break-
through do not corroborate the discontinuist account. Instead they must be
situated somewhere in-between the indeterminist and the continuist cate-
gories. In his Nobel lecture of 1920, Planck wrote (Planck, 1920, p. 127):
 Planck’s Quantum 225

‘‘Whereas [the constant h] was indispensable – for only with its help could
the size of the...’elementary domains’ or Spielräume of the probability be
determined – it proved to block and resist every attempt at fitting it into the
frame of classical theory.... The failure of all attempts...soon left no doubt:
either the quantum of action was only a fictitious quantity, or the derivation
of the radiation law rested on a truly physical thought....Experiments have
decided in favor of the second alternative. But science does not owe the
prompt and indubitable character of this decision to tests of the law for the
energy distribution of thermal radiation, and even less to my special deri-
vation of this law; it owes that to the unceasing progress of the researchers
who have put the quantum of action to the service of their investigations. A.
Einstein made the first breakthrough in this domain.’’
Clearly, Planck did not regard the physical significance of energy quanta
as established by his own work on radiation. In his own view, the decisive
steps in this regard were Einstein’s and others’ applications of the quantum
to non-thermal phenomena such as the photoelectric effect, specific heats,
and atomic spectra. Planck never claimed responsibility for the revolution of
which a long historiographical tradition makes him the hero. ‘‘The introduc-
tion [of the constant h],’’ he pondered at the close of his life, ‘‘meant a much
more radical break from classical theory than I had initially suspected’’
(Planck, 1948, p. 397).

2. Arguments

The multiplicity of the interpretations of Planck’s quantum corresponds to a

diversity in the kind of arguments that sustain these interpretations. This is
indicated in Table 2.

1) Tradition
Hund, Jost
2) Planck’s proof of 1900
3) Statistical thermodynamics Klein
4) Contemporary readings
5) Planck’s asserted goals Kangro, Needell, Galison Kuhn, Darrigol
6) Formal heuristics

Table 2: Arguments
226 O. Darrigol

Those who see quantum discontinuity in Planck’s work tend to refer to

historiographical tradition. Hund (1967) implicitly cites Sommerfeld by mak-
ing December 14th, 1900, the birthday of the quantum theory. Jost (1995)
accuses Kuhn of neglecting the collective memory of physicists. Klein (1962),
as a sophisticated historian, is more critical of traditional accounts and
spends much time correcting errors regarding Planck’s awareness of the Ray-
leigh-Jeans law or his use of Boltzmann’s method. Yet his insistence on the
revisionist character of Kuhn’s thesis betrays a basic trust in those aspects of
the history of quantum history that were never questioned before Kuhn. His
review of Kuhn’s book portrays it as an abusive revision: ‘‘By insisting so
strongly on this radical revision of all previous accounts, Kuhn has put him-
self in a position that requires him to explain away part of the available
evidence and to look at the rest of it from the standpoint of his central thesis’’
(Klein, 1979, p. 430).
The discontinuist historians also share a reference to the supposed trans-
parency of Planck’s original proof of the black-body law. They usually cite
the following extract of Planck’s talk of December 14th (Planck, 1900b, pp.
700–701):
‘‘When E [the energy to be distributed over N resonators with the fre-
quency n] is regarded as an indefinitely divisible quantity, the distribution can
be done in an infinite number of ways. However, we regard E – this is the
most essential point of the whole calculation – as made up of a completely
determined number of identical finite parts; and for this purpose we use the
constant of nature hΩ6.55¿10ª27 erg¿sec. This constant, multiplied by the
common frequency n of the resonators, gives the energy-element, in erg, and
by division of E by e we get the number P of energy-elements which are to
be distributed over the N resonators.’’
Therefrom it seems obvious that the energy of a resonator can only be
an integral multiple of e, since it is obtained by the distribution of energy-
elements.
Interestingly, the next sentence of Planck’s text is almost never quoted. It
reads: ‘‘When the thus computed ratio [E/e] is not a whole number, one can
take for P a neighboring whole number’’ (Planck, 1900b, p. 701). This sen-
tence should be problematic for the discontinuist interpreter of Planck, for
it clearly indicates that Planck did not expect the energy of his set of res-
onators to be an integral multiple of e; a fortiori, he could not expect the
energy of an individual resonator to be quantized. But this is the kind of
 Planck’s Quantum 227

detail that is easily filtered out by a prejudiced reader. The one physicist-
historian who does quote the embarrassing sentence, Res Jost (1995), pro-
nounces his anti-Kuhn ‘‘verdict’’ right after it.9
Another argument in favor of the discontinuist interpretation appeals to
the fact that statistical mechanics, if properly applied to a classical radiat-
ing system, necessarily leads to the Rayleigh-Jeans law. In many textbook
accounts, Planck is said to have been aware of this fact and to have intro-
duced quantum discontinuity in order to avoid this absurd consequence of
the classical theory. Klein rightly rejected this view and suggested instead
that Planck was too unfamiliar with statistical mechanics to understand
the fact. But Klein still used his own knowledge of classical implications to
infer that Planck’s demonstration only made sense if the resonators were
quantized. More exactly, Klein assumed that in 1900 Planck adopted
Boltzmann’s relation between entropy and probability; this relation could
only lead to Planck’s law if the resonators were quantized, against Boltz-
mann’s original intentions and against the consequences of classical stat-
istical mechanics (which included the equipartition theorem).10
By using this kind of argument, Klein imitated early interpreters of
Planck’s paper such as Hendrik Lorentz, James Jeans, and Albert Einstein.
In his review of Kuhn’s book, he insisted that these early readings of
Planck revealed the essence of Planck’s theory: ‘‘The quanta were there in
[Planck’s] theory, and some of his readers did draw attention to them: Lor-
entz in 1903, Ehrenfest and James Jeans in 1905.... Lorentz wrote in 1903
about Planck’s use of ‘a certain number of finite portions’ of energy. I see
this as Lorentz’s recognition of what Planck had done in his papers of
1900 and 1901’’ (Klein, 1979, p. 431). In his paper of 1962, Klein already
used Jeans’ and Ehrenfest’s readings of Planck to characterize Planck’s de-
parture from Boltzmann’s methods. In particular, he agreed with Ehrenfest
that ‘‘Planck’s energy elements amounted to a radical change in the a
priori weight function introduced into phase space’’ (Klein, 1962, p. 475):
whereas Boltzmann assumed a uniform weight, Planck allowed only dis-
crete energy values.
Klein was also sensitive to the programmatic aspects of Planck’s radiation
theory. He rightly emphasized that the meaning of entropy was the leading
thread of Planck’s program, and gave as much importance to the evolution of
Planck’s concept of entropy as to the steps necessary to his derivations of the
black-body law (Klein, 1963). However, Klein did not pursue his exploration
228 O. Darrigol

of Planck’ program far enough to perceive contradictions with the received

view on Planck and quantum discontinuity. We will see in a moment that the
indeterminist and continuist readings of Planck resulted from a more sustained
attention to Planck’s expressed goals.
Lastly, a few historians of Planck’s theory have tried to reconstruct the
formal and symbolic operations through which Planck reached the various
steps of his radiation theory. Although some of that is found in Rosenfeld’s
and Klein’s pioneering studies, Kuhn and myself made the most intense ef-
forts to capture the detailed interweaving of programmatic goals and formal
procedures. We thus hoped to shed light on Planck’s theoretical style in gen-
eral, and on the nature of his energy-elements in particular.
To sum up, the continuist historians are those who have most closely
studied Planck’s intricate considerations from the beginning of his program
in 1894, through the quantum papers of 1900 and 1901, to later elaborations
between 1906 and 1914. The indeterminist historians have focused on
Planck’s expressed goals. The discontinuist ones have given more weight to
historiographical tradition, to the formal structure of Planck’s reasonings of
1900 and 1901, and also, in Klein’s case, to contemporary readings of
Planck’s texts.

3. Methods

This variety of arguments raise methodological questions on the assessment

of individual discovery:
1) Can a historian give weight to traditional accounts?
2) Can the formal skeleton of a demonstration speak for itself?
3) Can the historian of a discovery rely on contemporary views not necess-
arily held by the discoverer?
4) Can he rely on contemporary accounts?
5) Does he need to retrace the discoverer’s path?
6) Should he worry about formal, technical details of this path?
(Note that I here use the words ‘‘discovery’’ and ‘‘discoverer’’ in a conven-
tional manner, to point to the work and to the actor who have been tradition-
ally regarded as bringing some essential new knowledge). The answer to these
questions depends on the historian’s interest. Table 3 schematizes this effect
for three kinds of interest to be now described.
 Planck’s Quantum 229

Discoverer’s way Impact of discovery Credit

(1) No Yes Yes

(2) No Yes Yes
(3) No Yes May be
(4) No Yes May be
(5) Yes May be not Not really
(6) Yes May be not Not really

Table 3: Interests/arguments

First suppose that we are trying to decide what the discoverer, Planck,
thought he was doing in 1900. In this case, we need to handle traditional
accounts with systematic suspicion, for it is well-known that the simple narra-
tives generated and stabilized by a given community of scientists may have
little relation with the actual performance of the discoverer. We must also be
weary of trusting the appearances of the formal apparatus of past demon-
strations, for we may unconsciously interpret this apparatus in terms of later
knowledge that may be incompatible with the discoverer’s views. Similarly,
we should be alert to possible contradictions between these views and those
projected by contemporary actors on the discoverer’s work. We must of
course retrace the discoverer’s path, because it usually contains important
implicit aspects of his final reasoning. Formal, technical details of this path
may thereby help, for they have every chance of being intimately connected
with more qualitative elements. They are basic data to anyone interested in
the process of theory construction.
A historian may, however, be less interested in the idiosyncrasies of
Planck’s approach than in the way his work was used by other proto-quan-
tum-theorists. Then he would give significantly different answers to our meth-
odological questions. If used with sufficient care, traditional accounts may
be more helpful to him, because they often reveal features of the early recep-
tion of the discoverer’s work: the canonical story of a discovery (if there is
any) is usually formed little after its impact has been recognized, on the basis
of the most common perception of the discovery, which may differ from the
discoverer’s own perception. The early users of the discoverer’s work are in
fact often reinterpreting this work: they extract some elements of this work
and combine them with other elements and views that may be incompatible
with those of the discoverer. The purely formal aspects of the discoverer’s
work may then be essential, for they may be all that the users retain. So can
230 O. Darrigol

be the users’ views, for they condition the process of reinterpretation. In

contrast, the discoverer’s path becomes irrelevant in so far as it is ignored by
his followers.
Lastly, a physicist-historian could be interested in considerations of credit:
Is it legitimate to make Planck the father of the quantum theory? Then our
methodological questions would receive still different answers, because the
factors contributing to the attribution of credit are more diverse and more
inconsistent than those allowed in strict history. What the discoverer actually
intended to do is only one factor. The impact of his seminal work, including
useful reinterpretations or misinterpretations, may be more important. His
overall achievements, both scientific and institutional, may also count. Owing
to this complexity, the weight of tradition is considerable, and revisions are
extremely rare.
Consider, for instance, the case of J. J. Thomson qua discoverer of the
electron. Historians (should) know that the new particle described by Thom-
son in 1897 had less resemblance with the modern electron than the one
announced a few months earlier by Emil Wiechert. Yet they would have a
hard time convincing physicists to change the traditional story of the dis-
covery of the electron, for this story serves pedagogical purposes, a popular
empiricist view of the formation of concepts, and the memory of J. J. Thom-
son – who certainly achieved much as the head of the Cavendish Laboratory.
As an other instance, Erwin Schrödinger is regarded as a cofounder of quan-
tum mechanics, even though his original interpretation of the wave equation
turned out to be incompatible with later quantum mechanics. In contrast,
Poincaré is usually not regarded as a cofounder of relativity theory, even
though he had all the formalism of Einstein’s relativity theory and an inter-
pretation of this formalism empirically equivalent to Einstein’s.
To sum up, the adequacy of the arguments used by the tellers of Planck’s
discovery cannot be judged in an absolute manner. It varies according as our
interest lies in the discoverer’s way, in the impact of discovery, or in credit
attribution. From the preceding discussion, it seems to follow that the conti-
nuist and indeterminist historians were more interested in Planck’s genuine
intentions, and the discontinuist ones in the impact of his work or in the
credit to be given to him (compare Tables 2 and 3). The reality is more subtle.
All historians and physicists write as if they were respecting Planck’s intended
meanings. Yet it seems likely that the physicist-historians were haunted by
considerations of credit. Martin Klein, being a professional historian, cannot
 Planck’s Quantum 231

have followed this inclination. However, as Ehrenfest’s biographer and as a

student of Einstein’s papers on radiation, he had an outstanding interest in
the exploitation of Planck’s work by contemporary physicists.
The discontinuist histories are acceptable and interesting, as long as they
are taken for what they truly are: histories motivated by credit attribution or
by assessment of the impact of discovery. They are misleading, however, if
they are taken to reflect Planck’s actual intentions. How credit is attributed
is a genuinely interesting historical question, which may teach us a lot on the
physicists’ community in a given period; but the attribution of credit is not
itself the historian’s business. As for the impact of discovery, it is of course a
legitimate interest of the historian’s. But the views of the discoverer should
be clearly distinguished from those of his interpreters. Unlike historians, a
scientist rarely seeks to penetrate a colleague’s mind very deeply; he rather
extracts whatever seems useful to him and reconstructs it in harmony with
his own views. These creative reinterpretations constitute an essential part of
the overall process of discovery. They deserve special attention from his-
torians and philosophers of science.11

4. Quantum continuity

The previous discussion raises doubts on the soundness of the method fol-
lowed by discontinuist historians. It is not sufficient, however, to disprove
their claim regarding Planck’s introduction of quantum discontinuity. For this
purpose, a more detailed consideration of the arguments of indeterminist
and continuist historians is necessary.
Kuhn’s continuist thesis is essentially based on three arguments (Kuhn,
1978, pp. 125–127). Firstly, Planck needed the classical theory of radiation
in order to derive the relation unΩ(8pn2/c3)U between the spectral density un
of thermal radiation and the average energy U of a resonator at the frequency
n. Therefore, he could not assume a quantization of the resonator without
contradicting himself. Secondly, in his combinatorial derivation of the black-
body law, Planck proceeded by analogy with Boltzmann’s combinatorics of
1877. Therefore, his recourse to a distribution of energy-elements is likely to
have been only a shortcut to Boltzmann’s fuller consideration of equiprob-
able energy-intervals (or cells in velocity space): for the counting of com-
plexions, it does not matter whether a complexion is defined by giving to
232 O. Darrigol

each molecule a discrete energy or by stipulating to which interval of the

energy axis the molecule belongs. Thirdly, in 1906 Planck explicitly defined
his complexions in terms of equiprobable intervals rather than discrete ener-
gies. It seems extremely unlikely that in 1900 he would have meant sharp
quantization, only to retreat to a more closely Boltzmannian approach in
1906.
Kuhn’s argumentation looks strong. Yet it failed to convince Martin Klein
and most of the physicists interested in the issue. It would be too easy to
interpret this persistent disagreement in terms of pride and prejudice. In fact,
there are flaws in Kuhn’s arguments, and a few errors and omissions in his
reading of both Planck and Boltzmann. For example, Allan Needell (1980)
and Res Jost (1995, p. 70) have rightly noted that Planck’s conversion to
Boltzmann’s statistical conception of irreversibility occurred several years
after 1900, whereas Kuhn dates this conversion in 1897–8 with the introduc-
tion of ‘‘natural radiation.’’ Martin Klein [1979, p. 432], who has the deepest
knowledge of Boltzmann’s kinetic theory, noted that Kuhn had overlooked
the aspects of Boltzmann’s work that anticipated statistical mechanics. Kuhn
had difficulty finding his way in the mathematical thicket of Boltzmann’s and
Planck’s theories, and he sometimes got lost despite the avowed assistance of
a few physicist friends (Kuhn, 1978, p. xii). Anyone who feels this lack of ease
in Kuhn’s investigations, tends to distrust his more iconoclastic conclusions.
Even so, Kuhn’s adversaries seem to have overlooked the gravest flaw of
his argumentation. If, as Kuhn insists, Planck in 1900 was faithfully following
Boltzmann’s procedures, he should have reached the Rayleigh-Jeans law in-
stead of Planck’s law, for in Boltzmann’s gas case the size of the cells (the
counterpart of Planck’s energy-elements) disappears from the final entropy
formula. Then there must have been some inconsistency in Planck’s appli-
cation of Boltzmann’s method. Now we face the following dilemma: in order
to accept Kuhn’s first argument about the derivation of the relation between
spectral density and resonator energy, we must assume Planck to be a consist-
ent thinker; in order to accept his second argument on the nature of Planck’s
combinatorial entropy derivation, we must assume Planck to be an inconsist-
ent thinker.
One way to avoid this dilemma is found in Klein’s review of Kuhn’s book:
we may assume that Planck was uniformly inconsistent. In this view, Planck
tolerated or overlooked two contradictions: that between the derivation of
unΩ(8pn2/c3)U and resonator quantization, and that between Boltzmann’s
 Planck’s Quantum 233

method and resonator quantization (Klein, 1979, p. 431). A more convincing

way out of the dilemma is given by the opposite assumption that Planck was
uniformly consistent: his resonator entropy calculation contradicted neither
the derivation of unΩ(8pn2/c3)U nor the relation between entropy and prob-
ability, because Planck understood this relation in a way different from Boltz-
mann’s.
The basic point overlooked by Kuhn is that in 1900 (or before) Planck did
not adopt the statistical conception of irreversibility. He did so only around
1914. Therefore, despite formal similarities Planck’s entropy/probability con-
siderations differed fundamentally from Boltzmann’s. This fact turns out to
be essential to a proper understanding of the status of Planck’s energy-
elements. Allan Needell, a student of Martin Klein, has solidly established
this new vision of Planck’s radiation theory in his dissertation of 1980. His
argument goes as follows.
Planck’s resonators, responsible for the thermalization of radiation, were
not meant as elastically bound ions or other similar atomistic systems. When
Planck began his program, he was hostile to microphysical speculations and
preferred considerations based on general, macroscopic principles. Accord-
ingly, he regarded his resonators as miniature versions (much smaller than
the corresponding wavelength) of Heinrich Hertz’s resonators with no ohmic
resistance and with indeterminate internal structure. He derived the relation
between the electric dipole of a resonator and the surrounding field by com-
paring the energy fluxes across well-chosen surfaces surrounding the res-
onator. As Edward Jurkowitz shows in a forthcoming publication, this type
of reasoning was typical of Planck and of the Berlin physics inaugurated by
Hermann Helmholtz.
The resulting relations between the observable, secular properties of res-
onator and radiation were not completely determinate, for they involved un-
known phase differences between radiation and resonator. Planck exploited
the indeterminate internal structure of his resonators to drop the phase-de-
pendent terms, so that the evolution of the controllable aspects of his system
became deterministic and irreversible. The idea was that the internal intric-
acies of resonator dynamics conspired to rigorously cancel the unwanted
terms. This assumption of ‘‘natural radiation’’ differed from Boltzmann’s
analogous ‘‘molecular chaos’’ in an essential manner: it yielded a strict valid-
ity of the entropy law, whereas Boltzmann only obtained a statistical validity
of this law.
234 O. Darrigol

When in 1900 Planck appealed to Boltzmann’s relation between entropy

and probability, he still avoided the statistical conception of irreversibility.
He reinterpreted the ‘‘probability’’ in Boltzmann’s relation as a measure of
the elementary disorder implied in natural radiation and bound to strict irre-
versibility. Accordingly, Planck’s W depends on the indeterminate internal
structure of the resonators. There follows an important corollary: the energy-
elements occurring in the calculation of W pertain to the finer details of
resonator dynamics and do not contradict the secular, large-scale application
of electrodynamics that Planck made in his derivation of relations between
radiation and resonator properties. The relevant connections of Planck’s pro-
gram are visualized on Table 4.

Table 4: Planck’s scheme.

These remarks of Needell’s explain why there is no contradiction between

Planck’s resonator-entropy calculation and his earlier derivation of the re-
lation between U and un. They also make clear that for Planck the deeper
significance of the energy-elements was an open question, having to do with
electrodynamics at a finer, non-observable scale. This is why Needell adopted
the indeterminist position with regard to Planck’s introduction of quantum
discontinuity. More generally, Needell excludes any application of the mod-
ern distinction between classical and quantum physics to Planck’s work, be-
cause for Planck and his contemporaries there was no such thing as a closed,
 Planck’s Quantum 235

uniform doctrine of physics which the black-body spectrum could be said to

contradict (Needell, 1988, pp. xi-xliii).
In a paper published in 1988, I came to conclusions very close to Needell’s
(without having seen his dissertation). My motivation was different. Needell
had followed the history of Planck’s understanding of irreversibility, from his
dissertation on the entropy law to his late conversion to Boltzmann’s statisti-
cal view. Instead I was interested in the formal analogies between Boltzmann’s
and Planck’s theories and in their role in the construction of proto-quantum
formalisms. Owing to this different emphasis, I pursued the formal conse-
quences of Planck’s interpretation of combinatorial probability as a measure
of elementary disorder, with the following results (Darrigol, 1988, 1992).
According to Planck, the kinds of disorder involved in gas theory and in
radiation theory are different: the disorder is spatial in the former case, and
temporal in the latter. Consequently, the ‘‘probabilities’’ measuring the dis-
order are expressed by different formulas. In other words, the characteriza-
tion of the macrostate is different for Boltzmann’s gas and for Planck’s res-
onator. For Boltzmann, a macrostate of the gas is given by the list of the
numbers Ni of the molecules with the energy ie (better: whose energy lies
between ie and (iπ1)e), whereas for Planck a macrostate of a set of resonators
is given by the total energy of this set. Naturally, Boltzmann chooses the
energy-element e so small that the distribution Ni is approximately continu-
ous (and large enough so that the Stirling approximation can be used for
Ni!). In contrast, this distribution has no observable significance in Planck’s
case, so that there is no upper limit on the size of e. This explains why e
survived in Planck’s final entropy formula whereas it disappeared in Boltz-
mann’s. Both this strange feature and the difference between Planck’s and
Boltzmann’s combinatorial formulas are intimately related to the temporal
nature of the disorder involved in natural radiation.
Accordingly, the singularity of Planck’s reasoning and his ability to retrieve
the new black-body law depended much more on his definition of macrostat-
es than on his definition of microstates or complexions. He could equally
well define the complexions in the discrete manner of Boltzmann’s fiction, or
in the continuist guise of energy-interval ascriptions. A rudimentary knowl-
edge of Planck’s psychology, and continuity with the presentation found in
his lectures of 1906 make it extremely plausible that he shared Boltzmann’s
preference for the continuist version. He only used the discontinuist version
because, for his definition of a macrostate, it led to a much quicker calculation
236 O. Darrigol

of the number of complexions. The last trace of doubt is removed by the

earlier quoted clause in the Academy paper of 1900 (Planck 1900b, p. 701)
(‘‘When the thus computed ratio [E/e] is not a whole number, one can take
for P a neighboring whole number’’) and by Planck’s reference to Johannes
von Kries’ Spielräume in the Annalen paper of 1901 (p. 722). By Spielräume,
Kries (1886, p. 36) meant domains of equiprobability for probability distri-
butions of continuous variables. Planck later used the word as synomymous
of his own Elementargebiete der Wahrscheinlichkeit. For all these reasons, we
must agree with Kuhn that Planck did not intend to restrict the energy of his
resonators to discrete values in 1900–01.12

5. Conclusion

In the previous section, we have tried to understand Planck’s theory of radi-

ation in its own terms, without the deforming lens of contemporary and later
interpretations by other physicists. This is of course what Kuhn himself
wanted to do. According to his methodology, the incommensurability of dif-
ferent systems of thoughts implies that the historian should penetrate the
thoughts of past scientists without injecting elements of later systems. This
task requires attention to small details and apparent contradictions, which
ultimately reveal the original coherence of the studied thought in a sort of
Gestalt switch. In his black-body book Kuhn wanted to reveal the true coher-
ence of Planck’s radiation physics, just as his teacher Alexandre Koyré had
revealed the coherence of Aristotelian physics.13
Kuhn partly succeeded in this enterprise by removing some apparent con-
tradictions of Planck’s theory and giving more continuity to its evolution.
However, he failed to notice the essential aspects of this theory which Allan
Needell discussed in his dissertation. He committed precisely the kind of
methodological error with which he reproached other historians: he confused
Planck’s reinterpretation of Boltzmann with Boltzmann himself. He failed to
perform the Gestalt switch that would have revealed the full coherence of
Planck’s approach.
So far I have spoken as if Kuhn’s methodology was obviously sound, as if
the reconstruction of the assumed coherence of a past system was a legitimate
goal. Yet we saw that Galison reproached Kuhn with assuming too much
coherence in Planck’ thoughts at a time of fast and chaotic change. Similarly,
 Planck’s Quantum 237

in his review of Kuhn’s book Klein wrote: ‘‘In my opinion Kuhn tries too
hard to establish the internal consistency of Planck’s position. He seems to
be unwilling to consider the possibility that Planck himself was not always
completely clear about what he was doing’’ (Klein, 1979, p. 431). The objec-
tion seems even more pertinent for accounts such as Needell’s and mine
which convey to Planck’s thoughts more coherence than Kuhn himself per-
ceived. The problem is whether the coherence is artificially introduced by the
historian or is a genuine characteristic of the described thoughts.
Klein’s doubts are understandable in the case of Kuhn’s book, for Kuhn
makes assumptions of internal consistency (his first point) and temporal conti-
nuity (his third point) for which he gives no textual evidence. Moreover, Kuhn
appears to have arbitrarily made Planck consistent on some issues and incon-
sistent on others. Do these charges also apply to the accounts given by Needell
and myself? I do not think so, for these accounts follow very closely Planck’s
expressed justifications. For example, Needell’s crucial point that Planck
understood the combinatorial probability as a measure of elementary disorder
is completely explicit in the quantum papers of 1900–01. Planck [1900b, p. 698]
starts his reasoning with the words ‘‘Entropie bedingt Unordnung,’’ and goes
on to characterize the kind of disorder affecting his resonators. It is in fact eas-
ier to locate statements of Planck’s corroborating Needell’s account than to
understand why previous historians overlooked them.
In short, I believe the coherence which Needell saw in Planck’s approach
to the radiation problem to be real. Coherence, however, should be confused
neither with consistency nor with completeness. Let us define consistency as
the lack of logical contradiction in an entirely explicit, closed, conceptual
system. Consistency in this sense is never achieved during the construction
of a theory. It can only be reached in a later stage of consolidation and
axiomatization. In contrast, coherence refers to a harmonious weaving of
arguments without easily perceptible contradictions. Planck’s theory was co-
herent in this soft sense; but it could not be made consistent in the hard
sense. For example, if Planck had provided a physical mechanism for the
thermalization of his resonators (such as encounters with gas molecules), he
would have been forced to the absurd Rayleigh-Jeans law. His theory was
incomplete in a way that hid potential contradictions.
Incompleteness was an essential characteristic of Planck’s theory, as empha-
sized by Needell and myself. Accordingly, Planck had no definite opinion of the
exact meaning of his new quanta. He strove to remain as close as possible to
238 O. Darrigol

received dynamical conceptions, but he did not know what the exact dynamics
of the resonators would be. He did not introduce quantum discontinuity, he did
not intend a sharp break from received theories, but he believed that his quan-
tum of action signaled yet unknown aspects of small-scale physics.

BIBLIOGRAPHY

Darrigol, O.
1988: ‘‘Statistics and combinatorics in early quantum theory,’’ Historical studies in the
physical sciences, 19, pp. 17–80.
1992: From c-numbers to q-numbers: The classical analogy in the history of quantum theory
(Berkeley, 1992).
Galison, P.
1981: ‘‘Kuhn and the quantum controversy,’’ British journal for the philosophy of science,
32, pp. 71–85.
Hermann, A.
1969: Frühgeschichte der Quantentheorie (1899–1913), Mosbach.
Hund, F.
1967: Geschichte der Quantentheorie, Mannheim.
Jammer, M.
1966: The conceptual development of quantum mechanics, New York.
Jost, R.
1995: ‘‘Planck-Kritik des T. Kuhn,’’ in Das Märchen vom Elfenbeinernen Turm. Reden
und Aufsätze, Berlin, pp. 67–78.
Kangro, H.
1970: Vorgeschichte des Planckschen Strahlungsgesetzes, Wiesbaden.
Klein, M. J.
1962: ‘‘Max Planck and the beginnings of the quantum theory,’’ Archive for the history
of exact sciences, 1, pp. 459–479.
1963: ‘‘Planck, entropy, and quanta, 1901–1906,’’ The natural philosopher, 1, pp. 83–108.
1966: ‘‘Thermodynamics and quanta in Planck’s work,’’ Physics today, 19:11, pp. 23–32.
1970: Paul Ehrenfest, vol. 1: The making of a theoretical physicist, Amsterdam.
1979: Contribution to ‘‘Paradigm lost? A review symposium,’’ Isis, 70, pp. 429–433.
Kries, J. v.
1886: Die Principien der Wahrscheinlichkeitsrechnung, Freiburg.
Kuhn, T. S.
1977: The essential tension, Chicago.
1978: Black-body theory and the quantum discontinuity. 1894–1912, Oxford.
Needell, A.
1980: Irreversibility and the failure of classical dynamics: Max Planck’s work on the quan-
tum theory: 1900–1915, Diss., University of Michigan, Ann Arbor.

1988: Needell, Introduction to Planck, The theory of radiation, Los Angeles: Tomash.
Planck, M.
1900a: ‘‘Über eine Verbesserung der Wien’schen Spektralgleichung,’’ Deutsche Physikalis-
che Gesellschaft, Verhandlungen, 2, pp. 202–204.
 Planck’s Quantum 239

1900b: ‘‘Zur Theorie des Gesetzes der Energieverteilung im Normalspektrum,’’ Deutsche

Physikalische Gesellschaft, Verhandlungen, 2, pp. 237–245, also in [Planck, 1958],
1, pp. 698–706.
1901: ‘‘Über das Gesetz der Energieverteilung im Normalspectrum,’’ Annalen der Physik,
4, pp. 553–563, also in (Planck, 1958), 1, pp. 717–727.
1920: ‘‘Die Entstehung und bisherige Entwicklung der Quantentheorie’’ (Nobel lecture),
in (Planck, 1958), 3, pp. 121–136.
1948: ‘‘Wissenschaftliche Selbstbiographie’’ (1948), also in (Planck, 1958), 3, pp. 374–
401.
1958: Physikalische Abhandlungen und Vorträge, Braunschweig.
Rosenfeld, L.
1936: ‘‘La première phase de l’évolution de la théorie des quanta,’’ Osiris, 2, pp. 148–196.
Sommerfeld, A.
1919: Atombau und Spektrallinien, Braunschweig, 3rd. ed. (1922).

NOTES

1. I thank John Stachel and Alfred Tauber for their hospitality at the Boston Colloquium,
Edward Jurkowitz and Jürgen Renn for helpful comments.
2. On this point see, e.g., (Klein, 1962, p. 471) and (Kuhn, 1978, pp. 110–113).
3. See J. Büttner, J. Renn, and M. Schemmel’s comments on the Einstein-Planck connec-
tion in their contribution to this conference.
4. Cf. (Rosenfeld, 1936), (Klein, 1962, 1963), (Kuhn, 1978), (Needell, 1980), (Darrigol,
1992, part I), and (Kangro, 1970) on the experimental side.
5. Planck could only obtain the spatial homogenization of radiation. The resonators could
not change the spectrum of the radiation, as Ehrenfest and Einstein later noted.
6. Cf. Dieter Hoffmann’s contribution to this issue.
7. Armin Hermann adopted a similar attitude in (Hermann, 1969, pp. 34–35).
8. J. Mehra and H. Rechenberg (see the first volume of their history of quantum theory),
could also be regarded as indeterminist, for they simply paraphrase Planck’s paper and
do not offer any interpretation of their own.
9. The sentence is not in Planck’s subsequent Annalen paper (Planck, 1901).
10. Cf. [Klein, 1962, p. 474]: ‘‘It is obviously of the very essence of Planck’s work that e
should not vanish, if the proper distribution law were to be reached. Planck apparently
did not even consider the possibility of taking this limit. This is undoubtedly related to
Planck’s apparent unawareness of the equipartition theorem and all it implied.’’
11. See the contribution to this conference by J. Büttner, J. Renn, and M. Schemmel.
12. For Planck’s identification of Spielräume and Elementargebiete, see (Planck, 1920, p.
127); Kuhn (1978, pp. 121, 286) notes Planck’s reference to Kries, but overlooks its
meaning. I thank Michael Heidelberger for discussing the matter with me and for lend-
ing me a copy of Kries’ book.
13. Cf. (Kuhn, 1977, p. xii): ‘‘I offer [students] a maxim: When reading the works of an
important thinker, look first for the apparent absurdities in the text and ask yourself
how a sensible person could have written them. When you find an answer, I continue,
when those passages make sense, then you may find that more central passages, ones
you previously thought you understood, have changed their meaning.