THE CREATION OF QUANTUM MECHANICS AND THE

BOHR-PAULI DIALOGUE
 STUDIES IN THE HISTORY
OF MODERN SCIENCE

Editors:

ROBERT S. COHEN, Boston University

ERWIN N. HIEBERT, Harvard University
EVERETT I. MENDELSOHN, Harvard University

VOLUME 14
 JOHN HENDRY

THE CREATION OF
QUANTUM MECHANICS
AND THE
BOHR-PAULI DIALOGUE

D. REIDEL PUBLISHING COMPANY

A MEMBER OF THE KLUWER . . ACADEMIC PUBLISHERS GROUP

DORDRECHT/BOSTON/LANCASTER
library of Congress Cataloging in Publication Data

Hendry, John, 1952-

The creation of quantum mechanics and the Bohr-Pauli dialogue.
(Studies in the history of modern science; v. 14)
Bibliography; p.
Includes index.
1. Quantum theory - History. 2. Bohr, Niels Henrik David, 1885-1962.
3. Pauli, Wolfgang, 1900-1958. I. Title. II. Series.
QCI73.98.H46 1984 530.1'2'09 83-27251
ISBN-13: 978-94-009-6279-8 e-ISBN-13: 978-94-009-6277-4
DOl: 10.1007/978-94-009-6277-4

Published by D. Reidel Publishing Company,

P.O. Box 17, 3300 AA Dordrecht, Holland.

Sold and distributed in the U.S.A. and Canada

by Kluwer Academic Publishers,
190 Old Derby Street, Hingham, MA 02043, U.S.A.

In all other countries, sold and distributed

by Kluwer Academic Publishers Group,
P.O. Box 322, 3300 AH Dordrecht, Holland.

All Rights Reserved.

© 1984 by D. Reidel Publishing Company.
Softcover reprint of the hardcover I st edition 1984
No part of the material protected by this copyright notice may be reproduced or
utilized in any form or by any means, electronic or mechanical,
including photocopying, recording or by any information storage and
retrieval system, without written permission from the copyright owner.
To Dee
 TABLE OF CONTENTS

Preface ix

Scientific Notation xi

Chapter 1. Introduction
Chapter 2. Wolfgang Pauli and the Search for a Unified Theory 6
Chapter 3. Niels Bohr and the Problems of Atomic Theory 24
Chapter 4. The Technical Problem Complex 35
Chapter 5. From Bohr's Virtual Oscillators to the New Kinematics
of Heisenberg and Pauli 51
Chapter 6. The New Kinematics and Its Exploration 67
Chapter 7. Wave Mechanics and the Problem of Interpretation 83
Chapter 8. Transformation Theory and the Development of the
Probabilistic Interpretation 102
Chapter 9. The Uncertainty Principle and the Copenhagen Inter-
pretation 111
Chapter 10. Concluding Remarks 129

Notes 134

Select Bibliography 166

Index 174
 PREFACE

Many books have been written on the history of quantum mechanics. So far
as I am aware, however, this is the first to incorporate the results of the large
amount of detailed scholarly research completed by professional historians of
physics over the past fifteen years. It is also, I believe, the first since Max
Jammer's pioneering study of fifteen years ago to attempt a genuine 'history'
as opposed to a mere technical report or popular or semi-popular account. My
aims in making this attempt have been to satisfy the needs of historians of
science and, more especially, to promote a serious interest in the history of
science among phYSicists and physics students. Since the creation of quantum
mechanics was inevitably a technical process conducted through the medium
of technical language it has been impossible to avoid the introduction of a
large amount of such language. Some acquaintance with quantum mechanics,
corresponding to that obtained through an undergraduate physics course, has
accordingly been assumed. I have tried to ensure, however, that such an
acquaintance should be sufficient as well as necessary, and even someone with
only the most basic grounding in physics should be able with judicious skip-
ping, to get through the book. The technical details are essential to the
dialogue, but the plot proceeds and can, I hope, be understood on a non-
technical level.
The research for this book has extended over a number of years, and
although the bulk of it is published here for the first time some aspects have
appeared in my Ph.D. thesis and in a number of published papers. In the
course of my work I have received assistance of varying kinds from many
teachers and colleagues, and I should like to offer my thanks to them and
also to those other colleagues on whose published research I have drawn. The
latter, very considerable, debt is recorded in the bibliography. In the fonner
class I should like to mention particularly Gerald Whitrow and Jon Dorling,
who oversaw my doctoral research, and David Cassidy, John Heilbron, Erwin
Hiebert, Karl von Meyenn and B. L. van der Waerden, who have provided
much-needed criticism and valuable encouragement since. I should also
like to thank all owners of copyright for permiSSion to reproduce private
correspondence, the Royal Society and the British Academy for invaluable
financial assistance, and the staff of the old History of Science and Technology
ix
x PREFACE

Department at Imperial College, London, of the Niels Bohr Institutet, Copen-

hagen, and of the Historisches Institut, University of Stuttgart, for hospitality
at various times.

Spring 1983 JOHN HENDRY

 SCIENTIFIC NOTATION

In most cases, scientific symbols used are defined upon their first appearance.
Exceptions are standard notations and the symbols for quantum numbers
in the old quantum atomic theory, which are given below.

Standard Notations

c Speed of light
J) Frequency
e Electron charge
m Particle mass
h Planck's constant

Quantum Numbers (for alkali atoms with single valence electron)

n Radial quantum number

I
k--i-
K--i-
j,J--i-
!
AZimuthal quantum number (orbital angular momentum of valence
electron)
Inner quantum number (total angular momentum)
s,R, r Spin, or angular momentum of atom minus valence electron
m Magnetic quantum number (total angular momentum in direction
of applied magnetic field)

xi
EverY sentence I utter is to be under-
stood as a question and not as an
affirmative.

NIELS BOHR

In my youth I believed myself to be

a revolutionary; now I see that I was
a classicist.

WOLFGANG PAULI
 CHAPTER 1

INTRODUCTION

Quantum mechanics has evolved considerably since it first received a defini-

tive formulation and interpretation in 1927. The quantum theory of fields
grew up quickly over the next few years and after the discovery of a new
range of elementary particles in the 1930s the formulation of field and
particle theories has developed apace since the war. But in all this, two
things have remained more or less unchanged. On one hand the quantum
theory has continued in all its formulations to show a remarkable predictive
power in respect of experimental observations. In this respect it must rank
as an extraordinarily successful physical theory, and as one that will not
easily be displaced. On the other hand, however, dissatisfaction with the
conceptual foundations of the theory has also apparently endured. Many
working physicists are seemingly content to accept what Einstein referred
to as the "gentle pillow" of the Copenhagen interpretation without asking
any further questions, and this has long been accepted as an orthodox posi-
tion. But if we restrict our attention to physicists (or indeed philosophers)
of the first rank, then we see immediately that such an orthodoxy is illUSOry.
It was created in the late 1920s when many of the leading quantum physicists,
among them Bohr, Born, Heisenberg, Pauli, Dirac, Jordan and von Neumann,
sunk their more philosophical differences in an effort to repel the challenge
of the semi-classical interpretations and get on with the job of developing
quantum electrodynamics. But those differences remained. Copenhagenism
was and is a generic term covering a whole range of related interpretations.
Even when these interpretations are taken together, they cannot be consid-
ered as an entirely dominant orthodoxy. Among their early opponents some
physicists might arguably be dismissed as narrow-sighted conservatives. But
such outright dismissal is very difficult to uphold in Einstein's case, and still
more so in those of SchrOdinger and de Broglie, neither of whose preferred
interpretations could reasonably be labelled classical. More recently attention
has shifted from the physical interpretation of quantum mechanics towards
the logical and mathematical consistency of quantum field theory, but the
issues remain closely connected and opposition to Copenhagenism remains
strong. However, and here lies the crux of the matter, the opponents seem
to be no nearer to providing a valid alternative than were their predecessors
1
2 CHAPTER 1

of the late 1920s. Beyond the limited compromise of Copenhagenism there

is still no such thing as a consistent and generally acceptable interpretation of
quantum mechanics, and the evidence of the last fifty years points unerringly
to the conclusion that there will not be one until either the structure of our
physical conceptions, or our expectations of physical theory, or the quantum
theory itself should undergo radical changes more far-reaching than any
yet seen.
Faced with this dilemma it is tempting to react as did Peter Debye to the
problem of electrons in the nucleus, a problem that arose in the immediate
wake of quantum mechanics, by treating it as something best ignored, "like
the new taxes". And many physicists have indeed taken this course, either
ignoring the interpretative problem altogether (paying the taxes without
question) or proceeding stubbornly to seek fundamentally classical inter-
pretations that are demonstrably not there (stalling the taxman). But whereas
such attitudes may be expedient in the short term they are ultimately incon-
sistent with the very spirit of the scientific enterprise. Wolfgang Pauli, who
will turn out to be one of the main actors in the story that follows, responded
to Debye's suggestion by announcing publicly his hitherto private specula-
tions on the existence of the neutrino. He was convinced that any positive
line of exploration was better than none at all, and this conviction quickly
bore fruit as the existence of the neutrino was confirmed. The interpretative
problem of quantum theory is several orders more fundamental than that of
nuclear electrons, and has proved immensely more resistant to attempts
at a solution. But a theory with innate inconsistencies, whatever its present
predictive success, cannot be expected to serve for ever. If the problem, like
the tax, does not bear thinking about, then that is the strongest indication
we can possibly have that it needs thinking about. And while it may not be
so easily solved we can at least try to understand how such an extreme
situation arose in the first place.
One aim of this study, then, is to approach the history of the theory of
quantum mechanics as a means of exploring its philosophy. We shall not
address ourselves explicitly to the various attempts to impose philosophical
positions upon the completed theory. But by looking at the conceptual
features surrounding its genesis, and at the philosophical and other precon-
ceptions built into its structure and foundations, we shall try to come to
a clearer understanding of these attempts, of the chronic differences of
opinion they reflect, and of the essential nature of the problem they address.
The second aim is to present the history of quantum mechanics in such
a way as to enable it to be related to the mainstream of intellectual history,
 INTRODUCTION 3

not through the study of surface parallels reflecting the prevailing themata
of the milieu, but through that of the essential conceptual changes underlying
these.
With these aims in mind, the history will necessarily be a selective one,
and the selection may sometimes appear arbitrary. The concentration will
be upon the course of matrix mechanics, for example, at the expense of that
of wave mechanics. This is not to deny the historical importance of the latter
theory, but it is the former, the work of an interacting group rather than of
a succession of individuals, in which the conceptual tensions can most easily
be seen. Again, though the listed contents of the history will inevitably have
much in common with those of the existing accounts, the emphasis will be
very different. In particular there will be no attempt made to give anything
like a complete account of the more technical developments within quantum
theory, which have been well treated elsewhere and are covered by the
bibliography. The same holds also for the prevailing philosophical milieu, so
long as its notions were not applied directly and consequentially to physical
problems. Instead the concern here will be with the middle area, with fun-
damental but specific epistemological and methodological beliefs, and with
the manifestation of these beliefs both in the general debate on the quantum
problems and, especially, in the major technical advances.
Straightforward as it is, this perspective requires some comment, for the
concern of historians to date has tended to be with either the technical
development of quantum mechanics or its conceptual background, as
expressed through largely undefined philosophical debates, such as that on
the causality issue. Even where the two aspects have been treated by the same
historian, as by Paul Forman in his wide range of scholarly papers, they have
remained imperfectly connected. But whatever the immediate concern of
the historian it is this very connection that is most crucial to the dynamics
of the events being considered.
The language of modern theoretical physics is that of symbolic models
and mathematical equations, and although a physicist may think in terms of
physical or metaphysical concepts he must continually translate his thoughts
into this working language. It provides not only the means of communication
within the physics community but also those between the underlying prin-
ciples of physics on one hand and the highly artificial experimental results
on the other. It is thus an omnipresent and irreducible part of the develop-
ment of physical science. And, this being the case, the innovations of the
1920s were inevitably of a highly technical, symbolic and mathematical
nature. Any attempt to reduce them completely to abstract conceptual
4 CHAPTER 1

developments automatically distorts the historical process and introduces

a gulf between story and evidence that is historically unacceptable. But
in seeking to understand why as opposed to how the innovations took place,
and in seeking to extract the true essence of these innovations, it is the
development of abstract concepts that is of the most relevance and interest.
This poses a problem for the historian. The physicists concerned with a
particular set of problems did not all share the same influences and concerns;
each one approached the current issues, both technical and philosophical,
from a different conceptual framework. Some no doubt did see the technical
problems of quantum theory as just that, and others no doubt saw the
historically elevated causality issue as the most fundamental imaginable.
But this was patently not the case for the most innovative of the quantum
physicists, who derived their views on both subjects from more fundamental
concerns.
It is usually impossible to connect a scientist's deepest, usually religious,
concerns with his scientific work, and no attempt will be made to do this
here. But it is possible to relate both the technical development of quantum
mechanics and the surrounding philosophical debate to more fundamental
issues concerning the concepts used to describe the physical world. The aim
here will therefore be to trace the development of these issues, and of their
manifestation in the evolution of quantum mechanics. This done it should
be possible to place the interpretative problem of quantum mechanics in a
somewhat clearer historical light than hitherto, to relate the interpretations
offered more closely and accurately to currents of thought prevailing in
other fields, and even to get a clear idea of what type of interpretative
problems remain.
The plan of the study is to begin with an analysis of the conceptual
background to quantum mechanics in terms both of the commonly perceived
issues (causality, energy conservation, the wave-particle duality) and of the
fundamental epistemological and methodological problems manifest at about
the same time in the context of general relativity theory and the search for a
unified field theory. Having shown how these latter problems were carried
over into the quantum context we shall then relate the conceptual background
first to the technical problems encountered in the old quantum theory,
then to the genesis and reception of the ill-fated virtual oscillator theory,
and then to the genesis in 1925 of Heisenberg's "new kinematics", the
foundation stone of quantum mechanics. Through this first part of the study
the evolving theme will be one of a debate, its origins in the quest for a
unified general relativistic field theory, between the established master,
 INTRODUCTION 5

Niels Bohr, and the up-and-coming Wolfgang Pauli. And in the second part we
shall take this theme as a guide through the evolution of quantum mechanics,
from Heisenberg's breakthrough to the establishment of a definitive formu-
lation and interpretation in 1927. We shall look first at the early exploration
of the meaning, significance and implications of the new kinematics, then
at the early evolution of wave mechanics and at the attempts of Heisenberg,
Pauli and Dirac to integrate the technical advances of this theory with the
fundamental concepts of Heisenberg's. A predominant theme in this part
of the story wijl be an ongoing attempt to base quantum mechanics upon a
foundational framework consistent with these characteristic concepts; and in
the ensuing sections we shall show how this attempt, building upon the
insights achieved in the wake of wave mechanics, led first to the defmitive
formalism of the statistical transformation theory in Hilbert space, and then
to Heisenberg's uncertainty principle. All these developments took place
largely within the framework of Pauli's ideas, but in the last section of
the main story we shall turn back to the other side of the fundamental
debate and trace the evolution of Bohr's principle of complementarity and
the compromise of the Copenhagen group of interpretations.
The characteristic of quantum mechanics that made it a truly revolutionary
theory of physics was its explicit rejection of the classical criterion of a
consistent visualisation of physical events. For centuries natural philosophers
had argued whether a consistent picture in space and time was the ideal to
which all theories should aspire or an idol imposed by the tyranny of the
senses, the worship of which could only stand in the way of truth. By the
time the issue was put to the test in the context of the quantum paradoxes,
however, the old debate had apparently been long forgotten, and it was
not immediately obvious just what was at stake. The role of visualisation
was subsumed under more general questions as to the operational foundations
of physical concepts, and indeed remained so subsumed throughout the
creation of the theory. Throughout our study too, then, the central theme
will be that of the status and applicability of physical concepts. As we reach
the end of the story, this theme will merge into that of Anschaulichkeit,
sometimes translatable as visualisability, sometimes as something less specific
to the visual sense. Broadly speaking the path will be from the competing
demands of visualisation on one hand and physical operationalism on the
other towards a common recognition that neither ideal was obtainable.
 CHAPTER 2

WOLFGANG PAULI AND THE SEARCH FOR A

UNIFIED THEORY

INTRODUCTION: MANIFESTATIONS OF DUALITY

It has been customary for historians investigating the background to quantum

mechanics to concentrate on one or other of the particular branches of
quantum physics. In some cases this approach may be vindicated by a similar
specialisation on the part of physicists whose work is being discussed. But
even though they may have published only within restricted areas, most
physicists took their perception of the problem complex from the wider
field of their reading, correspondence and conversation. And most of those
involved in the actual development of quantum mechanics had already
contributed important work in several distinct areas. In general, the major
conceptual issues ran through all such areas, and their most obvious and
widely discussed manifestations were not necessarily their most significant.
The wave-particle duality, which was almost certainly the most characteristic
conceptual feature of quantum theory, was most commonly discussed in
the contexts of experimental X-ray physics and individual absorption and
emission phenomena. This discussion reached its peak, in the years imme-
diately before the development of the new quantum mechanics, in the
context of the Compton effect, viewed by at least one historian as a turning
point in physics. 1 But by the time of Compton's experiments this whole
area had long been one of only secondary importance for the active develop-
ment of ideas pertaining to the duality problem, and dramatic though his
results were to a wider audience they had relatively little impact on the
physicists whose work was to be of importance in the creation of quantum
mechanics. 2
Ever since the tum of the century, and especially since the demonstration
of X-ray diffraction in 1912, discrete X-ray phenomena had offered the
clearest and best-known demonstration of the localised particulate properties
of light, just as the continuous absorption of interference phenomena pro-
vided the clearest portrayal of its non-localised wave-like nature. But following
the work of Poincare, Ehrenfest and Jeans it was clearly recognised by about
1913 that any proof of this dual nature of light rested not on the individual
wave and particle phenomena but on the statistical behaviour of black-body
6
 PAULI AND A UNIFIED FIELD THEORY 7

radiation, according to Planck's law. Although the fact had not been rec-
ognised at the time, it was in 1900, in his very first presentation of this law,
that Planck had, albeit unwittingly, introduced the duality to physics; and it
was in the details of the law's derivation that the duality manifested itself
as a logical necessity. Even with the demonstration of this necessity and the
full recognition of the wave-particle duality, this duality was not universally
accepted, and for the best of reasons: in the face of a long-standing tradition
requiring a physical description to be above all structurally visualisable and
as such self-consistent, it seemed better to have an intelligible classical wave
theory with flaws than a totally unintelligible "dual" theory.3 The issue
therefore remained a live one and a new wave of X-ray experiments in 1921
and the ensuing Compton effect results were the subject of heated discussion.
But to physicists such as de Broglie and Schrodinger, in their struggles towards
the wave mechanics, as Jordan and Dirac, who played a large part in the
development of matrix mechanics, and as the old masters Einstein, Lorentz,
and Planck himself, the phenomena of black-body radiation and quantum
statistics were of far greater importance. 4 And meanwhile the development
of theoretical spectroscopy through the Bohr theory of the atom had also
encountered the wave-particle duality as a force every bit as potent as in the
earlier contexts.
The problems of quantum spectroscopy were less explicitly related to the
wave-particle paradox than either those of X-ray phenomena or those of
quantum statistics. But partly because of this, because the paradox could be
approached obliquely and because it did not stand in the way of a host of
new results and predictions, many of the most brilliant physicists found this
field the more suitable context for their endeavours to solve the quantum
problem. In particular Bohr himself developed his ideas very largely within
this context, and the problem complex of the Bohr theory was, as has indeed
been recognised, the central source of quantum mechanics in the form of
Heisenberg's new kinematics. s It was not, however, the only such source.
For once we begin to look at just how the new kinematics developed out
of the Bohr the~ry, the full extent to which the fundamental problems
transcended the individual branches of physics becomes clear. And we find
that Heisenberg's work is also rooted in yet another problem complex, that
of the search for a unified general relativistic field theory.
The main debate over unified general relativistic theories, following
Weyl's attempt at such a theory in 1918, does not in fact feature in the
existing accounts of the development of quantum mechanics. Since neither
Bohr nor Heisenberg were at all involved, and since the theories bore little
8 CHAPTER 2

direct relationship to the quantum postulate, this is not perhaps surprising.

But the debate did occur in the years immediately preceding the genesis
of quantum mechanics; and apart from Weyl it did involve Einstein and
especially Pauli, who carried over what he had learnt into the quantum
context and into his creative interaction of the early 1920s with Heisenberg
and Bohr. The basic issues concerned a duality between field and particle
theories that was in fact closely akin to the wave-particle duality of quantum
theory. And the debate was so fundamental and important as to form an
integral part of the perceived problem complex for any physicist who was
actively involved in general relativity theory - including such important
teachers and contributors to the quantum theory as Sommerfeld, Born,
SchrOdinger, Dirac and Hilbert. In this chapter we shall therefore look at
the search for a unified field theory, and in particular at Pauli's response
to this search, and show how it came to interact in general terms with the
conceptual debate surrounding quantum theory.

WEYL'S THEORY AND ITS RECEPTION

The general theory of relativity had evolved gradually over a period of years,
but had finally appeared in fully developed form in 1916. 6 From the observed
equivalence between inertial and gravitational masses, Einstein had deduced
that the gravitational field imparted the same acceleration to all bodies, and
had postulated that it could therefore be transformed away completely
and replaced for each infmitesimal region, and for its effects on all physical
processes there, by a suitable choice of the space-time coordinate system.
This "principle of equivalence", combined with the assumption of the special
theory of relativity for each transformed infinitesimal region, led directly
to an identification of the gravitational field with a geometry, or coordinate
system, of space-time. Drawing on the postulate of "general covariance",
according to which physical laws were required to be independent of the
choice of coordinate system, Einstein had identified this geometry as Rie-
mannian: whereas in the traditional, 'flat', Euclidean geometry the directions
of vectors at different points could be directly compared, their relative
direction in the 'curved' Riemannian geometry was dependent upon the
choice of path joining the point. The parameters defming this choice could
be identified with those defming the gravitational field. 7
The development of Einstein's theory was closely monitored by David
Hilbert and Max Born in the mathematics department at GOttingen, and also
by Hermann Weyl, who was Hilbert's assistant and colleague until the end of
 PAULI AND A UNIFIED FIELD THEORY 9

1913, when he moved to ZUrich. 8 In ZUrich for a year, Weyl worked mainly
on mathematics; but he kept up his contacts with G6ttingen, and also made
direct contact with Einstein, then a professor in the same city. Weyl was
called up for service in the First World War, but returned to ZUrich late in
1916. Einstein, meanwhile, had moved to Berlin, but Weyl had become
sufficiently interested in his work, and was sufficiently impressed by his
newly published general theory of relativity, to drop his pre-war mathematical
investigations and make Einstein's work the basis of a new programme of his
own. 9 Early in 1918 he published a mathematical generalisation of Einstein's
theory, and suggested that this generalisation might encompass not only
gravitational but also electromagnetic phenomena. 10
Weyl argued that the essence of the change from Euclidean to Riemannian
geometry in the general theory of relativity lay in the change from a finite
geometry, in which a uniform metric could be applied, to an infinitesimal
one, in which a metric could only be applied directly to infinitesimally small
regions. One consequence of this change was the dependence in Riemannian
geometry already noted, of the relative direction of two vectors acting at
two distant pOints on the choice of path joining them. But, argued Weyl,
the Riemannian geometry still contained an element of finite geometry that
made it mathematically something of a compromise, for the specific choice
of which there was no sound a priori reason. Whereas the relative direction
of two vectors acting at distant points could not be directly compared, their
relative length could. Weyl therefore proposed a generalised geometry in
which such relative lengths were also dependent upon the choice of path
joining the two points, or, in other words, in which elements of length
were specified only to within an arbitrary function of position, denoted by
the term 'gauge'. In order to make this generalised geometry the foundation
of a physical theory, as the Riemannian geometry was the foundation of
Einstein's theory of gravitation, Weyl supplemented Einstein's principle of
general covariance by the requirement of gauge invariance, namely that
physical laws should be independent of the choice of gauge. He then pro-
ceeded to build a generalised analogue of Einstein's theory.
In Einstein's Riemannian theory, the connection between the coordinates
of a point in space-time and another point infinitely close was given by the
quadratic form corresponding to the 'distance', ds, between them:
ds 2 = ~ g/lV dx/l dx v ,
/lV
with g/lV the coordinates of the gravitational potential. In Weyl's generalised
theory this relationship still held. But the relationship between the two
10 CHAPTER 2

points, and the general metric structure of space, also depended upon an
additional, linear, form,
df/> = ~/L f/>/L dxw

The interest of this for Weyl lay in the fact that just as the gravitational
potential could be equated with the coordinates appearing in the quadratic
form, so the electromagnetic potential could be expressed by linear coor-
dinates of the form f/>/L' Pursuing this analogy he sought ways in which the
classical electromagnetic theory could be definitely connected with the
new geometry. Starting from the basic assumption, which he attributed
to Gustav Mie, that all laws of a generalised physical theory should rest
upon a generalised action invariant, or principle of least action, he suggested
a function that might fill this role. And, leaving the actual identity of such
a generalised action function undetermined, he showed that just as the laws
of energy and momentum conservation could be identified as by Einstein
and others with the general relativistic invariance of the normal action
function, so the conservation of electricity could in principle be identified
with the gauge invariance of this new, as yet arbitrary, function. This
achieved, he interpreted his theory as a unification of gravity and electro-
magnetism, and, while stressing the mathematical difficulties in the way
of further developments, he also held out hope for a deduction of the exis-
tence of the electron and of the quantum behaviour of the atom from the
theory. 11
Extraordinarily difficult though Weyl's work was, it was enthusiastically
received. Mie wrote to him that it was the most fascinating mathematical
work he had ever read,12 Sommerfeld that he found it "truly wonderful", 13
Eddington that it was his "constant companion", 14 and Einstein that it was
a "master-symphony". 15 The young Pauli started his research career by
developing aspects of the theory, 16 and E. T. Whittaker's recollections
confirm that it made an enormous impression upon many theoretical
physicists of the period. 17 But with the praise came criticism.
Einstein had originally welcomed Weyl's interest in general relativity
theory, but he had already recognised from the first exchange of ideas that
they were unlikely to agree, and had expressed himself pleased in particular
because "he who has the most powerful opponent excels".18 When in March
1918 Weyl sent him proofs of his book, Raum-Zeit-Materie, incorporating
his theory, Einstein responded with extravagant praise and agreed immediately
to communicate a paper on the new theory to the Berlin Academy.19 On
receiving the paper he was again deeply impressed,20 but a week later he
 PAULI AND A UNIFIED FIELD THEORY 11

wrote to Weyl that, despite the beauty of his theory, "I must say frankly that
it cannot possibly correspond in my opinion to the theory of Nature, that in
itself it has no real meaning."21 The problem, Einstein explained, was that
since the relative lengths of two space-time vectors depended in Weyl's theory
upon the path joining them, the lengths of two measuring rods or the periods
of two clocks, separated and then brought together again, would depend
upon the paths each had taken in the meantime. This discrepancy would
increase with the time of separation, and Einstein argued that if it really
existed it would show up in the measurements of atomic spectra, the observed
frequencies corresponding to the periods of the atomic clocks. Since no such
discrepancy was observed, he could not accept that Weyl's theory had any
real physical significance. 22 A few days later Einstein reported with apologies
that Nernst and Planck were even less happy with Weyl's paper than he was,
and that they were demanding on behalf of the Berlin Academy that his
opinion be appended to Weyl's paper as a postscript. Together with a reply by
Weyl, it was. 23
In his reply, Weyl argued that he did not in fact see his theory as counter
to experience, for the formal mathematical process of vector displacements
that was the foundation for his geometry could not be identified with the
physical motions of clocks and measuring rods, the behaviour of which had
still to be worked out. He conceded that Einstein's own theory maintained
the lengths of measuring rods and periods of clocks. But he insisted this
theory had only been developed for the case of a static gravitational field,
with no electromagnetic field present. In strongly varying gravitational and
electromagnetic fields such as would be needed to effect the separation and
return proposed, the constancy of the period of a clock could be predicted
according to Weyl by neither Einstein's theory nor his own, as they had so
far been developed. This being the case he returned to his claim that the
geometry of his theory was the only one that could be mathematically
justified, and that it would be strange indeed were Nature to have chosen a
compromise geometry such as Einstein's, arbitrarily pasted onto the electro-
magnetic field. 24
Weyl had already expressed the opinion in his book that a priori considera-
tions could only specify what relations were possible in Nature, and that the
choice of actual physical laws had to be based upon empirical considera-
tions. 25 And he conceded that in this respect his own theory remained a
speculative one to be proved or disproved. 26 But he had also set out his
view that a priori considerations were fundamental to theoretical physics.
Writing to Einstein in May 1918 he repeated his reluctance to "accuse God of
12 CHAPTER 2

mathematical inconsistency", and insisted that he still believed in his theory

and in the importance of carrying it through to testable physical conclu-
sions. 27 After initially retorting that it seemed to him just as bad to accuse
God of a theoretical physics that did not do justice to human observations,28
Einstein for his part accepted the validity of Weyl's defence - though not of
his philosophy of physics - and while stating his belief that Weyl's theory
would be proved false accepted that it had not yet been so proved. 29
Although Einstein remained convinced that Weyl's theory would prove to
be irreconcilable with physical reality, he was prepared to accept Weyl's
defence and wait for further research to establish for certain which of their
theories was correct. 30 But Wolfgang Pauli, in his famous encyclopedia article
of 1921 surveying relativity theory, criticised Weyl's defence. According to
Pauli, Weyl's argument implied that the coefficientsg~p and cp~ of his theory,
in contrast to the g~p of Einstein's theory, were not directly observable; and
this in itself meant in Pauli's view that the field of Weyl's theory could not
have any physical significance. 31 The same year Eddington noted that although
Weyl's geometry was indeed more general than Einstein's, it was itself still
subject to one apparently arbitrary restriction, namely that zero lengths were
preserved unchanged under all its allowed transformations. If this restriction
were removed, the resulting theory would contain not only the 10 coefficients
g~p of Einstein's theory and the 4 more cp~ added by Weyl, but also another
26 coefficients, bringing the total to 40. 32 Weyl rejected this further generali-
sation on the grounds that non-zero values of the additional coefficients
could not be reconciled with a generalised principle of least action such as his
physical theory required,33 but this defence was difficult to reconcile with
that he had given against Einstein's objection. While accepting that only the
14 coefficients might be important in practice (no physical quantities could
yet be associated with the other 26), Eddington argued in principle that
Weyl's geometry was neither one with an arbitrary metric (as he claimed was
his own version, with the 40 coefficients) nor one with a unique metric
corresponding to observed physical measurements (as was Einstein's), and
that it could therefore be defended on neither mathematical nor physical
grounds. 34 Eddington agreed with Weyl in seeing the a priori mathematical
theory as defming the possible laws of the world, and as providing a genuine
if metaphysical connection between these laws. But he also agreed with
Pauli that any account of actual material behaviour, defmed in this case by
the natural gauge corresponding to Einstein's theory, had to be based directly
upon observation.3s
Weyl's theory provided the context for the expression and exploration of
 PAULI AND A UNIFIED FIELD THEORY 13

fundamental differences of opinion as to what a physical theory could, and

should, achieve. The issues raised were not peculiar to the theory but, as the
disputants recognised, transcended the whole of physics. As we shall see, they
were to impinge in several respects upon the debate surrounding quantum
theory and its problems. And in one particular pOint, which was taken up by
the young Pauli, they bore directly on the quantum problem of wave-particle
duality.

THE CONTINUUM PROBLEM AND PAULI'S CRITICISMS

Although Weyl's work was stimulated directly by Einstein's, the prime

concern of the Gottingen mathematicians had not been with Einstein's
theory of gravitation but with a theory proposed by Gustav Mie in 1912.36
Mie had attempted to explain both gravitational phenomena and the existence
of the fmite electron on the basis of special relativity and a purely electro-
magnetic field theory. 37 And in 1915 Hilbert, combining Mie's approach to a
unified theory with the insights of Einstein and with his own mathematical
facility, had independently derived the general relativistic gravitational field
equations. 38 Born had also worked on Mie's theory while at Gottingen, and
so had Weyl, whose basic concern with general relativity was to develop
Hilbert's work into a fully-fledged unified theory, including an explanation of
electron structure. 39
The reconciliation of the corpuscular properties of the electron with the
continuous electromagnetic field theory was one of the most persistent and
difficult problems of classical physics, and had attracted considerable atten-
tion during the early years of the century. Mie's approach had been to seek
additional electromagnetic field terms that would be significant only ins~de
elementary particles, where they would combine with the traditional terms to
allow a stable distribution of electricity.40 Weyl, though incorporating
gravitational as well as electromagnetic terms, essentially followed this
approach and, as we shall see, based his physics upon an action principle
derived from Mie's work.41
The terms required were however too complex to be realistically incor-
porated into the field equations. The approach failed in practice, and in 1919
Pauli attacked it strongly on principle. 42 Since the theory was symmetric
with respect to positive and negative charges, he argued that it was contrary
to experience. 43 Since it incorporated no direct relationship betwee~ its two
basic entities, charge and mass, he argued that it was incomplete, in that it
could not account for the physically observed elementary particles. And most
14 CHAPTER 2

fundamentally of all he argued that as a pure field theory it could not in

principle offer any explanation of the cohesive structure of matter. As he put
it two years later, in his survey of relativity theory,44
The continuum theories make direct use of the ordinary concept of electric field strength,
even for the field in the interior of the electron. This field strength is however defined as
the force acting on a test particle, and since there are no test particles smaller than an
electron or hydrogen nucleus, the field strength at a given point in the interior of such a
particle would seem to be unobservable by definition, and thus be fictitious and without
physical meaning.
According to Pauli, any attempt at a complete unified theory would have to
account for the internal structure of particles, and would therefore have to
incorporate complex but well defined properties of matter independent of
and prior to those of the electromagnetic-gravitational field, which were
therefore operationally meaningful only on a scale large compared with that
of the elementary particles. But more important than this conclusion was the
philosophy behind it. For Pauli's requirement that a quantity, to be physically
meaningful, must be in principle capable of measurement had implications
beyond the narrow realm of Weyl's theory. In a wider sense too Pauli's
critique drew attention once more to the fundamental problem of the struc-
ture of matter (electrons) in a pure field theory and acted as a catalyst in the
fusion of this problem with the related one of the structure of light in the
electromagnetic field theory - the wave-particle problem of quantum theory.

DUALITY AND CAUSALITY

The strongest connection between the debate surrounding Weyl's theory and
the genesis of quantum mechanics was to be through the development of
Pauli's 'operationalist' ideas on the requirements of a physical theory.45 But
the two fields also interacted in other ways and it is instructive to look at
these, if only to establish the general potential for cross-fertilisation between
different branches of physics.
The greatest exponent of such cross-fertilisations was perhaps Einstein. He
had himself been long searching for a unified pure field theory, 46 and he now
treated Pauli's criticisms as applicable equally in respect of material particles
and light-quanta. Writing to Born in January 1920 he recognised that "Pauli's
objections apply not only to Weyl's but to any continuum theory, including
one which treats the light-quanta as singularities."47 Einstein had earlier
reacted to Mie's theory of electron structure by suggesting how the additional
terms of that theory might be replaced by purely gravitational terms connected
 PAULI AND A UNIFIED FIELD THEORY 15

with the solution of the cosmological problem. 48 This suggestion had run into
the problem that (as with Mie's theory) all spherically symmetric distributions
of charge seemed to be equally possible: there was no way in which the
particular size, charge and mass of the electron could be singled OUt. 49 But
Einstein did not give up hope of a pure field theory. In the letter to Born he
suggested that the extra determination necessary might be imposed on the
field by boundary conditions, and that both material particles and light
quanta might be derived from the continuous pure field by overdetermination
of differential equations. 50 The idea was still subject to Pauli's criticisms, but
Einstein made it clear that he could not accept these, and his continued
attempts to pursue the idea through the 1920s and to derive the light-quanta
from the continuous field may be seen both as rooted in the problem complex
of general relativistic field theory and as an assertion of his philosophical
stance of objective physical realism against Pauli's more subjective position. Sl
Another physicist to recognise an intimate connection between the light-
quantum and electron structure problems in this context was Eddington,52
who in 1921 took Pauli's ope rationalist criticism a step further by arguing
that the shape and size of an electron were observationally indeterminable,
since they could only be measured using another electron. S3 In practice, he
argued, one adopted the convention that the electron was spherically sym-
metric, and the mathematical expression of this convention corresponded to
the choice of natural guage, and to Einstein's gravitational field equations. 54
Despite this restriction on the observed world Eddington hoped that one
might get some further information on the possibilities of electron structure
from the 26 parameters of his theory that could not be related to the gravita-
tional and electromagnetic fields. 55 But writing to Weyl in July 1921 he
suggested that this programme was severely hampered by ignorance as to the
structure of light. Though his ideas were as yet "too vague to be formulated",
the devel.opment of relativistic matter theory would have to be closely
related to that of the quantum theory. S6
Weyl too saw a close connection between the material and light-quantum
problems, and in response to Pauli's criticism of a pure field theory he suggested
that there might be an element of reality prior to the field, corresponding to
acausality on the quantum level. In May 1919 he wrote to Pauli that he
thought the non-equivalence of positive and negative electricity would reduce
to that of past and future, to the unidirectionality of time. 57 As to his
advocacy of a pure field theory, Pauli should not accuse him of being a
dogmatist: he did not think he'd found the philosophers' stone, and he was
himself qUite sure that there was something in matter independent of the
16 CHAPTER 2

field. 58 Writing again in December, he approached the problem of deriving

the world of physical observation from that of his predominantly mathe-
matical theory, and suggested that the past-future distinction was more basic
than that between the two signs of electric charge, and that it completely
determined the latter through the world geometry. The process of events
represented by this geometry was however fundamentally statistical and
acausal, and it was the origin of this acausality, in terms of "independent
decisions", that formed the existent in matter independent of the field. 59
From Weyl's highly metaphysical perspective it was the independent decisions
that were 'real', and not the physical field; but it was the field that constituted
what might be called the physical reality, the reality of the observed world. 60
Weyl did not explain just how the independent decisions operated to produce
the observed phenomena, but in the fourth edition of Raum-Zeit-Materie,
completed in 1920, he linked them with the apparent acausality ~f quantum
phenomena. 61 And in the letter to Pauli he suggested that the complex
structure of the electron to which they would lead might account for the
quantum phenomena of the atom: 62

I believe, for example, that the fact that the electron does not radiate in the stationary
Bohr orbits is an indication of the acceleration produced through the internal changes
of the electron, which enables it to retain its energy; then why should it also in non-
quasistationary acceleration behave as a rigid body with fixed charge?

Weyl's response to Pauli's objections may be seen, as by Forman, as reflecting

a hostility towards causality pervasive in the prevailing intellectual milieu. 63
It was certainly based on philosophical as much as physical considerations,
and in a paper of 1920 Weyl expressed his views on the rejection of causality
in a thoroughly philosophical framework. 64 But these views were not without
their merits, even on more restricted, purely scientific, grounds. The problem
of the asymmetry between the elementary particles was one that Weyl had
already tackled without success in the context of Mie's theory.65 And Pauli's
analysis had shown that an asymmetric theory of the type he had then
proposed would inevitably fail to be invariant under time reversal, an important
conclusion that must have been partly responsible at least for his identification
of the problem with that of the unidirectionality of time. 66
Turning back to the quantum context we should note also that Einstein's
attempt at a causal pure field theory was not only subject to Pauli's methodo-
logical criticisms but also had every appearance of being untenable on empirical
grounds. Einstein himself found all his attempts to derive the light-quanta
from a continuum theory foundering on the necessity in this case of abandoning
 PAULI AND A UNIFIED FIELD THEORY 17

strict energy conservation in order to allow for the discrete absorption of

light, and on the resultant possibility of an infinite energy arising from the
gradually increasing Brownian motion made possible by energy non-conserva-
tion. 67 He had indeed demonstrated in 1911 and 1916that the only alternative
to the light-quanta, within the existing conceptual framework, was the
rejection of energy conservation. 68 And although he seems to have been
hopeful that by deriving the light-quanta from a continuum theory he might
somehow retain the energy-momentum conservation, he could draw on
nothing whatsoever to support this hope.
Einstein's paper of 1916 was a critical one for the future of quantum
mechanics, and will often be referred to. In it he introduced for the discrete
quantum behaviour of the atom probabilities of the absorption, free emission
and stimulated emission of radiation in the presence of a field. These corres-
ponded to the transition intensities of the classical continuum theory and
were defined so that the total absorption and emission over a period of time
would be the same on the two theories. Using the new terminology, Einstein
offered a newaerivation of Planck's Law and, in a second section of the
paper, he also put forward convincing arguments to the effect that, assuming
the conservation of energy, the emission or absorption of light by an atom
was always accompanied by a transfer of linear momentum as well as energy.
This result provided strong confirmation of the particulate aspect of the
nature of light.
Following this paper of Einstein's it became widely accepted that the
mildest possible modification to classical electromagnetic theory if quantum
effects were to be explained would be the abandonment of energy conserva-
tion;69 and to extend this to the abandonment of causality, as did Weyl and
later Schottky, SchrOdinger and Von Mises, was not at all unreasonable.
Given a preference for a field over a particle theory it was indeed the only
path that seemed to offer any hope for advance, and even Weyl's meta-
physically-based introduction of independent decisions must therefore be
associated with the fundamental wave-particle and field-particle problems. 70
I t should by now be apparent that the issues of the search for a unified field
theory and those of quantum theory, whether in the form of conservation,
causality or duality, could not be taken in isolation. In particular both Weyl's
rejection and Einstein's defence of causality, though expressed in the
quantum context, were deeply rooted in the problems of a unified general
relativistic field theory. The same is true also of Schottky's rejection of
causality in 1921, which was closely related to Einstein's WO.rk.71 So far as
the actual development of quantum mechanics was concerned, Weyl and
18 CHAPTER 2

Schottky played relatively minor parts (though Weyl did contribute on a

technical level to both matrix mechanics and wave mechanics). But Einstein's
ideas were to be of the greatest importance for the genesis of wave mechanics,
and Schrodinger himself, the author of that theory, also appears to have
drawn on the general relativistic context.
Schrodinger joined Weyl in Zurich in 1921,72 and the following year he
published two papers, both of which bear the stamp of Weyl's ideas. In 1920
Forsterling had sought to reconcile Bohr's quantum condition, E = hv, with
the special and general theories of relativity; but his treatment had been
criticised, by Pauli once more, on the grounds that it contradicted Einstein's
result on the linear momentum (particulate light-quantum nature) of emitted
light. 73 In June 1922 SchrOdinger rederived the reconciliation with special
relativity, apparently upholding the light-quantum hypothesis. 74 But the
following November, responding to further criticisms by Pauli, he explained
that he now interpreted Einstein's result as applicable only to the recoil of
the emitting atom, with which Einstein's calculations had in fact been
concerned, and not to the light itself.7s For he stressed that he saw no way
out of the quantum paradox other than to abandon energy-momentum
conservation on the atomic level. And he put forward his own suggestion that
one might replace the quantum identity, E = hv, by an equivalence between
the effect produced by a material particle of energy E and that produced by a
(spherical) wave offrequency E/h, but without the energy E necessarily being
attributed to the wave, and thus without the wave having to be localised as a
light-quantum. This would not be possible for a conservative mechanical
system of course; but "the devil knows", he argued, whether atomic systems
are strictly causally determined, let alone whether they are conservative
mechanical systems. 76 A few weeks later, in an inaugural lecture, he explicitly
rejected the "rooted prejudice of causality". 77
Writing to Pauli, Schrodinger insisted that their differences were purely
philosophical,7s and his rejection of causality was explicitly a philosophically
based decision rather than one dictated by the requirements of physics. 79
In 1924-5 he changed his views on both light-quanta and causality and moved
closer to Einstein's position,80 but in contrast with Einstein he maintained
his metaphysical bent, and he read de Broglie's These in 1925 not as offering
a straight-forward fusion of wave and particle theories (which was more or
less how it had been put) but as revealing the "inner, spiritual" connection
between the two theories. 81 This perspective must have brought him dose to
Weyl, with whom he worked closely in 1922, and it would seem likely that
his rejection of causality was linked with Weyl's. His second paper of 1922,
 PAULI AND A UNIFIED FIELD THEORY 19

completed in October, was explicitly based upon some of Weyl's work. 82

In this paper Schrodinger again pursued the programme of reconciliation
initiated by Forsterling, and following up the lead given earlier by Wey1 him-
self he attempted to derive the Bohr atomic orbits directly from Weyl's
unified field theory.83 His efforts bore no immediate fruit, but as Raman,
Forman and Hanle have shown, this work appears to have played a crucial
part in his later creation of wave mechanics. 84

THE DEVELOPMENT OF PAULI'S VIEWS

Important as it was, we shall return to Schrodinger's work only briefly, for it

was in the development of matrix mechanics that the flow of ideas found
clearest expression, and in this context the link with Wey1's work arose
through the developing ideas of Pauli.
We have referred to Pauli's criticisms of both Weyl and SchrOdinger, and
have described his philosophical position loosely as 'operationalist', a term
that was coined by Bridgman in 1927 and is also used sometimes of Edding-
ton. But what more precisely was Pauli's position and how did it translate
into the quantum-theoretical context? One outcome of the debate over unified
field theories was the explicit development of Eddington's natural philosophy,
which found its first extended expression in the closing sections of his book
Mathematical Theory of Relativity, in which both Weyl's theory and his own
extension of that theory were presented. 85 And it was in response to this
book, and in a letter to Eddington, that Pauli was prompted to give the most
detailed account we have of his own position. It was already clear from his
publications that this entailed some form of methodological or epistemologi-
cal operationalism, summed up in the statement from his 1921 survey that
"we should hold fast to the idea that in physics only quantities that are in
principle observable should be introduced."86 But in the letter to Eddington
the status of this demand was clarified and the demand itself expanded and
carried into the quantum context.
So clearly did Pauli express himself in his letter to Eddington, and so
crucial were his views to be to the development of quantum mechanics, that
it is worth quoting them directly and at length. Having first discussed the
formal connection between electromagnetism and gravitation in Eddington's
theory, Pauli restated the limitations of a pure field theory such as the
classical electromagnetic theory:87
The most unsatisfactory thing about the electron theory of Lorentz and Larmor is
indeed that it is not a theory of the electron, as one might somewhat jokingly say. Why
20 CHAPTER 2

are there only two kinds of elementary particle, negative and positive electrons (Hydrogen
nuclei)? Why are their electric charges the same, when their masses are so completely
different? Generally, how can an elementary particle stay together despite the Coulomb
repulsion between its parts? This cannot happen according to the Maxwell-Lorentz
theory, nor can either the energy or momentum of a uniformly perturbed electron be
calculated in its entirety on the basis of this theory of the electromagnetic field. The
reduction of the whole inertia to the electromagnetic field also does not succeed, so
long as we hold to this theory.

He then reaffirmed his belief that the problem of the electric elementary
particles could not be solved by any theory using the concept of continuously
varying field equations, satisfying known differential equations, for the region
inside the elementary particles.
Pauli next turned to quantum theory, and used the problems in that con-
text to expand upon his viewpoint:
As is shown in quantum theory, classical electrodynamics also fails in the domain in
which dimensions are large compared with those of the elementary particles, as soon as it
treats swiftly varying fields. It holds in this domain only when the fields are static or
quasistatic, as for example when an electron is moving with constant velocity. For the
other cases it leads to the right answer only in respect of the statistical average over
many individual processes within a certain range.
I might emphasise especially that the quantum theory in no way calls for only a
modification of light theory, but generally calls for a new definition of the concept of
electromagnetic field for non-static processes. Indeed the classical theory fails also in the
description of electron-atom collisions, as well as for the behaviour of the electron in the
atom in time-varying fields (dispersion theory). The famous contradiction between the
interference capability of light emitted in various directions and its property of remain-
ing constantly disposed in energy quanta (Einstein's "light-quanta"), for example in the
photoelectric effect, comes only from the fact that we give up the laws of classical
theory, but still always work with the concepts of that theory.

Thus in Pauli's view the wave-particle duality was neither to be dismissed (as
many more conservative physicists still maintained) nor to be attributed to
any amendment of the existing theory, such as the absence of causality or
energy conservation. Its origins were rather more fundamental, lying in the
fact that the concepts being used in physics were themselves inappropriate to
the situations concerned. Just how they were inappropriate, Pauli went on to
explain:
We now know both from experimental results and from quantum-theoretical considera-
tions, which I shall not however go into here on account of brevity, that even an electron
in a light wave does not in fact move at all as specified [by the classical theory). The
field of a light source is therefore at present undefmed. But the path of an electron (in
general in a swiftly varying field) is just as little so. Indeed one can only define this
 PAULI AND A UNIFIED FIELD THEORY 21

through the action of the electron on other elementary particles, and in a non-static
field this again fails to satisfy the classical laws. Some physicists have fallen upon the
idea, considering the above-mentioned paradox of light, of completely abandoning the
concept of the field of a light wave and only describing the motion of the electron.
This we wholly reject; for on the one hand the idea scarcely helps us over the difficulty,
and on the other hand it makes a connection with the classical theory, which has proved
itself so brilliant for large scale phenomena, quite impossible. We can proceed only
through a modification, not through an abandonment of the field concept. So much
would the return to the pre-Maxwellian standpoint be considered a retrogression, so
much must we hold before our eyes the fact that the field concept only has a meaning
when we specify a reaction, which is in principle possible, in which we can if we want
to measure the field strength at each point of space-time. We can attribute a reality to
the field intellectually, even if we do not exactly execute the reaction. It is only essential
that we can always execute it in principle, if we want to. But as soon as the reaction
ceases to be always specif"table or in principle executable the respective field concept
is no longer defined. *

In a footnote, Pauli illustrated this point by an example:

* It has often been said that one could retain the Maxwell equations in the vacuum
(charge-free space). Only the interaction between light and material systems could not
be treated with the help of the classical theory. In fact the two are inseparable. Only by
means of this interaction can the electromagnetic field of a light wave be defined. The
Maxwell equations in a vacuum are neither correct nor incorrect; they are meaningless
in every application where the classical theory fails and where it is not a question of
statistical averages.

If we replace "reaction" by "operation", then Pauli's position is very close

to that later adopted by Bridgman. Only whereas Bridgman was concerned
clearly with epistemology, Pauli's concern oscillates between a form of
effective epistemology - what is "meaningful" - and methodology. Perhaps
under the influence of Bohr, under whom he was working at the time, Pauli
was concerned not only with building up a system of concepts from scratch,
but also with a correspondence between the new concepts and those that had
proved themselves valuable in the classical context. He was certainly deeply
concerned that one should seek, within the limits set by the operational
definition of concepts, to go behind the phenomena, to provide an expla-
nation not just an account. And he had some ideas too on the direction in
which the new concepts required should be sought:

In the definition of the electromagnetic field strength in the classical theory the reaction
concerned at present is the force on a charged test particle. This means that already in
the defmition of the electro-magnetic field strength we use mechanics (force and mass
concepts). Already therefore the reduction of mechanics to electrodynamics is merely
22 CHAPTER 2

apparent. It would be much more satisfying if the measurement of the field strength
could be based exclusively on a counting process, without making use of mechanical
concepts. Instead of introducing discontinuous functions one would perhaps associate
suitable of these with continuous functions; and as the laws of relativity theory are
invariant with respect to coordinate transformations, so would the laws of the future
quantum physics perhaps also be invariant with respect to alterations of the field func-
tions, so long as these left unchanged only certain counting process results.
To summarise, may I say this: we do not yet know today which fast-varying electro-
magnetic processes we can in principle observe, and which not. As soon as we know
this, we shall have solved the quantum riddle. Because of the relationship between
the dermition of electromagnetic field strength in the classical theory and mechanics,
I believe that the same dissociation of the field concepts from mechanics, with the
simultaneous introduction of some kind of atomism (countable quantities instead of
continuous ones) will bring the understanding of the quantum problem and of the
elementary electric particles.

In the last passage, Pauli made two specific recommendations that were
indeed to be relevant in the development of quantum mechanics. Insisting
on an operationally based definition of any concepts used in the new theory,
he suggested that this would entail the removal of the existing mechanical
content of the field concepts, and the replacement of continuous concepts
by discrete ones. This last point had obvious implications for the quantum
problem, and it was closely linked with Pauli's earlier rejection of pure
continuum theories and his repeated defence of the particle aspect of the
wave-particle duality as being if anything the more fundamental. Since our
observations of the physical world consist of instantaneous and localised,
effectively discrete measurements, then the concepts used to describe that
world had, according to Pauli's principles, to reflect that discreteness.
In his long letter to Eddington, Pauli next returned to the original context
of Eddington's theory, and to his conviction that the connection therein
between the electromagnetic and gravitational fields had no physical signifi-
cance. In the course of discussing this very difficult problem, which he saw
as "rooted in the quantum theory", he extended his discussion to suggest
that, in the absence of any solution to the quantum problem, one could
legitimately proceed "phenomenologically", that is without any insight
to the fundamental nature of the elementary particles:

We do not know a priori whether it will be possible to answer this question without
regard for quanta. It can indeed be the case. But then the question must in my opinion
be answered purely phenomenologically, without regard to the nature of the elementary
electric particles. Further, I stand by my position (naturally not demonstrable) that
this question must begin with a dermition of the field quantities used, which specifies
 PAULI AND A UNIFIED FIELD THEORY 23

how the quantities can be measured. It must further disclose relations between electro-
magnetic and otherwise measured effects. (The greatest achievement of relativity theory
was indeed to have brought the measurements of clocks and measuring rods, the orbits
of freely falling mass points, and those of light rays into a rum and profound union.)
Logically or epistemologically, this postulate cannot be demonstrated. But I am con-
vinced that it is correct.

Pauli had already criticised the adoption of what might be called a negative
or passive phenomenological approach, but he now indicated how such an
approach might be positively used, in the absence of any sound foundation
for a theory, to explore the relationship between parts of that theory. If the
discrete and non-mechanical conceptual foundations he had advocated
could not yet be found it was important above all to ensure that those
concepts that were used, though they could not of course provide a unified
base for the whole theory, were at least operationally well defmed; and that
the relationship between their operational definitions was explicit. Then one
could at least proceed, albeit only on a phenomenological level, without
danger of contradiction. Pauli's immediate concern was that neither Weyl's
theory nor Eddington's satisfied these conditions, and they were indeed
far from trivial. But once again the perspective was one that was to be of
continuing use to Pauli himself in the course of the development of the new
quantum mechanics.
 CHAPTER 3

NIELS BOHR AND THE PROBLEMS OF ATOMIC THEORY

INTRODUCTION

Having completed his survey of relativity theory in 1921, Pauli switched

his concern to quantum theory. 1 He worked first as Max Born's assistant
for the academic year 1921-1922, sharing in Born's own first attack on
the problems of the atom. By the time he wrote the letter to Eddington
in September 1923 he had also worked for a year in Copenhagen with
Bohr. 2 Neither of the physicists with whom he worked most closely in this
period, Bohr or Heisenberg, had played any active part in relativity theory
or the search for a unified theory. Indeed Bohr, uniquely among the great
theoreticians of the period, appears to have displayed little interest in unifi-
cation and little if any sympathy with relativistic thinking and analysis.
But despite this difference, the new environment was far from unconnected
with the old. Born started work on quantum theory at about the same time
as Pauli, at the urging of the experimentalist James Franck, who was his
colleague at Gottingen and a close ally of Bohr's. 3 Before then his orbit
had been much closer to the problems of relativity theory than to those
of quanta, and he had, moreover, shown some sympathy with Pauli's ideas. 4
Heisenberg had been Pauli's colleague as a student in Munich under Sommer-
feld, and had already then discussed the fundamental problems of physics
with him. Sommerfeld himself was part author and leading exponent of the
quantum theory, but equally at home with relativity theory, and again of
course familiar with Pauli's ideas. But of all his new colleagues the one who
provided the greatest stimulus to Pauli's thought, and whose ideas had most
relevance to his own, was the one whose work had been most removed from
his, the father of quantum spectral theory, Niels Bohr. Though developed in
a completely different context Bohr's ideas applied to the same fundamental
problems as did Pauli's - and were completely opposed to them. As in the
earlier discussions of unified field theories the argument was conducted with
friendship and respect. Pauli never ceased to admire Bohr as a truly great
physicist and as the teacher who made the greatest impression upon him. 5
Bohr was delighted to find in the young Pauli a worthy if troublesome
opponent, the soundness of whose judgement he could trust completely. 6
24
 BOHR AND PROBLEMS OF ATOMIC THEORY 25

But argument there was, for the division between their views was a deep
one, and if Pauli's position was based on a strong conviction Bohr's, devel-
oped over a longer period and greater experience, was even more so.

THE CONFLICT OF BOHR'S ATOMIC MODEL

Like the quantum mechanics it preceded, Bohr's theory of the atom appeared,
and by any reasonable standards was, a success. But from its first enunciation
in 1913 the theory's foundations were problematic. Bohr had early become
convinced that in the solution of the quantum problem classical mechanics
would have to be superceded, and his model of the atom was explicitly at
variance with this theory, which could not accommodate the discrete transi-
tions between stationary states. It was also at variance with classical electro-
dynamics, which could not account for the absence of radiation in these
states. But at the same time it was framed entirely in the language of, and
rested entirely upon the foundations of, these classical theories. The concept
of electrons orbiting the atom was a classical mechanical concept, and as the
Bohr theory developed the choice of possible orbits continued to be governed
entirely by the requirements of the classical theory. Bohr's orbital model
of the atom was explicitly a model, of course, and Bohr himself was the first
to insist that it could not be interpreted as in any sense structurally true. 7
Rather as in mediaeval planetary theory, the details of the atomic model
were varied at leisure, within the basic form, so as to produce the required
observed results. But still the model was in some sense intended, and taken,
as real. It was not intended, as for example were J. J. Thomson's models,
as a merely heuristic device directing the way to further investigation, but
rather as a genuine if inadequate image of reality. Bohr sacrificed elements
of the classical mechanics and electrodynamics, but he did not depart from
the basic conceptual entities of those theories. Apart from the fundamental
but conceptually undefined postulate governing the energy differences be-
tween stationary states, E = hv, he did not introduce anything to replace
what had been lost. And so the Bohr theory entailed a fundamental conflict,
as on the one hand it departed from the classical theories radically, while on
the other hand it remained both technically and conceptually dependent
upon them.
As the theory developed, this conflict heightened. Since the structure of
the model took second place, as it had to in the absence of any consistent
foundation, to the results derived from it, the details of this structure were
varied at leisure. Just as astronomers of old had adjusted their epicycles
26 CHAPTER 3

to agree with any new data, so the quantum theorists, notably Sommerfeld,
Lande and their colleagues, adjusted the choice and relative configuration
of electron orbits at will in trying to apply the model to higher elements. 8
They did not concern themselves much with why this or that orbit should be
preferred, let alone with why any such choice made sense, given the admitted
failure of the foundations upon which the model was based. Their only
criterion was whether it would give the observed spectra. With the introduc-
tion of quantum numbers to describe the stationary states this trend was
exaggerated, for although the numbers were originally descriptive of physical
properties they eventually came to precede these properties, with the choice
of a new quantum number, i.e. a new degree of freedom, preceding the de-
bate as to what this number represented physically. The positivistic attitude
in the choice of orbits was thus continued, and strengthened, in the treatment
of quantum numbers.
But still the theory rested upon the very classical concepts it spurned.
The orbits were restricted to those that, given the absence of any continuous
radiation from the orbiting electrons, were possible according to the classical
theory. The quantum numbers, to be accepted, had ultimately to be inter-
pretable in terms of the classical concepts. Even if it were recognised that the
planetary model was too naive to be a genuine structural representation,
there was simply no alternative visualisation. This situation became even more
acute, moreover, when in the early 1920s the theory progressed from the
mere prediction of transition frequencies to take into account also the
intensities of the emitted and absorbed radiation of each frequency. The
extension was made possible in 1916, when Einstein introduced the notion
of transition probabilities as the quantum equivalent of the intensities of fhe
corresponding radiation on the classical theory. 9 But it was possible only
through such a connection, justified conceptually by Bohr's correspondence
principle of 1919.10
Since the Bohr atom transitions were discrete, and since there was no
known mechanism governing when they took place, Einstein had seen that
the classical intensities of continuous emission and absorption could only
be replaced on the quantum theory by probabilities. In order to provide a
theoretical derivation of these probabilities Bohr then argued that there must
be a "necessary connection" between the quantum and classical theories
"in the limit of slow vibrations" (hv small). He then assumed that such a
connection also held in the slightly wider limiting range of high quantum
numbers, and, more generally still, wherever the classical theory gave what
seemed to be the right answer. The correspondence principle was based on
 BOHR AND PROBLEMS OF ATOMIC THEORY 27

this connection, which was made through the quantum theory of condi-
tionally periodic systems in the limit in which these gave frequencies that
could be associated with those of classical oscillators. Since, however, the
only cases of physical importance tended to be those in which the frequencies
could not be so associated, the principle tended to be used in isolation from
its theoretical foundations, as a blanket excuse for carrying over empirical
results from the classical to the quantum theory. 11
Bohr was of course well aware of these problems, and also of the closely
related problem of the wave-particle duality, clearly manifest in Einstein's
1916 paper. The dominant theme of that paper was that of light-quanta.
Einstein's analysis of the momentum exchange in radiative processes sup-
ported the light-quantum hypothesis, and his introduction of transition
probabilities opened the way for the description of such processes in terms of
discrete quanta. At the same time, however, his accompanying derivation of
Planck's law, though apparently based on the light-quantum hypothesis, in
fact included a limiting process that was valid only for the wave theory of
light. Moreover Einstein's hypothesis of stimulated emission in the presence
of a radiation field, which was also necessary for his derivation of Planck's
law, could be explained naturally only in terms of wave resonances and not
in terms of particle mechanics. 12 Bringing the same considerations to bear
on this complex of problems as on those of the atomic model, Bohr came
to a series of fundamental and closely related conclusions.

THE DEVELOPMENT OF BOHR'S POSITION

As we have noted, Bohr's atomic model represented a departure from both

classical electrodynamics and classical mechanics. But although the former
departure appears to have exercised his mind the more by the 1920s, it
was the latter that was most central to his theory, and with which he had
originally been primarily concerned. Indeed, while he had from the beginning
aimed to depart from classical mechanics, Bohr had not actually sought
to depart from classical electrodynamics. He had, on the contrary, a very
strong belief in the power of that theory, 13 and even after 1913 he argued
persistently in its favour, trying to retain as much of it as possible. Thus,
although his atomic model appeared to imply the necessary discrete absorp-
tion of light, and hence its localisation in light-quanta, Bohr could not accept
this implication, but believed as did many others that the demonstrated
success of the classical electrodynamical wave theory of light argued con-
vincingly for its retention. Over the years he spoke out gradually more
28 CHAPTER 3

strongly in favour of the wave, and against the particle concept of light. In a
survey of quantum theory written in 1918, he adopted Einstein's transition
probabilities but pointedly ignored the arguments for light-quanta that had
accompanied the derivation of these probabilities in Einstein's paper. 14 In
1920, he wrote that "I shall not here discuss the familiar difficulties to which
the hypothesis of light-quanta leads in connection with the phenomenon of
interference, for the explanation of which the classical theory has shown
itself to be most remarkably suited."1s And the same year he explained to
his close colleague Darwin that since one could only define a frequency
through a wavelength, and a wavelength through interference, and interference
through the wave concept of radiation, this concept had to be retained. 16
In 1921, at the Solvay congress of that year, he did discuss the light-quantum
concept for the first time in a published paper, and concluded that it "presents
apparently insurmountable difficulties from the point of view of optical
interference." 17 And the following year, in the course of the most clear
and extensive account of quantum theory to date, he brought together his
criticisms in a sharp attack on the light-quantum hypothesis: 18

As is well known, this hypothesis introduces insuperable difficulties, when applied to the
explanation of the phenomena of interference, which constitute our chief means of
investigating the nature of radiation. We can even maintain that the picture, which lies
at the foundation of the hypothesis of light-quanta, excludes in principle the possibility
of a rational definition of the conception of a frequency II, which plays a principal
part in this theory.

Apart from Sommerfeld, who wrote to Bohr in 1918 of his conviction that 19

The wave process occurs only in the aether, which obeys Maxwell's equations and acts
quantum-theoretically as a linear oscillator with arbitrary eigenfrequency II. The atom
merely furnishes a definite amount of energy and angular momentum as material for
the process.

and who continued to follow this line until about 1923, Bohr was the only
one of the group working on the quantum theory of spectra to hold such
firm views on the wave-particle duality. In this respect his position bore
more resemblance to those of the physicists on the fringe of quantum theory
than to those in its midst. 20 Unlike many of those physicists, however, Bohr
did not underestimate the force of the arguments for light-quanta. On the
contrary, his opinion was considered, well supported, and far from lacking
in subtlety. In respect of both the wave-particle duality and the atomic
theory he had sought hard for a more satisfactory solution. And although he
 BOHR AND PROBLEMS OF ATOMIC THEORY 29

had failed to find such a solution he had succeeded in tracing the problem
back to the concepts used to describe the physical world.
Whereas Pauli was to call for a radical revision of the concepts of physics so
as to ensure their consistency both with each other and with the operations
used to defme them, Bohr came to the conviction that such a replacement
was impossible. The concepts of classical physics were demonstrably insuf-
ficient to describe the processes of physics, but their choice seemed to Bohr
to be psychologically as well as physically determined, and there seemed to
him to be no alternative to them. In his survey paper of 1922, which was
written and may be read as his manifesto of the period, he admitted that
"a description of atomic processes in terms of space and time cannot be
carried through in a manner free from contradiction by the use of concep-
tions borrowed from classical electrodynamics." But in the opening paragraph
of the very same paper he insisted that "from the present point of view of
physics, however, every description of natural processes must be based on
ideas which have been introduced and defmed by classical theory." 21 And as
is clear from his attitude to the wave-particle duality, Bohr saw the concepts
of classical wave theory, of classical electrodynamics, as peculiarly funda-
mental in this respect.
In purely philosophical terms, Bohr was not troubled by this situation.
As he explained in September 1923 to his philosophical friend and mentor
H~ffding, he saw no reason why atomic science, or anything else for that
matter, should conform completely to the visual world of our perceptions
and of the classical physical conceptions: 22

In general, however, and particularly in some new fields of investigation. one must
remember the obvious or likely inadequacy of pictures: as long as the analogies show
through strongly, one can be content if their usefulness - or rather fruitfulness - in
the area in which they are being used is beyond doubt. Such a state of affairs holds not
least from the standpoint of the present atomic theory. Here we find ourselves in the
peculiar situation, that we have obtained certain information about the structure of the
atom which may surely be regarded as just as certain as anyone of the facts in natural
science. On the other hand, we meet with problems of such a profound kind that they
seem to defy solution; it is my personal opinion that these difficulties are of such a
nature that they hardly allow us to hope that we shall be able, within the world of the
atom, to carry through a description in space and time that corresponds to our ordinary
sensory perceptions.

However, while this may not have been philosophically disturbing it did
have very serious consequences for the development of physics. The key
problem for Bohr was how to overcome the conflict he had identified, and
30 CHAPTER 3

in particular how to preserve the continuity both of and with the classical
conceptions. Although even his own correspondence principle rested neces-
sarily upon Einstein's considerations of 1916, he described these as providing
"only a preliminary solution", and throughout his 1922 survey paper he
treated the discreteness of quantum theory not only as its defming charac-
teristic but also as its major problem. The task as he saw it was to reconcile
the discontinuities of quantum phenomena with the continuous nature
of our natural understanding, manifest in the classical continuum electro-
dynamics, and to do so by effecting a continuous transition from the old
theory to the new.
Setting about this task, Bohr proceeded in two directions. On one hand
he sought to extend the range of the classical concepts into non-classical
phenomena, and this he did through his famous correspondence principle.
At the time of its enunciation the status of this principle was still somewhat
obscure. But Bohr stressed in his survey paper that it was intended as far
more than a mere link between the quantum and classical theories and that
it, and the classical theory of electrodynamics itself, must be treated logically
as part of the foundations of quantum theory. This was consistent with the
dependence of his quantum theory upon results that could not be obtained
without reference to the classical theory, and it also fitted in with the status
he afforded to the classical concepts. On the other hand, Bohr also sought
to limit the range of these classical concepts so as to allow completely non-
classical phenomena to enter into the theory without leading to serious
contradictions. Here he pursued two related possibilities, the abandonment
of strict energy-momentum conservation, and the abandoment of a causal
description in space and time.
We have already noted in the context of SchrOdinger's work that Einstein
had mooted the possibility of energy non-conservation - a possibility that
he himself had found totally unacceptable - in 1911. At the first Solvay
congress he had argued that the only alternative to acceptance of the light-
quantum hypothesis was to "resort to abandoning the law of conservation
of energy in its present form, giving it for example only a statistical kind of
validity, as one does already for the second principle of thermodynamics."23
For the particular case he was examining, he stressed that "one can only
choose between the structure of radiation and the negation of an absolute
validity for the law of conservation of energy."24 Einstein himself was
not prepared to sacrifice energy conservation, a position he justified later
by its implications for causality and by consideration of a possible infmite
Brownian motion that might result in this case. 25 He preferred to keep
 BOHR AND PROBLEMS OF ATOMIC THEORY 31

both the light-quanta and "the indispensable Maxwell equations",26 hoping

to derive both in due course from a unified field theory. But as we have
noted before, this position was very difficult to uphold, and the pressures
in support of the alternative of energy non-conservation were therefore
strong. Both Bohr's theory of the atom and Planck's 'standard', 1911, deri-
vation of his black-body radiation law seemed to require discrete changes in
the energy content of matter at the same time as their authors upheld a
continuous theory of the transmission of energy, and the abandonment of
strict conservation was the most-natural response to this situation. 27 Such
at least was the view of the British physicist Darwin, who wrote to Bohr
at length on the subject in 1919,28 and in apparent reply to this letter Bohr
agreed that "on the quantum theory, conservation of energy seems to be
quite out of the question." 29 His own feeling at that time was that something
funny must go on inside the atom, triggered by the incident light. 30
In the paper to the 1921 Solvay congress in which Bohr first came out
strongly against the light-quantum concept, he nevertheless noted that this
concept seemed "to offer the only possibility of accounting for the photo-
electric effect [one of the phenomena discussed by Einstein in 1905 as
characteristic of particulate light-quanta], if we stick to the unrestricted
application of the ideas of energy and momentum conservation."31 In a
manuscript of the same year that did not reach publication, Bohr repeated
the same statement verbatim, but followed it up with the comment that 32

At this stage of things it would appear that the interesting arguments brought forward
recently by Einstein [those in support of the light-quantum concept in his 1916 paper]
... rather than supporting the theory of light-quanta will seem to bring the legitimacy
of a direct application of the theorem of conservation of energy and momentum to the
radiation processes into doubt.

In his survey paper of 1922 Bohr insisted simply that "a general description
of the phenomena, in which the laws of the conservation of energy and
momentum retain in detail their validity in their classical formulation, cannot
be carried through."33 In support of this contention he cited with approval
Schrodinger's treatment of Einstein's analysis of momentum exchange
during radiative processes as applying only to the recoil of the atom and
not necessarily to the light, and he pointed out that because of the relative
mass of the atom any such recoil would be of small velocity and effectively
unobservable.
One problem with the abandonment of energy conservation was its im-
plication for the causality principle. There was no a priori reason why a
32 CHAPTER 3

stati~tical energy law, or systematic changes in energy, should not have been
incorporated in a deterministic physics much as was the statistical entropy
law. But this would not have been possible without some refmement of either
the energy concept or the structural conception of matter: it would have
been necessary for example to keep track somehow of the deviation of
energy from its statistical norm. The simplicity of the energy concept and the
conservation principle gave classical physics a large measure of its security,
and the abandonment of energy conservation was therefore likely to be seen
by many as tantamount to, either logically (within the existing conceptions)
or at least in the sense of being as bad as, an abandonment of causality.
This appears, for example, to have been true in Einstein's case,34 and the
connection may also be seen in Schrodinger's work as discussed in the last
chapter. Planck's use of an a priori emission probability in his 1911 derivation
of the black-body law and Einstein's introduction of transition probabilities
in 1916 both had to be accompanied by clarifying statements to the effect
that causality was somehow, though how they could not tell, maintained. 35
The absence of any mechanism for the discrete changes in eithe( Planck's
theory or Bohr's cried out for the abandonment of causality and the resort
to a purely statistical theory. Moreover, beyond the realms of the quantum
theory itself there were also strong ideological pressures in the same
direction. 36
Bohr is unlikely to have been swayed by the anticausal pressures of the
German Weimar milieu, strong as these seem to have been, but he was acutely
conscious of those within quantum theory, and he also appears to have
been influenced by aspects of Danish philosophy. He himself talked of such
an influence in respect of his principle of complementarity, 37 and there are
striking parallels between the exposition of his atomic model and that of
some aspects of Kierkegaard's philosophy, to which he was strongly drawn. 38
His philosophical mentor, H~ffding, had made the rejection of causality an
explicit part of his system. 39 It was not surprising that Bohr himself should
move in this direction. But whereas the abandonment of energy conservation
seemed to Bohr to offer a freedom that was at once sufficient to allow
the incorporation of non-classical behaviour in the theory and at the same
time restricted enough, given for example statistical conservation, to allow
the theory to retain some positive predictive potency, a mere abandonment
of causality did not have this virtue. It gave up too much. And in these
circumstances Bohr's thoughts seem to have run along lines, perhaps tied
in with his philosophical background, that anticipated in some respects
his later principle of complementarity. Some of his colleagues rejected the
 BOHR AND PROBLEMS OF ATOMIC THEORY 33

causality principle outright. Others talked of the impossibility of a consistent

description in three-dimensional space.40 But Bohr, recognising that these
factors were intimately linked but that neither a spatial description nor a
causal description could be sacrificed altogether, moved only tentatively
toward the idea that they might be incompatible. He referred to the role
of probability in determining the atomic transitions, but always qualified
his reference as applying "in the present state of the theory." In his 1922
survey he piled caution upon caution, stating only that 41
In the present state of the theory, it is not possible to bring the occurrence of radiative
processes, nor the choice between various possible transitions, into direct relation with
any action which finds a place in our description of phenomena, as developed up to the
present time.

It was only in a manuscript dating from the Winter of 1923-1924, by when

the pressures had increased considerably, that he was even so bold as to
suggest that, in relation to the wave-particle duality, 42
It is more probable that the chasm appearing between these so different conceptions
of the nature of light is an evidence of the unavoidable difficulties of giving a detailed
description of atomic processes without departing essentially from the causal description
in space and time that is characteristic of the classical mechanical description of nature.

Bohr was cautious. But there can be little doubt that he was already heading
towards the abandonment of a causal space-time description for atomic
processes. Writing strongly under his influence, Darwin, in the letter of 1919,
suggested the last resort of endowing electrons with free will.43 Von Mises's
rejection of causality in 1921 appears to have been directly related to Bohr's
views as they were expressed in his paper to the Solvay congress that year. 44
And in 1923 H. A. Senftleben, who was also strongly influenced by Bohr,
observed that "Planck's constant h limits in principle the possibility of
describing a process in space and time with arbitrary accuracy."45 Moreover,
he drew the conclusion that in this case the principle of causality, expressed
as "Given a situation A in space-time we can determine a later situation B",
was simply inappropriate: we were not given such a situation A.

THE STAGE IS SET FOR A DIALOGUE

Bohr's views on the causality principle at the time he was joined by Pauli are
unclear and, in the last analysis, unknown. For although we have tried to
indicate a trend in the expression of his ideas the evidence is slight. Although
the causality issue was the most emotionally potent and the most publicised
34 CHAPTER 3

and argued aspect of the quantum theoretical problem complex, it was not
however the most fundamental. And on the most fundamental issue Bohr's
views, and their relationship to Pauli's, are clear. At the very end of his
1922 survey paper Bohr expressed "a hope in the future of a consistent
theory, which at the same time reproduces the characteristic features of the
quantum theory, important for its applicability, and, nevertheless, can be
regarded as a rational generalisation of classical electrodynamics."46 The
quantum concepts were "important for [their] applicability", but the con-
cepts of classical electrodynamics were fundamental. The quantum theory
was based upon these concepts, both through the atomic model itself and
through the correspondence principle, and they were treated as basically
immutable, so that any account of realms in which they were not directly
applicable had to proceed by their limitation rather than their replacement.
In Pauli's view, on the other hand, these concepts had to be replaced by
new ones, to be derived in accordance with the criteria of internal and
operational consistency. Bohr's fundamental concepts of understanding
were those of the wave theory, and he rejected the light-quantum hypothesis
in favour of energy conservation. Pauli's fundamental concepts were to be
derived from discrete counting processes, and he insisted on the validity
of the light-quantum hypothesis,47 and on the retention of energy and
momentum conservation. 48 The concerns of the two phYSicists were similar,
but their convictions opposed.
 CHAPTER 4

THE TECHNICAL PROBLEM COMPLEX

The dialogue between Bohr and Pauli was to be central to the development
of the new quantum mechanics, but it could be so only once it had been
incorporated into the technical problem complex of quantum theory, and
this was no easy matter. Pauli's ideas, developed outside the quantum context,
seemed strange to some of his colleagues and could have little impact until
applied to detailed quantum problems. Even Bohr's views, though developed
in the quantum context, lacked the precision conferred by concrete appli-
cation. Moreover, there was a further complicating factor in the anti-causal
pressures of the Weimar intellectual milieu. That such pressures existed,
were strong, and were recognised and to some extent accommodated to by
German physicists in the early 1920s has been clearly demonstrated by
Forman. 1 And at first sight their existence would seem to offer support for
Bohr's views on the absence of causality in quantum theory. But so far as
the mainstream quantum physicists were concerned, the pressures do not
seem to have been anything like so strong as has sometimes been suggested,
and their existence tends historically to conceal as much as it reveals. 2
Not only for Bohr and Pauli but also for most of the main quantum atomic
physics community, causality was an issue, but only a secondary one, a
decision on which was to be derived from other more fundamental considera-
tions. Within this community there was an awareness that a fully causal
qauntum theory did not yet exist, and that the retention of causality in any
future theory could not be assumed. But a discussion of causality per se did
not seem a very useful way of going about things, and the subject was not
really an issue. 3
More popular among the quantum physicists was Bohr's rejection of
strict energy-momentum conservation. Heisenberg has recalled that after the
implications of Einstein's 1916 paper had been absorbed most members of
the quantum physics communities in Munich and Gottingen were open to
the possibility of statistical energy conservation. 4 And in 1921 Sommerfeld
included in the latest edition of his famous and influential book Atombau
und Spektrallinien, the "bible" of quantum atomic theory, the statement
that 5

35
36 CHAPTER 4

The mildest modification that must be applied to the wave theory is, therefore, that of
disavowing the energy theorem for the single radiation phenomenon and allowing it
to be valid only on the average for many processes.
A little later, Einstein and Ehrenfest found themselves obliged to consider
seriously, though they could not accept, the possibility of energy non-con-
servation. In the Winter of 1921-1922 Stern and Gerl~ch demonstrated that
if a beam of silver atoms was passed along a strong magnetic field gradient
the beam split into two well-defmed beams, deflected from the original
path in opposite directions, and with no atoms at all remaining on the
original path. This result was in clear agreement with the 'space quantisation'
prediction of quantum atomic theory, according to which the atoms could
possess only certain discrete values of magnetic moment, producing the
discrete deflections observed. It was in complete disagreement with the
classical theory, according to which any change in magnetic moment had to
be continuous and should have resulted in spreading of the beam of atoms
with the peak of the resulting distribution remaining along the undeflected
path. What concerned Einstein and Ehrenfest, however, was that while the
behaviour did give the result predicted by quantum theory, they could see
no way in which the process leading to that result could take place, no
way in which in the time available and in a radiation-free vacuum the con-
tinuously varying field could impart sufficient energy to the atoms to make
possible the discrete change in magnetic moment observed. Either one had to
give up altogether any description of the process, and merely rest content
with the result, or else one had to assume at the least that energy was not
conserved. 6 Finally, in a study of 1923 Born and Heisenberg ran into some
closely related problems when considering the behaviour of an atom in
crossed electric and magnetic fields. A continuous change in the field specifi-
cation, which change could be infmitely small, led according to the quantum
theory to a discrete change in the state of the atom, but without apparently
being able to provide the discrete amount of energy required for such a
change. Their provisional conclusion had to be that energy was not strictly
conserved. 7
Bohr's ideas on the failure of mechanics were also reflected quite widely
at about this time. In February 1923 Heisenberg admitted in correspondence
that he was beginning to follow Bohr and Pauli in accepting the failure of
mechanics, 8 and suggested that "either new quantum conditions, or proposals
for the modification of mechanics" were needed. 9 And that Summer Lande
and Born also wrote of this same "failure of mechanics". 10
In general at this stage, Bohr's views dominated over Pauli's. He was
 THE TECHNICAL PROBLEM COMPLEX 37

after all the accepted and undisputed master of quantum theory. But to some
extent, insofar as they talked of the problems rather than the remedies, their
views did overlap; and even where they differed Pauli's more radical attitude
was not without support. Thus Born, who had earlier shown sympathy with
Pauli's views on the inapplicability of field theory to the inside of the atom,
wrote at this time that "the whole system of concepts of physics must be
reconstructed from the ground up." 11 Heisenberg was now becoming in-
creasingly sympathetic to Pauli's viewpoint. And there was a growing feeling
that while statistical energy conservation might indeed be the "mildest
modification" needed to solve the immediate problems, it would not ulti-
mately be enough; that more radical conceptual changes would be necessary.
Only a few weeks after advocating statistical conservation in his textbook, and
after arguing in its favour against Heisenberg, whose study of the anomalous
Zeeman effect had led him to "place ourselves deliberately in opposition to
classical radiation",12 and in support of light-quanta, Sommerfeld admitted
to Einstein that "inwardly I too no longer believe in the spherical waves." 13
Statistical conservation had been introduced as an alternative to the light-
quantum hypothesis, and before the results of Geiger and Bothe in 1925 there
was nothing that could experimentally distinguish between them.14 But as
the light-quantum gradually found wider and more convincing application so
energy non-conservation appeared more and more as an inadequate substitute,
avoiding rather than addressing the crucial problems involved. At the end
of 1922 the discovery of what quickly became known as the Compton
effect provided what is still the clearest and most natural application of
the light-quantum hypothesis. The scattering of X-ray light by an electron
was explained perfectly by the simple laws of particle collisions, and Comp-
ton drew the "obvious conclusion" that the light was composed of discrete
and localised quanta. IS A few months later Debye also came upon the same
effect, and drew the same conclusion. 16 Soon after, the same light-quantum
treatment was extended by Compton and Duane to Fraunhofer diffraction,
and this new run of success for the light-quantum was consolidated by Pauli
who, building on Einstein's 1916 paper and combining this with the insights
of Compton and Debye, derived the first successful probabilistic treatment
of the temperature equilibrium between radiation and free electrons. 17
The Compton effect did not "prove" the existence of light-quanta. Ehren-
fest and Epstein failed in their attempt to extend the light-quantum analysis
to Fresnel diffraction; and Compton himself, between writing and submitting
a paper on Compton scattering, published another important article on the
wave-like total internal reflection of X-rays, a phenomenon which, as he
38 CHAPTER 4

admitted, was "not easy to reconcile" with the conclusions he had drawn
from the scattering effect.18 Moreover it was still the case that the admission
of light-quanta entailed what appeared to be more fundamental inconsistencies
than the absence of strict energy conservation. But the Compton effect and
related results did help to redress the balance between the view of Pauli,
that far more radical conceptual changes were needed than the mere accep-
tance of energy non-conservation, and the previously dominant view of
Bohr, that the way lay through modification rather than replacement of the
classical wave theory. To see how these two viewpoints interacted so as to
lead to the creation of the new theory of quantum mechanics we must now
turn away from general statements and issues to the specific and highly
technical problems of the quantum theory of the atom.

PROBLEMS WITH STATIONARY STATES

Prior to the advent of quantum mechanics the quantum theory of the atom
was divided into two distinct parts, one dealing with the determination
of stationary states and allowed transition frequencies, the other with the
transition probabilities or intensities. The theory of stationary states was
effectively based upon a mechanical model of the atom as a conditionally
periodic system, with the quantum-theoretically possible orbits or vibrations
of the atomic electrons given by Sommerfeld's generalisation of Bohr's
original quantum conditions,19
(1) h =I Pkdqk =nkh
(h action, Pk momentum, qk displacement, h Planck's constant and nk
integers).
The theory was able to predict correctly the spectral frequencies of the
hydrogen atom, both with and without the application of external electric
or magnetic fields. 20 But as soon as it encountered the more complex prob-
lem of mutH-electron atoms, and especially that of the anomalous Zeeman
effect (splitting of spectral lines in a magnetic field) in alkali atoms, which
on Bohr's shell theory of periodic structure had a single electron in the outer
orbit or uppermost set of energy levels, it ran into serious problems. One did
not in fact have to go very far up the periodic table to run into trouble and
in 1921-1922 Langmuir, Epstein and Van Vleck in America, and Kramers
in Denmark, all reported apparently insurmountable difficulties with the
helium atom. 21
Meanwhile in GOttingen, Born embarked on a programme of pushing
 THE TECHNICAL PROBLEM COMPLEX 39

the theory as far as it would go, trying with Pauli's assistance to find out
its limitations and shed light upon the modifications it needed. 22 But they
were able to make little progress, and Born could only conclude, writing
to Einstein, that "the quanta really are a hopeless mess."23 In 1922 Pauli
applied the theory to the hydrogen molecular ion and showed that even for
this structure, which possessed just one electron, it gave the wrong answer.24
Indeed the only success obtained at this period was with a model of the atom
introduced by Heisenberg, while still a student under Sommerfeld at Munich,
and developed by Lande. 2s This "core model" of the atom, however, repre-
sented an obscure but essential departure from the Bohr theory. Abandoning
any detailed decription of the electron orbits such as the theory required,
Heisenberg restricted the description of the atom essentially to that given by
the quantum numbers, and treated the multi-electron alkali atom as composed
of just two parts; the single outer or "series" or "valence" electron, and a
"core" composed of the nucleus and other electrons but treated as effectively
a single particle. 2s And while this model was applied with some success it
was a glaring feature of that success that it entailed hypotheses which, if
interpreted physically, were quite contrary to the established theory. To
obtain the observed splitting of the spectral lines in a magnetic field, half-
integral quantum numbers had to be introduced for some components. The
selection rules determining which transitions between stationary states were
possible violated those established on theoretical grounds by Rubinowicz
and appeared to be quite inconsistent, even arbitrary. And the failure of the
Larmor theorem on the precession of an electron orbit in a field manifested
itself in the fact that to obtain the correct atomic energy in a magnetic
field, the core of the atom (i.e., all but the single outer electron) had to be
counted twice in its contribution to the "g-factor" defining the atom's
angular momentum and magnetic moment. 27
Of these inconsistencies the anomalous g-factor and the half-integral
quantum numbers were seen as the most problematic. Bohr objected of the
former that the break with the classical theory of conditionally periodic
systems that it involved "immediately removes any ground for the calculation
of the energy of the atom in the field of the sort that Heisenberg under-
takes."28 And of the half-integral quantum numbers he wrote simply that
"the entire method of quantisation ... appears not to be reconcilable with
the fundamental principles of quantum theory." 29
But what was the alternative? By the Winter of 1922-1923 Pauli had
moved from Gottingen to Copenhagen and Heisenberg had followed in his
footsteps from Munich to Gottingen; and that Winter was devoted to a
40 CHAPTER 4

thorough investigation of the status of Bohr's atomic theory, in which the use
of perturbation theory was treated as acceptable, but half-integral quantum
numbers and their associated contradictions were not. Born and Heisenberg,
pursuing the perturbation theory approach, published two papers, the first
of which ran into the troubles with crossed fields already mentioned. 3O In
the second paper, in which they also acknowledged the assistance of Pauli,
they subjected the helium atom to its most thorough investigation yet, and
their conclusions were quite clear: 31

We have now set ourselves the problem of examining all possible orbital types in excited
Helium atoms, of selecting the quantum-theoretically possible solutions, and of cal-
culating the energy values so as to be able to establish whether or not orbits are present
which give the empirical terms correctly. The result of our investigation is negative:
one reaches through the consequent application of the known quantum rules no expla-
nation of the Helium spectrum.

This paper was taken as decisive proof that the old quantum atomic theory,
with integral quantum numbers, failed for the helium atom. 32
Meanwhile, late in 1922, Bohr had already anticipated that Born's per-
turbation theory approach must fail, if only because the perturbations were
themselves of the same order as the unperturbed results. 33 And in his major
survey paper on the foundations of quantum theory he had expressed strongly
his conviction that classical mechanics, and with it the existing quantum
atomic theory, must fail. 34 Though differing in their beliefs as to how the
problems of the theory might eventually be resolved, both Bohr and Pauli
were convinced of the need for major innovations, and their approach that
Winter was to push the existing theory as far as it would go, especially in
the context of the anomalous Zeeman effect, and to look for a resolution
through the exploration of the resulting paradoxes. 35 But sticking, as they
felt they must, to the integral quantum numbers, they made little progress. 36
And the situation continued to deteriorate.
Further confirmation of the failure of the theory continued, moreover, to
arise. Towards the end of 1922 Van Vleck had obtained additional evidence
of the failure of integral quantum numbers in a situation in which the appli-
cation of half-integral numbers gave results to within the experimental
error. 37 And this had increased Heisenberg's ( and Sommerfeld's) conviction
that "the half-integral quantum numbers are right." 38 Then in early 1923
Lande made explicit yet another problem arising from the successful use of
the core modeP9 In 1921 Bohr had attempted to construct a quantum
periodic table by means of a "building-up principle".40 For conditionally
 THE TECHNICAL PROBLEM COMPLEX 41

periodic systems, the quantum numbers and statistical weights of stationary

states could be treated as adiabatic invariants, that is as invariants with respect
to certain gradual "adiabatic changes" in the systems. This enabled Bohr to
build up the theory to describe the higher elements by starting from hydrogen
and adding electrons one by one such that the change was in each case
adiabatic. Although the procedure could be applied rigorously only for true
conditionally periodic systems, already recognised as insufficient repre-
sentations of the higher elements, it gave qualitatively good results, and
Bohr was sufficiently convinced of its general validity to stress the role of
the "adiabatic principle" and to formulate a corresponding "principle of
the existence and permanence of quantum numbers" in his 1922 survey
paper.41 No sooner had this been published, however, than Lande demon-
strated that the core model not only necessitated half-integral quantum
numbers but also involved a change in the quantum numbers and statistical
weights during the building-up process. To make matters worse, it was the
building-up process itself that generated the higher elements for which the
core model was used.
Although the required change of quantum numbers in the building-up
process was a new result, the general problem of statistical weights was not.
In terms of the core theory it could be expressed by saying that the theory
gave the core of the atom either one too few or one too many degrees of
freedom, and the outer electron one too many; but it had already existed in
essence before the advent of that model. As early as 1920 Bohr had introduced
a Zwang ("constraint") to regulate the choice of possible orbits, despite the
fact that such a Zwang Gustified in 1922 on the basis of the correspondence
principle) could not be explained in terms of the mechanical model of the
atom.42 In 1922 Lande had referred to the problems of the core model as
involving an "unmechanical adjustment of the core",43 and it was to a
combination of these ideas that Bohr turned with Pauli in 1923. To preserve
what consistency they could, they insisted on integral quantum numbers,
and the invariance of quantum numbers, for all those numbers that were
interpreted as characterising the individual electron motions in the atom. But
for the "inner" quantum number, j, which was associated with the relative
orientations of the electrons and with the coupling between the outer electron
and the core, both conditions were finally sacrificed. And their absence was
explained by an "unmechanischer Zwang": 44

In the electron assemblage in an atom, we have to do with a coupling mechanism that

does not permit a direct application of the quantum theory of mechanical periodic
42 CHAPTER 4

systems; in particular, there can obviously be no question of accounting for the complex
structure in terms of the exclusion, based on the consideration of adiabatic transforma-
tion, of certain motions, compatible with this theory, as stationary states of the atom.
Rather, we are led to the view that the interplay between series [outer] electron and
atomic core, at least so far as the relative orientation of the orbit of the series electron
and those of the core electrons is concerned, conceals a "Zwang" that cannot be described
by our mechanical concepts and that has the effect that the stationary states of the
atom, in essential respects, cannot be compared with those of a mechanical periodic
system. According to our view it is just this Zwang that fmds its expression in the
regularity of the anomalous Zeeman effect, and, in particular, is responsible for the
failure of the Latmor theorem.

The introduction of the unmechanischer Zwang did not solve the problems of
atomic theory, and the paper was not indeed published; but it did mark an
important step in their development. At this stage neither Bohr nor Pauli
could accept Heisenberg's rather buccaneering approach to the quantum
theory. As Pauli wrote to Bohr at the beginning of 1924,45
If I think about his ideas they seem monstrous and I curse to myself a lot about them.
Because he is so unphilosophical, he pays no attention to clear presentation of the basic
assumptions and their relationship to previous theories.

Bohr's feelings appear to have been rather stronger,46 and the fundamental
differences between Bohr and Pauli also remained. But the new development
did represent a general recognition that while Heisenberg's methods might be
appalling, and many of his suggestions unacceptable, his was the approach
that was getting results. Moreover, it also provided a common language in
terms of which the problems could be further explored.
When Pauli wrote up his work on the anomalous Zeeman effect in April
1923 he stuck hard by his criterion of internal consistency and adopted what
he described as a "purely phenomenological" description which "abandoned
all use of models".47 In June, he wrote to Sommerfeld that the quantum
theory supplied "no sufficient grounding" for the treatment of complex
spectra, and that something "in principle new" was needed for the anomalous
Zeeman effect.48 The old quantum theory had failed: 49

This failure can scarcely be doubted any longer, and it seems to me to be one of the
most important results of the last few years that the difficulties with many-body
problems lie in the physical atom, not in the mathematical treatment (when, e.g. the
Helium term comes out wrong in Born and Heisenberg, this certainly does not lie in the
fact that the approximation is insufficient.

But beyond the adoption of his phenomenological approach and a consequent

 THE TECHNICAL PROBLEM COMPLEX 43

rejection of the orbital model of the atom, Pauli had no idea what to do about
this failure. In July he wrote to Bohr optimistically that "perhaps after all
you may in the course of the Summer get a saving idea about complex
structure and the anomalous Zeeman effect."50 But it was not to be, and
rather than pursue further the details of what appeared to be an innately
inconsistent theory Pauli turned his attention elsewhere. In August he sub-
mitted his paper on the thermal equilibrium between radiation and free
electrons. 51 In September he wrote the letter to Eddington based on the
latter's mathematical theory of relativity. 52 Later in the year he returned
to Copenhagen for a two months "holiday", and early in 1924 he explained
to Bohr how he felt about the quantum atomic theory: 53
The atomic physicists in Germany today fall into two groups. The one calculate a given
problem fIrst with half-integral values of the quantum numbers, and if it doesn't agree
with experiment they then do it with integral quantum numbers. The others calculate
fIrst with whole numbers and if it doesn't agree then they calculate with halves. But
both groups of atomic physicists have the property in common, that their theories
offer no a priori reasoning which quantum numbers and which atoms should be calcu-
lated with half-integral values of the quantum numbers and which should be calculated
with integral values. Instead they decide this merely a posteriori by comparison with
experiment. I myself have no taste for this sort of theoretical physics, and retire from
it to my heat conduction of solid bodies.

HEISENBERG, BORN AND DISCRETE ATOMIC PHYSICS

As it turned out, Pauli's absence from atomic physics lasted only a few months.
But in that time the theory saw two further striking developments, both of
which were to play a large part in shaping its future development. One of
these, to which we shall return, was in the branch of the theory devoted to
transition probabilities or intensities, and was to lead to the infamous virtual
oscillator theory of Bohr, Kramers and Slater. The other was a continuation
of Heisenberg's work on the core model, and although it is far less well
known, it was to be equally important for the development of a new quantum
mechanics.
Early in 1923 Heisenberg's reaction to the failure of the existing atomic
theory had been to continue developing this theory but to suggest that
"either new quantum conditions or proposals for the modification of
mechanics" were needed.54 And returning to the anomalous Zeeman effect
in October of that year he did in fact derive some new quantum conditons: 55

= f-112
+ 112
(2) H quantum HcIa&<Acal dj =F(j +~) - F(j - ~),
44 CHAPTER 4

with H the Hamiltonian and F a new function to be determined. The immediate

context of this formula was the problem of the anomalous g-factor (governing
the magnetic moment) in Heisenberg's core model of the atom, and Serwer
has suggested that the problem of statistical weights may also have played an
important part in its origin. 56 In some of its manifestations this required that
the number of atomic states should be one less than that given by the quantum
numbers, and by associating each single state with a pair of quantum numbers
instead of a single quantum number, Heisenberg's new formalism overcame
this difficulty. From the letter to Pauli in which the formalism was first
developed, however, its origins appear to have been more fundamental. 57
Heisenberg wrote that so far one had obtained the frequency of a transition
simply by taking the difference between two energies as in the Bohr quantisa-
tion conditions,
(3) hv= aH,
but that this was only appropriate for the simple case of hydrogen. In the
general case, one would also have to derive the energies themselves from a
difference equation, say H = a F: 58

The new Gottingen theory of the anomalous Zeeman effect runs roughly as follows:
(1) The model representations have in principle only a formal sense, they are the classical
analogues of the 'discrete' quantum theory,
(2) Up to now it was usual to go over from model symbols to the real radiation
frequencies by taking over the energy H(J .. ... , I n ) from the symbols to the Vqu
through the difference equation ....
(3) This is only a special case, which is right for Hydrogen. In other problems one must
take from the symbols other functions than H. A general theory, as to which functions
of the J .. ... ,In is still outstanding.
(4) For the anomalous Zeeman effect, the function in question reads F(k, r, j, m) =
f H (k, r, j, m) dj = f H dj. From F one gets to the Hqu through AF = Hqu, from H to
v through AH = v.

Heisenberg gave no real justification for this approach, but he stressed through-
out the letter the fundamental role of the difference equations. He always
referred either to "our" approach or to "the G6ttingen" approach, rather
than claiming it as his own, and he noted Born's summing up of their future
programme as the "discretisation of atomic physics."s9 Combined with
Heisenberg's later recollection of a seminar in Hilbert's department at this
time on the very subject of difference equations/,c) this all goes to suggest
that it was the concept of difference equations that lay behind the new
formalism. If the seminar did playa crucial role, then Heisenberg and Born
 THE TECHNICAL PROBLEM COMPLEX 45

would indeed have shared in the idea to develop this concept. And it is
possible, moreover, to reconstruct the argument in the letter to Pauli from
the difference equation basis. Whereas Bohr's original quantum condition
was indeed a difference equation, the more general Bohr-Sommerfeld con-
ditions, Equation (1), were not. Having decided to express everything in
difference equation form, Heisenberg would therefore have had to search
for a new quantisation condition, and the requirement that it gave the correct
g-factor was in fact sufficient to determine his choice. Earlier in 1923 Lande
had given a general formula for the g-factor, and this gave for the required
H in terms of the quantum numbers,61

This could be obtained from the formula proposed by Heisenberg in the

letter to Pauli with

i.e., as the difference between two terms each expressible in terms of a single
clearly identifiable set of quantum numbers.
Although it caused quite a stir, the Heisenberg-Born formalism was not
at first universally well received. From Bohr's point of view it merely com-
pounded the felony of the core model and half-integer quantum numbers,
and went in completely the wrong direction, even further away from any link
with the classical conceptions. 62 Heisenberg's happy acceptance of, even
advocacy of, the fact that his approach abandoned any attempt to make
physical sense out of the basic formulae was not for Bohr. 63 Even Pauli,
who unlike Bohr was favourably predisposed to the idea of a new discrete
form of mechanics, could only despair at the lack of any conceptual foun-
dation for the new ideas, writing to Lande that 64
I don't in any way share your opinion of Heisenberg's new theory. I even hold it for
ugly. For despite radical assumptions it provides no clarification of the half quantum
numbers or the failure of the Larmor theorem (especially the magnetic anomaly). I
don't think much of the whole thing.

Faced with such opposition, Heisenberg withheld publication of his new for-
malism, and concentrated first on establishing the need for the half-integral
quantum numbers, and on trying, unsuccessfully, to derive these from the
46 CHAPTER 4

formalism instead of having to put them in to it. 65 And by the time it was
published, in the Summer of 1924, it was in many respects out of date.
But by then it had already served its purpose, for it had provided Born and
Heisenberg with a vital link between the problems of stationary states and the
idea of discreteness on the one hand, and the theory of transition intensities
and Bohr's emphasis upon continuity on the other.

THE BOHR-KRAMERS DISPERSION THEORY

Whereas the part of quantum atomic theory dealing with stationary states
and frequencies was already well established by the end of the Great War,
the part dealing with transition probabilities, or intensities, could not get
under way until after the enunciation of Bohr's correspondence principle
in 1919. Even then it developed only slowly as the problems with the basic
model, some of which we have outlined above, dominated the researches of
the physicists. One aspect of the intensities problem did however receive
some attention, and that was the theory of dispersion.
In principle, any theory of dispersion had to be related in the Bohr theory
to the orbital model of the atom and to the theory of transition frequencies
derived therefrom. The Bohr atomic model did not itself incorporate any
mechanism through which the transition probabilities could be predicted,
and they had therefore to be derived from the classical intensities by way
of the correspondence principle. But they could only be derived in this way
as functions of frequencies. However, although the correspondence principle
rested upon an application of the orbital theory in a limit in which the
quantum transition frequencies were comparable with the classical absorp-
tion and emission frequencies, the rigorous application of the orbital theory
had failed and the only cases of physical importance lay outside the limit of
comparison. The only feasible approach to the quantum theory of dispersion
was therefore to take the observed frequencies and intensities, to express
the latter in terms of quantum transition probabilities, and to try thereafter
to reconcile both with the quantum theory of the atom.
In 1921 Ladenburg took the first step in this direction by showing that,
given the observed absorption and emission frequencies, both the observed
absorption intensities and the observed dispersion coefficients corresponded
to those predicted by the classical theory.66 He could only draw very limited
connections with the quantum theory of the atom, relating transition prob-
abilities very roughly to statistical weights of atomic stationary states for a
few simple cases. But his work did provide strong evidence that the classical
 THE TECHNICAL PROBLEM COMPLEX 47

theory of dispersion could be carried over to the quantum context, dispersion

being of course a fundamentally wave-type of phenomenon. Working with
Reiche in 1923, Ladenburg was able to extend this evidence and to draw
the explicit conclusion that: 67
We believe on the grounds of the observed phenomena that we must consider the end
result of a process in which a wave of frequency v is incident upon the atom as not
fundamentally different from the effect that such a wave exerts on classical oscillators:
. .. Even the force of scattered waves seems repeatedly to agree with that from an
oscillator.

The dispersion theory of Ladenburg and Reiche was restricted to disper-

sion by atoms initially in their ground state, that is to dispersion as a pure
absorption process, and apart from a vague reference to the correspondence
principle it was without any foundations in the quantum theory of the atom.
There was, however, no viable alternative to its basic approach, and in an
unpublished manuscript of 1921 Bohr wrote encouragingly of Ladenburg's
work: 68

Although it is at present an unsolved problem, how a detailed theory of dispersion can

be developed on the basis of the quantum theory, a promising beginning on the indicated
basis might nevertheless seem to be contained in the interesting considerations about
this phenomenon, recently published by Ladenburg.

Writing to Darwin in December 1922, Bohr suggested that dispersion should

be attributed to some "mechanism" called into play when the atom was
illuminated, "with the effect that the reaction of the atom corresponds to
that of a harmonic oscillator in the classical theory, with the frequency
coinciding with that of a spectral line."69 In a major survey of quantum
theory, completed that November, he had expressed himself more carefully
and at greater length: 70

On the one hand, as is well known, the phenomena of dispersion in gases show that the
process of dispersion can be described on the basis of a comparison with a system of
harmonic oscillators, according to the classical electron theory .... On the other hand,
the frequencies of the absorption lines, according to the postulates of the quantum
theory, are not connected in any simple way with the motions of the electrons in the
normal state of the atom ... According to the form of the quantum theory presented
in this work, the phenomena of dispersion must then be so conceived that the reaction
of the atom on being subjected to radiation is closely connected with the unknown
mechanism which is answerable for the emission of radiation on the transition between
stationary states. In order to take account of these observations, it must be assumed
that this mechanism, which is designated in the preceding paragraph the coupling
mechanism, becomes active when the atom is illuminated in such a way that the total
48 CHAPTER 4

reaction of a number of atoms is the same as that of a number of harmonic oscillators in

the classical theory, the frequencies of which are equal to those of the radiation emitted
by the atom in the possible processes of transition, and the relative number of which is
determined by the probability of occurrence of such processes of transition under the
influence of illumination.
Extracted from the familiar Bohr dressing, this long exposition amounts to
the statement that one should use the classical theory for the calculation of
transition probabilities, whether in dispersion phenomena or elsewhere, and
that this is quite justifiable as the cause of the transitions is unknown anyway.
Considering X-ray absorption phenomena in a paper completed in the
Autumn of 1923, Bohr's assistant Kramers noted that "the quantum theory
in its present state tells us nothing about the mechanisms of absorption and
does not therefore permit the direct calculation of the probability that an
absorption process may occur ."71 In this context and in that of the X-ray
emissions from electron-atom colliSions, Kramers therefore adopted the
procedure laid down by Bohr, and worked with the classical theory. In the
latter context he explained that
the only procedure which offers itself at present seems to consist in estimating the
statistical result of a great number of emission processes - in a way suggested by Bohr's
correspondence principle - from the radiation which on the classical theory would be
emitted by the free electrons in consequence of the change in motion produced by the
forces owing to the electrical particles in the atom. 72

The question remained as to how this classical treatment was to be tied in

with the orbital quantum model of the atom. In November 1922 Bohr had
been still prepared to hope that the orbital model might survive its current
problems, and this hope was reflected in his reference to "a number of atoms"
for comparison with the classical theory, it being impossible to incorporate all
the possible transition frequencies within a single orbital atom. For atoms
initially in a given stationary state, as in Ladenburg's theory of dispersion, it
would have been desirable if all possible transition frequencies from that state
could somehow have been contained in the appropriate electron orbit, and
Kramers suggested in his X-ray paper that "one should expect that every
possible transition corresponds to a certain frequency present in the motion
of the electron."73 But it was difficult to conceive how even this requirement
might be satisfied. Kramers, who had not been directly involved with the
problems of the core model and unmechanischer Zwang, appears to have held
fast to a relatively literal interpretation of the Bohr model. 74 But to Bohr
the pressures on this model were building up, and the need to incorporate
the classical theory of dispersion on top of all the other problems prompted
 THE TECHNICAL PROBLEM COMPLEX 49

him to give further thought to the problem of conceptual foundations. Just as

Heisenberg held back his new difference equation formalism from publication
on account of criticisms from Bohr, so Bohr himself had refrained from
publishing his thoughts on the unmechanischer Zwang while the implications
of Heisenberg's ideas were worked out. And in the Winter of 1923-1924 the
work of the previous Spring was finally superseded by a new manuscript draft
which sought to accommodate both the problems of the core model and the
results of dispersion theory, and which fell back on Bohr's ideas on energy
conservation and causality. 75
Both the general classical nature of the dispersion theory and the fact that
it could not be reconciled with the orbital model of the atom other than on a
statistical basis, for a large number of atoms, supported Bohr's earlier conclu-
sions on statistical energy conservation. And in the manuscript of 1923-1924
he carried his argument slightly further than he had before, noting the need
to "depart essentially from the causal description in space and time that is
characteristic of the classical mechanical description of nature." 76

PAULI, HEISENBERG, AND THE REJECTION OF ELECTRON ORBITS

While Heisenberg and Born pushed their discretisation of atomic physics and
Bohr and Kramers pushed their classical theory of dispersion, with its
emphasis upon the continuous wave formulation and associated changes in
the laws of energy conservation and causality, Pauli concentrated on other
things. But he did not refrain from all comment, and nor did he restrict his
comments to criticisms. Thus in June 1923 he wrote to Sommerfeld with the
idea that rather than adapting the dispersion theory to fit in with the hypo-
thetical orbital model of the atom, it was the latter that should be adjusted to
fit in with the former: 77
I often think that not only in dispersion, where they are under the influence of a simply
harmonic periodic external force, but also in the mutual effects of the electrons in the
atom, the individual electron orbits control themselves more as a system of oscillators in
which the frequencies are associated not with the motion but with the transition.
Writing to Bohr in February 1924 he decided to press his view that the
concept of electron orbits, upon which the whole atomic theory still depended,
had to gO:78
The most important question appears to me to be this one, to what extent one may in
general speak of flXed orbits of electrons in stationary states. I think that this can no
way be assumed as self-evident, especially in view of your observations about the balance
of statistical weights in coupling. Heisenberg has in my view hit the mark precisely when
so CHAPTER 4

he doubts the possibility of fIxed orbits. Doubts of this kind Kramers has never con-
sidered as reasonable. I must nevertheless insist upon this, because the point appears to
me to be very important.
Pauli's views on the inadmissibility of the concept of electron orbits within
the atom were by this time well established, and they may be traced back to
his criticisms of Weyl's attempt at a unified field theory. It was impossible
to specify any operational means of defining the orbit, for once bound in the
atom the electron could be 'observed' only by means of the transition inten-
sities and frequencies. The suggestion of the letter to Sommerfeld, that the
theory of the atom should be therefore based upon these observables rather
than upon the hypothetical orbits, followed. But what of Heisenberg's
opinions? From Pauli's strong and unbending criticisms of Heisenberg's work
it would seem at first that the two physicists were in fundamental opposition,
but this was not in fact so. Heisenberg recalled that the rejection of electron
orbits in the atom had been a shared feature of their student days together in
Munich, and although there is no direct evidence of Heisenberg discussing this
matter before 1925, Pauli's letter to Bohr just cited confirms their agreement
on it at an earlier date. 79 Heisenberg continued to work with his core model
version of the orbital model and to use this, to Pauli's perpetual despair,
without any regard for its foundations or physical consistency. But in corres-
pondence with Pauli he recognised that "the model conceptions have
principally only a symbolic sense."80 And far from abiding by Bohr's in-
sistence that the correspondence principle must be founded upon the orbital
model, Heisenberg had claimed that "the correspondence principle renounces
any model insight."81 In 1921 he had actually compared the correspondence
principle and the orbital model as rivals, and had concluded that the former
was the more sound, since founded directly upon experiment. 82 In one sense,
of course, Heisenberg's attitude to the correspondence principle in these
examples represented the very "unphilosophical" approach that Pauli so
lamented. But if Heisenberg did not provide sound foundations for his
arguments he did not, at least, cling to unsound ones. And this ensured that
while on one level he was subject to Pauli's criticisms on another, deeper,
level he was open to them. 83 From the other side Pauli, though always
critical, was far from dismissive: 84
But if I talk to him, he strikes me as all right, and I see that he has all sorts of new
arguments - at least in his heart. I therefore think of him ~ aside from the fact that
he is also personally a very nice fellow - as very thoughtful, even a genius, and I think
he will once again greatly advance science.
He did, but only after Bohr and Pauli had finally met head on, in the context
of the virtual oscillator theory of radiation.
 CHAPTER 5

FROM BOHR'S VIRTUAL OSCILLATORS TO THE

NEW KINEMATICS OF HEISENBERG AND PAULI

INTRODUCTION

Between them, the various lines of thought current at the end of 1923
contained most of the ingredients that would be needed for the formation
of a new quantum mechanics. But these ingredients had not yet been brought
together. The two parts of the quantum theory of the atom were still largely
independent. And the different approaches of Bohr, Pauli and Heisenberg still
ran alongside each other rather than engaging in any fruitful union or dynamic
conflict. For further developments to take place a new element was needed,
and this was provided early in 1924 in the form of the Bohr-Kramers-Slater
theory of virtual oscillators.l Paradoxically, although the theory has been
widely treated as the central and most fundamental feature of the evolution
of quantum mechanics, its basic principles were almost universally rejected.
But this rejection was itself important, setting as it did the stage for the
development of quantum mechanics upon the rival ideas of Pauli. Equally
important, the theory also drew attention to ways in which the techniques
of the quantum theory of dispersion could be applied to those parts of the
theory previously in the domain of the orbital model. 2

THE GENESIS OF THE VIRTUAL OSCILLATOR THEORY

In essence, the virtual oscillator theory entailed only a very slight, but very
crucial, development of Bohr's position. Departing from the somewhat
hesitant and reserved attitude to the problem of an intuitive picture that had
characterised his previous publications, Bohr came out openly in support of
the view that such a picture was essential, and that it would be possible if and
only if causality and energy-momentum conservation were abandoned, and
the classical oscillator representation of the atom, preViously no more than
a heuristic device tenuously justified on the basis of the correspondence
principle, reinterpreted as a physically meaningful model. In describing this
model in the Bohr-Kramers-Slater (BKS) paper, he was typically vague. He
described the oscillators as having a "virtual" existence - a characterisation
that was at best ambiguous and at worst quite meaningless - and he did not
51
52 CHAPTER 5

commit himself as to the relationship between the new oscillator model and
the old orbital one, the validity of which he appeared to uphold. Even the
rejection of causality, for which the paper is perhaps most famous, was left
open to interpretation as a temporary measure. 3 But there can be little doubt,
especially in view of his later reaction to its refutation, that Bohr took the
new model interpretation very seriously, and that he saw his decision to
concentrate on an intuitive description in terms of the classical wave concep-
tions as both fundamental and necessary. 4
This decision was of course a natural one for Bohr to make, and the virtual
oscillator paper continued the line of argument already manifest in his
manuscript of 1923 - 1924, without introducing anything dramatically new.
The new theory may indeed have been formulated independent of any further
outside influences. But its public expression at least was stimulated by the
arrival in Copenhagen of a young American physicist, Slater, and by his
advocacy of a theory similar to, and possibly derived from, that which had
been recently proposed by Louis de Broglie.
De Broglie had started work on the problems of quantum theory in 1921
and had since completed a systematic study of those phenomena that revealed
most clearly the fundamental wave-particle duality of light. Influenced by his
brother Maurice's work on the particulate properties of X-rays, and by his
tremendous admiration of Einstein, he had started out convinced of the
necessity of the light-quantum concept, but intrigued by the problem of how
the frequency of such a particle could be defined. He first looked at the
quantum phenomenon described by Stokes's law in which light, affected by
matter, always passes from a higher to a lower frequency, and from this he
drew an analogy between frequency and temperature, and between Stokes's
law and the second law of thermodynamics. s But the concept of the tempera-
ture of a single material particle raised just as many problems as did that of
the frequency of a light-quantum, and although this confirmed the strength
of the analogy it did n9t actually get him anywhere. De Broglie therefore
turned to the problem of Planck's law, of which there was still no satisfactory
interpretation, and having derived Wien's law from the hypothesis of in-
dependent light-quanta he investigated the dependence necessary to modify
this to Planck's law. 6 The research was still unproductive so far as any funda-
mental insight was concerned, but it did lead to the introduction of a rather
unusual conception. For de Broglie treated his derivations of the radiation
laws as exercises in straight-forward relativistic particle mechanics, and the
light-quanta as traditional material particles of negligible (but not necessarily
zero) rest mass and with velocities approaching (but not necessarily equal to)
 FROM OSCILLATORS TO A NEW KINEMATICS 53

c. The aim of this approach appears to have been to sharpen the wave-par-
ticle paradox by treating the light-quanta as thoroughly traditional particles. 7
But for whatever reason they may have been introduced, the light-quanta
that did not move at the speed of light remained a permanent feature of de
Broglie's endeavours. And in the course of 1923 these resulted in his famous
wave theory of matter.
Although de Broglie's theory was of crucial importance· for the detailed
development of wave mechanics it actually contributed little to the generally
accepted conceptual foundations of quantum mechanics, and had no direct
influence on the development of matrix mechanics. 8 To consider it in detail
in this context would therefore be out of place. But before writing his full
These, de Broglie published a series of short papers describing his main
results, and an English summary of these was sent to Fowler in Cambridge
for publication early in 1924 in the Philosophical Magazine. 9 And in Decem-
ber 1923 Slater wrote to Kramers from Cambridge, where he was under the
supervision of Fowler, proposing a treatment of light-quanta along the lines
used by de Broglie. 1o
It would seem that before leaving Harvard in the Autumn of 1923, Slater
had been committed to the classical wave theory of light. He recalled that
he had been unable to accept the abandonment in the Bohr theory of the
classical relationship between the width of the spectral lines (interpreted by
Bohr as being due to a statistical spread of the energies for each stationary
state over different atoms) and the length of a finite wave train (or the period
of its emission).l1 He found the concept of instantaneous transitions between
stationary states "quite silly", and was convinced that a finite emission period
was neededP In December, however, he wrote to Kramers that he had come
to the "rather surprising conclusion that the only possible way of getting a
consistent explanation was in the direction of light-quanta.,,13 And he
expounded an idea as to how this might be done: 14

Of course, the quanta can't travel in a straight line with the speed of light: but it seems
possible to suppose that there is an electromagnetic field, produced not by the actual
motion of the electrons, but with motions with the frequency of possible emission lines
(or, in an impressed field, of possible absorption lines), and amplitudes determined by
the correspondence principle, the function of this field being to determine the motion of
the quanta. If this motion is determined by the condition that Poynting's theorem shall
hold over an average taken over a long period of time, defmite patterns are described,
and the probability of moving along the paths is such, for example, as to account for
interference, many quanta being led to the bright spots in the field.

Slater's idea involved a remarkable fusion of diverse elements. His new

54 CHAPTER 5

found preference for the light-quanta could reflect a number of influences,

including that of American discussion of the Compton effect. 15 The concept
of light-quanta governed by a guiding wave field had featured repeatedly in
Einstein's speculations, recently expounded in California by Lorentz,16 as
well as in de Broglie's theory, in which the quanta moved, as for Slater, at less
than the speed of light. Which if any of these was Slater's source remains a
matter for speculation.17 But he had proceeded to fuse the guided light-
quantum concept with the oscillator theory of dispersion advocated by that
concept's strongest opponent, Bohr, at which he had been looking prior to his
visit to Copenhagen. IS Throwing in Poynting's theorem, which he took from
some of Cunningham's lectures in Cambridge,19 he derived a theory in which
his original problem of spectral widths was, somewhat spectacularly, solved. 2o
The first recipient of Slater's idea was presumably Fowler, his host, who
seems to have been quite pleased with it. 21 Back in America, Kemble was also
impressed. 22 But in Copenhagen the reception was mixed. Kramers either
never mastered the full subtlety of Bohr's approach or never quite accepted
this approach, but as his student and assistant he was strongly influenced by
it, and in a popular introduction to quantum theory completed in 1922 he
had followed Bohr's rejection of the light-quantum concept, which he likened
to "medicine which will cause the disease to vanish and kill the patient".23
On receiving Slater's idea he apparently reacted warmly to the extension of
the oscillator approach to include some kind of wave field, but rejected
outright the introduction of light-quanta. 24 Bohr then incorporated the idea
into his own developing conceptualisation, accepting the guiding field as a
physical realisation of the oscillator technique, and again rejecting the light-
quantum component. When Slater arrived in Copenhagen, and showed them a
paper describing his idea, Bohr and Kramers first edited this paper so as to
present Slater's idea as leading on to Bohr's conception: 25

An atom may, in fact, be supposed to communicate with other atoms all the time it is in
a stationary state, by means of a virtual field of radiation, originating from the oscillators
having the frequencies of possible quantum transitions, and the function of which was to
provide for stationary states conservation of energy and momentum by determining the
probabilities of quantum transitions.

Slater recalled that this published version was a third draft, written under the
strong influence of Bohr and Kramers. Moreover, writing to van Vleck that
July he claimed, albeit somewhat emotionally, that this draft was actually
written by them. 26 They then wrote up Bohr's conception more fully, adding
Slater's name gratuitously to the paper:27
 FROM OSCILLATORS TO A NEW KINEMATICS 55

We will assume that a given atom in a certain stationary state will communicate con-
tinually with other atoms through a time-spatial mechanism which is virtually equivalent
with the field of radiation which on the classical theory would originate from the virtual
harmonic oscillators corresponding with the various possible transitions to other
stationary states. Further, we will assume that the occurrence of transition processes .. .
is connected with the mechanism by probability laws which are analogous to those ... in
Einstein's theory .... The occurrence of certain transitions in a given atom will depend
on the initial stationary state of the atom itself and on the states of the atoms with
which it is in communication through the virtual radiation field, but not on the occurr-
ence of transition processes in the latter atoms ....
We abandon ... any attempt at a causal connexion between the transitions in distant
atoms, and especially a direct application of the principles of conservation of energy
and momentum, so characteristic for the classical theories.

In citing Slater as a joint author of this paper, Bohr and Kramers probably
meant nothing but kindness and respect. But Slater, who seems to have been
quite opposed to the new theory, was naturally a little disturbed, and he left
Copenhagen prematurely. 28

THE REJECTION OF BOHR'S INTERPRETATION

Slater was not the only physicist to reject Bohr's interpretation. Sommerfeld
and Compton, for example, insisted that the Compton effect provided
definite evidence of the necessity of the light-quantum concept and of the
energy-momentum conservation associated with it. 29 Bohr did attempt an
explanation of the effect in the BKS paper, but this involved a velocity of the
virtual oscillators different from that of the particles to which they were
supposed to be attached, and was not remotely convincing. 30 Sommerfeld
also referred contemptuously to the BKS "compromise",31 while Stoner,
hitting on the limited achievements of the theory in comparison with its
assumptions, argued that "it seems unnatural to assume that [conservation1
does not hold in individual processes when there is no definite evidence of its
breakdown, unless the supposition leads to a much more complete and
satisfying explanation of observed phenomena than has hitherto been put
forward."32 Einstein objected violently to the absence of conservation and
causality, arguing among other things that "a box with reflecting walls
containing radiation, in empty space that is free from radiation, would have
to carry out an ever increasing Brownian motion,"33 and Ehrenfest wrote to
him that "this time, as an exception, I frrmly believe you are right."34 As
reported by Pauli, Einstein also objected that there were now two explana-
tions of spectral widths, from the decay time and from the state uncertainty,
56 CHAPTER 5

and that the theory thus needed a "pre-established harmony" that he did not
like. 35 Unlike most of Bohr's critics, Pauli himself must have been at least
partially aware of the thinking behind the interpretation. But from his own
philosophical viewpoint this interpretation, restricted as it was to a classically
intuitive description, evaded the fundamental issue of a new conceptual frame-
work. His first reaction, in response to a preprint of the paper, was mocking: 36

I have tried on the basis of the definition of the two words [kommunisieren, virtuellJ to
guess what your work is really about. But it is not easy. In any case, it is very interesting
to me and if I can be of any help with the grammar I shall gladly oblige.

In October he wrote to Bohr that he could not reject the theory on scientific
grounds, that Einstein's objections did not worry him, and that his own were
not strictly logical. But he had to admit that he was completely opposed to
the theory and that he shared this opposition with "many other physicists,
perhaps even the majority" - in fact a considerable understatement. 37
Bohr's interpretation did receive some support, from Schrodinger, who
had also recently abandoned causality and conservation for the sake of a pure
field theory ,38 and possibly from Kramers. But even Kramers's support is
uncertain. In response to Breit's criticism that the emission component of his
formula corresponded to classical oscillators in which e2 1m was negative (a
feature later incorporated in the thflory of holes), he described the virtual
oscillators as "meant only as a terminology"; on the other hand, working
with Heisenberg later, he ignored their virtual nature altogether and treated
the oscillator model as naively as he had the orbital model, and in none of his
own work did he mention the causality issue. 39 Of those who expounded or
developed the new technique,Bom's assistant Jordan was the most sympathe-
tic to a pure continuum treatment, but he was also a strict positivist so far as
physical interpretations were concemed. 40 Neither Fowler nor Becker, who
both discussed the technique, made any reference to Bohr's interpretation. 41
Born adopted the technique "independent of the critically important and still
disputed conceptual framework" ,42 and Heisenberg became interested in it
only after Born's work had tied it in with the discretisaton programme they
had worked on together the previous Autumn;43 his first reaction had been
that "Bohr's work on radiation is indeed very interesting, but I do not really
see it as an essential progress."44 Ladenburg wrote to Kramers that he and
Reiche were glad his work coincided so well with their own considerations,45
but this was a response to a further development of the dispersion theory by
Kramers, which was effectively independent of the interpretation, and not to
the BKS theory itself.46
 FROM OSCILLATORS TO A NEW KINEMATICS 57

In summary, it seems clear that Bohr's interpretation did not have the
most enthusiastic of receptions. And before it was even conceived, Ramsauer
had published the results of his experiments on the penetration of atoms
by slow electrons, results that posed similar problems to those encountered
in radiation phenomena, but in a context where the rejection of conserva-
tion was impossible. 47 These results disturbed Bohr more and more. By the
time Geiger and Bothe announced in April 1925 that coincidence counting
of X-rays and recoil electrons confirmed Compton's light-quantum expla-
nation of X-ray scattering,48 implying that conservation of energy and
momentum had to be upheld in radiation phenomena too, Bohr was already
anticipating this refutation of his own interpretation. 49 It still came hard,
especially when accompanied by yet another exposition of the advantages of
de Broglie's theory, this time by Born. so But with Pauli's help he managed
to recondition himself, and admit that his "revolution" was overY As
Pauli wrote to Kramers in July, in the wake of Heisenberg's new kine-
matics,s2

[The ideas of BKS] thus move in completely the wrong direction: it is not the energy
concept that is to be modified but the concepts of motion and force. One can indeed
derme no fixed paths for the light-quanta where interference phenomena are present, but
nor can one define any such paths for the electrons in atoms; and to doubt the existence
of light-quanta on the grounds of interference phenomena is just as little justified,
therefore, as to doubt the existence of the electron would be.

HEISENBERG, BORN AND THE DEVELOPMENT OF THE

OSCILLATOR TECHNIQUE

Fortunately for the development of the oscillator technique, Bohr's closer

colleagues knew of his prejudices and were not, apart from Pauli, put off
by his interpretation. Toward the end of 1923, before the intervention of
Slater's idea, Kramers had generalised the old quantum theory of dispersion
to incorporate dispersion by emission from atoms not originally in the
ground state,S3 and following the discussions with Slater and Bohr he dressed
up the new formula in virtual oscillator language and presented it as the first
derivation from the new theory .54 Ladenburg's formula based on pure
absorption had taken a form equivalent to

(1) 9)1 (t) ex: J L fa

1 a (v~ - v2 )
I
cos 21TVt
58 CHAPTER 5

for the scattering moment with frequency v, and Kramers simply generalised
this to

(2) IDl(t) ex {~fa

a (va - v2 )
_ ~e (v; fe_ v2 } cos 21TVt,
)

where the coefficients fa and fe, and the transition frequencies va and Ve,
corresponded to absorption and emission processes respectively.
Kramers did not at first give any derivation of his new formula, which
was almost certainly guessed from its classical equivalent,55 but in a second
paper, in July, he did sketch a derivation, apparently taken from the first
important development of the oscillator technique, by Born.56 Born had not
shown any previous interest in dispersion phenomena, but he wa~ struck by
the similarity between Kramers's form1,lla and the difference equations
occurring in the discretisation programme he had been developing with
Heisenberg. 57
This difference equation formalism was, as we have seen, completely
symbolic, all questions of physical interpretation having been explicitly set
aside. 58 It was also completely general, with implications for radiation
phenomena such as dispersion as well as for the structure of the atom. 59 And
it was thus a matter of course that Born and Heisenberg should reinterpret
Kramers's dispersion formula (2) as a difference equation,

(3) IDl (t) ex ~{ fa _ fe },

(va - v2 ) (vi _ v2 )

and seek to incorporate it in their general programme. That this is indeed

what happened is confirmed by Heisenberg's letter to Pauli reporting his first
serious interest in the oscillator approach, where he wrote that the difference
equations were the key to the whole thing,60 and by both Born's and
Heisenberg's recollections. 61 The problem of how and where to apply the
difference equation approach so as to derive Kramers's formula became a
recurring subject of seminar discussions, through which Born's new theory
gradually emerged. On 13 June two papers, one by Born on the new theory
and one by Heisenberg on the (previously unpublished) difference equation
quantum conditions, were submitted for publication simultaneously.62
Born's paper acknowledged the help not only of Heisenberg but also of
Bohr, who visited Gottingen during its preparation. 63 Having been composed
effectively in seminar, the theory in its final form shows few traces of its
genesis, and any attempt to isolate either the order of ideas or their origin in
 FROM OSCILLATORS TO A NEW KINEMATICS 59

individual minds would here be so speculative as to be worthless. 64 But the

outcome was the use of perturbation theory, at which Born was the expert
among quantum physicists,65 to extend the oscillator treatment of dispersion,
in terms of difference equations, to the general behaviour of the atom. Born
argued that ,66
Since one knows that ... atoms react to light waves completely 'non-mechanically', it is
not to be expected either that the interactions between the electrons of one and the
same atom should comply with the laws of classical mechanics; this disposes of any
attempt to' calculate the stationary orbits by using a classical perturbation theory com-
plemented by quantum rules. For as long as one does not know the laws for the interaction
of light with atoms, i.e. the connection of dispersion with atomic structure and quantum
jumps, one is left all the more in the dark about the laws of interaction between several
electrons of the same atom.

He therefore considered whether it might be possible to extend Kramers's

treatment of dispersion, closer study of which "leads one to investigate
whether the method of quantisation used by him is not based on some
general property of perturbed mechanical systems."67 This introduction
followed the basic Gottingen approach: given the failure of classical mechanics
(noted explicitly by both Born and Heisenberg the previous year), and the
failure of the orbital model within this mechanics (as stressed by Heisenberg),
the success of the oscillator approach to dispersion suggested a search for new
quantum conditions based upon this approach, interpreted in terms of
difference equations. The reference to electron-electron interactions was
presumably a hangover from the seminar discussions, where Pauli's idea of
extending the oscillator approach to these interactions would have been
discussed, perhaps in connection with some of Born's work on the orbital
model published in 1923, in which he ran into considerable difficulties with
high frequency electron-electron coupling in the atom. 68 In fact, the generali.
sation proposed by Born did not yet extend to such interactions, the oscillators
continuing to be applied in effect to the atom (any interactions being repre-
sented as perturbations), rather than to the individual electrons.
To derive his new "quantum mechanics" Born first reviewed classical
perturbation theory; using a Fourier expansion of the Hamiltonian, suggested
presumably by the oscillator approach, he brought the formulae for systems
with and without external forces into the same form, and then went over
classical dispersion theory as an application of these formulae. Next he
introduced the oscillator representation, associating a given stationary state
with "emission resonators", v(n, n'), and "absorption resonators", v(n', n),
each corresponding to a higher harmonic of that state in the classical theory
60 CHAPTER 5

«vr); T = In - n' 1).69 The division of the oscillators into two classes, one of
which could only emit and the other only absorb, was conceptually unsatis-
factory, but it was a natural consequence of the reinterpretation of Kramers's
formula (2) as a difference equation, (3), and since Born did not afford the
oscillator treatment too much physical significance he was no more disturbed
than Kramers had been by. Breit's objections. Physically, an atom in a given
state could undertake infinitely many absorptions, but only finitely many
emissions (in the ground state, for example, none), and given the mathemat-
ical symmetry between emissions and absorptions in (3) the condition had to
be imposed that the relevant transitions were possible, or that the relevant
absorption or emission oscillators existed. Next, guided by the quantisation
process, h =nkh, of the old theory, Born attempted to connect the quantum
and classical frequencies of an unperturbed system, H 0 : 70
The following quantitative connection exists between the classical frequency (VT) and
the quantum-theoretical absorption frequency v(n', n). Let us imagine that the transition
nk ---+ nk = nk + Tk is performed in a 'linear' way; i.e. let us set for the action integrals
Jk = h (nk + iJ.Tk); 0 .:;; iJ. .:;; 1. Then we obtain on the one hand,

(vr) = ~ vkTk= ~ aHo Tk =~~ aHo ~ = J:.. dH o

k k afk h afk a iJ. h dJ.I. '
and on the other,
, 1
v(n,n)="h [Ho(n+T) -Ho(n»);

therefore v(n + T, n) = n
(VT) diJ. .... One can say that the ways in which v(n + T, n)
and (vr) are obtained from H 0 stand in the same relationship as differential coefficients
stand to difference quotients.

Born next considered the interaction process described by the perturbation

function AH1. For the classical perturbation energy he had obtained a

r
Fourier series expansion that was characterised, like the classical frequency,
aId
by the operator Tk ah = h dJ,L ,and he concluded that "we are as good as
forced to adopt the rule that we have to replace a classically calculated
quantity, wherever it is of the form f Tk ajk = -tz : ' by the liner average
or difference quotient r i a% Tk dJ,L =~ [4'(n + T) -lP(n)]." Applying
this transformation to the perturbation results he obtained Kramers's disper-
sion formula and an analogy in simple cases with Heisenberg's difference
equation quantum conditions.
 FROM OSCILLATORS TO A NEW KINEMATICS 61

Heisenberg himself began to work with the oscillator technique in the

Autumn, when he went to Copenhagen for a semester. His first paper there
was on a simple extension of the dispersion theory to allow for additional
spontaneous atomic transitions during the scattering process, and was written
jointly with Kramers.71 He recalled that both the idea, in response to a
treatment by Smekal based on the light-quantum concept, and the redaction
were due to Kramers,n and this was reflected in the terminology adopted:
opposed as he was to the light-quanta, Kramers relapsed into the terminology
of purely classical oscillators and waves. No mention was made of the virtual
oscillator concept, and BKS was referred to only as an example of how the
correspondence principle, on which the treatment claimed dependence,
might be applied.
Heisenberg's role in the dispersion paper seems to have been merely to
contribute some rigour to Kramers's physically inspired guesswork,73 but at
the same time he was also working by himself, extending the application of
the oscillator technique and removing it still further from Bohr's interpretative
context. This work, which was to be critical for the development of the new
kinematics, was on fluorescent polarisation. 74
Wood and Ellett had found that if a polarised light source was used to
stimulate fluorescent resonance radiation from mercury, the stimulated
radiation showed about 100% polarisation in a weak magnetic field, a result
that seemed compatible with the classical theory but not with that based on
the orbital quantum model of the atom. 7S In the absence of a magnetic field,
the polarisation of the incident light was maintained in the resonance radiation
on both theories, but the empirical and classically acceptable extension of
this result to the presence of a magnetic field clashed with the natural assump-
tion of equal statistical weighting of the magnetically induced multiplet states
in the orbitals model. In a short paper on the subject, Bohr had suggested that
the results might be obtained from the virtual oscillator theory, and Heisen-
berg now took up this possibility. 76 But whereas Bohr, though preferring the
virtual oscillator model explicitly here to the orbital model, had continued
to view the two models as compatible, and had insisted that the results did
not contradict his theory of atomic structure,77 Heisenberg took a slightly
different approach. He had always treated the virtual oscillator approach and
its parent the correspondence principle as constituting an empirical approach,
neutral as to any physical models. He had earlier compared the correspondence
principle, interpreted in this way, with the orbital model, and he had since
come to the personal conclusion that the model was untenable. What he did
now was to demonstrate this conclusion, comparing the orbital model with
62 CHAPTER 5

the oscillator representation, as physical model approach against symbolic

phenomenological approach.
Heisenberg based his oscillator treatment on some work by Dorgelo,
Ornstein and Burger, who had been studying related problems of multiplet
structure by analogy with the classical theory. 78 Consideration of this work
led him to the empirically-based assumption that the multiplet degeneracy
revealed in the presence of a magnetic field was a permanent feature of the
atom, the multiplet states being present (though indistinguishable) even in
the absence of a field. His use of this idea (and with it the suggestion of
solving a problem in the absence of a field by first introducing one, then
applying the usual quantisation, then setting its strength to zero) seems to
have led to an argument with Bohr, who criticised him as usual for ignoring
the fundamental principles of quantum theory,79 but it was undoubtedly a
major breakthrough in the treatment of complex spectra. He introduced it
together with the orbital model approach, in the context of the polarisation
problem with unpolarised incident light and no field; then, considering the
introduction of a weak field he used the new situation to decide between his
approaches. Classically, the resonance radiation remained unpolarised, and
this conclusion was carried through in the oscillator approach, being inter-
preted statistically as equal intensities of parallel and perpendicular polarised
components. On the orbital theory, however, the field acting on the orbit
produced a polarisation effect. And as Heisenberg wrote, "we have every
reason to believe that polarisation is not present."80 Having justified the use
of the oscillator approach and rejected that of the model, he reproduced
the Wood and Ellett results, obtaining predictions in full agreement with
experiment.
Heisenberg interpreted both his own work and that of Dorgelo, Ornstein
and Burger as extensions of the empirically based correspondence principle
approach to the problem of statistical weights, previously the domain of the
orbital model. And, successful as it was, he clearly found this extension
encouraging. He could not reject the orbital model on principle, for it was
still theoretically essential for both the calculation of frequencies and the
justification of the correspondence principle itself. But he could conclude
that any new theory should be developed from the "symbolic" , or essentially
phenomenological, nature of the oscillator treatment, this being the feature
that ensured its success. 81 The models were no longer useful, but restrictive.
 FROM OSCILLATORS TO A NEW KINEMATICS 63

THE RETURN OF PAULI AND THE GENESIS OF HEISENBERG'S

NEW KINEMATICS

Meanwhile, Pauli had refused to have anything to do with the virtual oscilla-
tors. But in the course of 1924 he had come back to quantum atomic theory
from his self-imposed exile, and by developing his earlier "phenomenological"
approach he had begun to unravel some of its problems. Before turning to the
virtual oscillator theory Heisenberg and Born, together with Lande, had
published a series of papers in support of the core model of the atom and the
associated use of half-integral quantum numbers. 82 Pauli had never been
happy with this model, and when he returned it was to show that it was
innately inconsistent. Looking once again at the anomalous Zeeman effect he
was able to show that according to the core theory the Zeeman splitting had
to depend on the atomic number, which it empirically did not. 83 To get
round this he suggested that the anomalous angular momentum and magnetic
moment, previously attributed to the hypothetical structure of the core,
should instead be transferred to the outer electron. Instead of counting the
core twice in its contributions to atomic properties, as Heisenberg had done,
Pauli counted the electron twice: 84

According to the interpretation suggested here, Bohr's 'Zwang' does not manifest itself
in a violation of the permanence of the quantum numbers in the coupling of a series
electron to the atomic core, but only in a characteristic Zweideutigkeit ["two-valued-
ness"] in the quantum-theoretical characteristics of the individual electrons in the
stationary states of the atom.

Soon after having reached this conclusion, Pauli read a paper by Stoner
containing a new scheme for the shell structure by which the atom was
related to the periodic table of elements, a scheme that was different from
Bohr's but completely natural and in excellent agreement with experiment. 85
One of Stoner's innovations was to assign values of the "inner" quantum
number, j + f' to each electron rather than treating it as part of the electron-
core interaction, a procedure that was clearly similar to Pauli's own treatment
of the anomalous Zeeman effect. This gave a set of three quantum numbers
for each electron, (n, I, j), and Stoner found that the number of electrons in
each shell was equal to twice the inner quantum number of that shell, 2(j + t ).
From this Pauli saw that the whole shell structure could be obtained very
naturally, independent of any atomic model, and without any recourse to
arguments from the correspondence principle, by giving each electron a fourth
quantum number, mj, - j :E;;; mj :E;;; j, and insisting that each state defined by
64 CHAPTER 5

a set of the four quantum numbers (n, /, j, mj) could represent, orbe occupied
by, just one electron. The Zweideutigkeit was absorbed into the new quantum
number, and the whole idea was expressed as his famous exclusion principle,
that no two electrons should occupy the same state in the atom. 86
Given this tremendous success of his approach, it was with greatly increased
fervour and confidence that Pauli once again expounded his views on the
future of quantum theory to Bohr in December 1924. 81 He suggested that for
"weak" people, who needed the support of well-defined orbits and mechanical
models, one could justify the exclusion principle on the basis that electrons in
the same orbit would crash. 88 But he explained that he had consciously
avoided the use of such terminology in his paper, and that he thought that
the future would involve not only an abandonment of the orbital concept but
also some fundamental changes in the kinematic concepts themselves: 89
The relativistic doublet formula [the Zweideutigkeit) appears to me to show unques-
tionably that not only the dynamic concept of force but also the kinematic concept of
motion of the classical theory shall have to undergo fundamental changes (it is for this
reason that I have avoided entirely in my work the designation 'orbit') ... I think that
the energy and momentum values of stationary states are something much more real
than 'orbits'. The (still unattained) goal must be to deduce these and an other physically
real, observable characteristics of the stationary states from the (fixed) quantum numbers
and quantum theoretical laws. However, we should not want to clap the atoms into the
chains of our preconceptions (to which in my opinion belongs the assumption of the
existence of electron orbits in the sense of the usual kinematics), but must on the
contrary adjust our ideas to experience.

In a footnote to the first sentence of this quotation Pauli, referring to the

"children" who lapped up Kramers's "picture book" (and meaning children
in terms of wisdom rather than years), noted that
Even though the demand of these children for visualisation [Anschaulichkeit) is partly
legitimate and healthy. this demand should still never count in physics as an argument
for the retention of a certain set of concepts. When the system of concepts is once
clarified. then will there be also a new visualisation.

The fundamental ideas of Pauli's proclamation were of course those that he

had advocated consistently over the years: the rejection of orbits, the use of
a phenomenological approach in the absence of anything better, and the
operational basis of concepts. But the last idea, which was the most funda-
mental, was here expressed more clearly and positively than it ever had been
in the context of the quantum thoery. Whereas Bohr stood by the permanence
of those fundamental physical concepts corresponding to the classical visuali-
sation, Pauli insisted that this visualisation must be put aside and new concepts
 FROM OSCILLATORS TO A NEW KINEMATICS 65

derived from experience. He even suggested an equation between "physically

real" and "observable" properties. Moreover, Pauli was also able for the first
time to make positive recommendations as to how these ideas should be
brought to bear on the quantum problem, adding to the above suggestions the
thought that the solution to the whole problem would come through the
hydrogen atom. If this relatively simple problem could be solved through a
new mechanics, he argued, then the principles of that mechanics would allow
the solution to be extended to the more general case. 90
This new and specific expression of Pauli's ideas was well timed, for it
coincided with Heisenberg's move in the very same direction. And in March
1925 Pauli visited Copenhagen for a few weeks, and argued out his views on
the future of quantum theory with both Heisenberg and Bohr. 91 Following
his conclusion on the fluorescent polarisation problem, disturbed by
Sommerfeld's recent discovery that the use of half-integer quantum numbers
failed after all for helium,92 and not yet happy with Pauli's treatment of the
anomalous Zeeman effect, Heisenberg had turned back to this problem. 93
And considering the application of the correspondence principle to the
theoretical derivation of Zeeman intensities he noted in his work that: 94

The application of the correspondence principle to the derivation of the selection rules
and intensities is legitimately only possible through the possession of unequivocal
mechanical models.

But as Heisenberg had himself demonstrated the existing mechanical models

were themselves invalid. And he argued that whether Pauli's Zweideutigkeit
was attributed to the core or the outer electron was essentially immaterial,
as both representations were equally artificial. 95 To obtain a satisfactory
treatment, he reasoned, the approach had to be an essentially empirical one,
but at the same time a proper theoretical derivation required some kind of
model. The problem was therefore to derive a new model, and a radically
new one at that, from the empirical results. And it was to this task that he
turned next.
Heisenberg's programme, based on a rejection of the existing model of the
atom and a reliance primarily upon the observed results, was now close to
that advocated by Pauli. And it was presumably at Pauli's suggestion that he
decided to concentrate on hydrogen and so arrived at G6ttingen in April
armed with a book of Bessel functions with which he hoped to improve the
mathematics of the correspondence principle. He then intended to "guess" a
symbolic scheme for the reaction of hydrogen to an external field, and so to
deduce a new model for hydrogen.96
66 CHAPTER 5

In fact, Heisenberg had to give up on hydrogen as being too difficult. But

by June, as is well known, he had found his scheme and completed his paper
on 'A theoretical reinterpret~tion of kinematic and mechanical relations'.97
In preparing this paper, he adopted both Pauli's phenomenological approach
and, after talking out the interpretation of the scheme with him during a
short visit to Hamburg, his operational ideas as well. 98 In its final form,
Heisenberg's new presentation was based on a restriction to quantities that
were in principle observable, and on a complete revision of kinematics. The
electron orbits were finally abandoned, and the electrons themselves were
replaced, as Pauli had earlier suggested, by systems of complex oscillators.
Pauli himself could write, following the completion of this paper, that he and
Heisenberg were as much in agreement as any two individuals could be. 99
Bohr, meanwhile, turned away from the world of publication to that of
contemplation.
 CHAPTER 6

THE NEW KINEMATICS AND ITS EXPLORATION

HEISENBERG'S NEW KINEMATICS

The avowed aim of Heisenberg's paper was "to try and establish a theoretical
quantum mechanics, analogous to classical mechanics, but in which only
relations between observable quantities occur." 1 The attempt was a confused
one, if only on account of the variety of conflicting notations,2 but the
fundamental idea was clear: to take over the classical equation of motion,

(l) ij+f(q)=O,

but to replace the classical acceleration ij and potential f(q) by quantum-

theoretical representations derived from their series or integral Fourier
expressions. In the case of the position function q(t) of the periodic motion
of an electron in an atom, the position itself was not observable: as Heisen-
berg and Pauli had long recognised, the kinematic conception of an orbit
was in this case operationally meaningless. 3 But the terms of the Fourier
expansion of the position could be directly related to observables. Classically
the position vector of an oscillating electron could be expanded as a Fourier
series,

(2) q(n, t) = ~ qoln) eiw(n)at,

0:=-00

and the radiation corresponding to each harmonic was proportional to the

real part of the Fourier component, IR {qa(n) eiw(n)at}. In quantum theory
the Bohr frequency condition,

1
(3) v(n, n -a)="h {W(n)- Wen -a)},

with Wen) the energy of the nth state, led to the requirement that the har-
monic components take the form of expressions q(n, n - a) eiv(n, n - a) t,
corresponding to pairs of states or transitions. Heisenberg assumed that the
observable radiation was again given by their real parts. In order to construct
a mechanics of observable quantities he extended this form of representation
67
68 CHAPTER 6

as an ensemble of functions of possible transitions to a general quantity,

deriving for the product of two quantities (now assumed to be scalars)
x(t) = {x(n, n - a) eiv(n, n - a)t},
yet) = {yen, n - a) eiv(n, n - a) t},

(4) z(t) =x(t)y(t) ={zen, n - (3) eiv(n, n - Int}

= {~ x(n, n - a) yen - a, n - (3) eiv(n, n -Int}.
a=-oo

Since the occurrence of quantum transitions was necessarily ordered, the

multiplication was non-commutative, and Heisenberg noted that one would
therefore have to replace ambiguous classical products by symmetric alter-
t
natives, x(t)y(t) -+ (xy + yx). But apart from this the classical expressions,
both scalar and by Implication also vector, could simply be replaced in the
classical equations of motion by their quantum equivalents,
qa(n) eiw(n)at -+ q(n, n - a) eiv(n, n - a)t, etc.

The solution to the equation of motion (1) had been given in the old
quantum theory by a quantisation of the action,

(5) J = 1 pdq = 1 mqdq = nh,

which could not be translated into the new quantum terminology. But by
differentiating with respect to 11 Heisenberg was able to obtain a form that
could be so translated, giving a new quantisation condition,4

(6) h =41Tm ,£00 {Iq(n, n +a)1 2 v(n, n +a)

a=O
- Iq(n, n - a) 12 v(n, n - a)}.

This condition had already been derived from the virtual oscillator theory
from which, on the technical level, Heisenberg's own theory had devel-
oped. But in the former case one had had to represent the electron by a
set of classical oscillators with the observed transition frequencies and then,
working with these classical oscillators, replace the resulting differential
equations of a certain type by difference equations. This procedure had had
no real theoretical foundation, and the use of the oscillator model, on which
the oscillator radiation took the classical continuous form but the transi-
tions remained discrete, had entailed a departure from energy-momentum
 THE NEW KIN EM A TICS AND ITS EXPLORATION 69

conservation that had now been shown, by the experiments of Geiger and
Bothe, to be impossible. s In Heisenberg's energy-conserving theory, the
quantum-theoretical solution, given by the equation of motion (1) and the
quantisation condition (6), followed straight from the replacement of the
kinematic expressions.
The new kinematics constituted a major breakthrough in the treatment
of quantum phenomena, introducing explicitly the break from classical kine-
matics and restriction to observables that had long been advocated by Pauli.
Despite these conceptual innovations, however, it was still in many ways
unclear and lacking in direct applicability. Heisenberg was able to derive
the dispersion formula of the old virtual oscillator theory as well as the
approximate energy levels of the one-dimensional anharmonic oscillator
and rotator, but even these derivations lacked rigour. He was unable to
establish energy conservation as a general feature of the theory (though
he had established it for his simple examples and convinced himself of its
general existence),6 and was unable to apply this theory even to the relatively
simple problem of the hydrogen atom, let alone to systems of more than
one electron. 7 The physical significance of the restriction to observables
was unclear and, as he wrote to Pauli, the fundamental problem remained
as to "what the equations of motion really mean, when one treats them as
relations between transition probabilities."s If the traditional kinematics
was invalid, how could the equations of motion, derived conceptually from
this kinematics, be justified?

THE RECEPTION OF HEISENBERG'S THEOR Y, AND BORN'S

MATRIX MECHANICS

Not surprisingly, since he was himself responsible for most of its concep-
tual innovations, Pauli greeted Heisenberg's theory with delight, reporting
to Kramers that "on the whole I believe that I am now close to Heisenberg
in my scientific opinions, and that our opinions agree in everything as much
as is in general possible for two independently thinking men."9
Also enthusiastic was Born, under whom Heisenberg was nominally
working at the time. Born later recalled that he had discussed with Heisen-
berg and Jordan, before Heisenberg's innovations, the possibility that
"[transition amplitudes] might be the central quantities and be handled
by some kind of symbolic multiplication" ,10 and on seeing Heisenberg's
paper, that 11
70 CHAPTER 6

I began to ponder about his symbol multiplication, and was soon 80 involved in it that
I thought the whole day and could hardly sleep at night .... And one morning ... I
suddenly saw the light: Heisenberg's symbolic multiplication was nothing but the matrix
calculus, well known to me since my student days.

In fact Born, who was trained as a mathematician, must have been rather
more familiar with matrices than thest:' well-known recollections suggest,12
and this suggests that his initial concern may have been with the physical
implications of Heisenberg's work, and with its advantages over the virtual
oscillator theory and de Broglie's wave theory of matter, rather than with its
"symbolic multiplication" .13 But however we may reconstruct the details of
his reaction it is clear that it was favourable and that it did lead him within a
few days to the matrix formulation in terms of which Heisenberg's theory
came to be known. Born had long ago expressed a willingness to abandon
the space-time description of the inside of the atom;14 and like Heisenberg
he had appreciated and contributed to the heuristic value of the virtual
oscillator theory, though without attributing to this any physical signi-
ficance. 1S He had anticipated the Geiger-Bothe results on the preservation
of energy conservation, and had responded to Einstein's extension of the
Planck-Bose statistics to a material ideal gas, the electron scattering results
of Davisson and Kunsman, and the barrier penetration by slow electrons
demonstrated by Ramsauer by adopting a version of de Broglie's theory of
matter waves as a physical alternative to the virtual oscillators. 16 Just before
assimilating Heisenberg's paper he had expressed the view that "the wave
theory of matter could be of very great importance." 17 The results of Ein-
stein, Ramsauer, Davisson and Kunsman all displayed a wave-particle duality
of matter that found natural expression in de Broglie's theory, and it was
with these results that Born and Franck had been largely concerned during
the Spring,18 and with the Planck-Bose statistics of light that Born and
Jordan had been concerned in their most recent applications of the virtual
oscillator theory .19 But Bohr had rejected the matter-wave hypothesis out-
right and Born, as he recalled, had also been following closely Heisenberg's
attempt to derive a new and consistent physics from the starting point of
the observable transition amplitudes. 20 His immediate sympathy with Heisen-
berg's theory, despite the confusion and complication of its presentation,
was therefore natural; and given his mathematical ability the re-expression
of this theory as a matrix mechanics was more or less inevitable.
Adopting the matrix notation and looking at the physical implications of
non-commutativity, Born saw at once that the quantisation rule (6) gave the
value h/21Ti for the diagonal elements of the position momentum commutator,
 THE NEW KINEMATICS AND ITS EXPLORATION 71

pq - qp. Convincing himself that the off-diagonal elements of the ex-

pression would probably be zero, he suggested the resulting commutation
relationship,21

(7) h
pq - qp =-2 . I,
1fl

and retired to Switzerland for a much needed holiday,22 leaving his assistant
Jordan to prepare a joint paper that was completed on his return and received
on 27 September. 23 In this paper Heisenberg's sets of Fourier coefficients
were expressed as Hermitian matrices, x(n, n - q) -+ x(nm) and the com-
mutation relationship, the frequency law and energy conservation were
derived as general results for non-degenerate systems. 24
By defining an artificial 'symbolic differentiation' of one matrix with
respect to another, Born and Jordan were able to express the classical equa-
tions of motion, taken as matrix equations, for a general Hamiltonian energy
function, H, in the canonical form

(8) . aH • -aH
q = ap' P=aq'
For non-degenerate systems, where any change of state was associated with
a non-zero change of energy, they were able to show that the time deriva-
tive of a general quantity g(nm) e 21riv (nm)t, given by g = 21fiv(nm) g(nm)
e 21rill (nm)t, was zero if and only if its matrixg(nm) were diagonal. From this
and from Heisenberg's quantisation rule (6) they deduced the commutation
relationship (7) and, from substitution in the general result,

(9) • 21fi
g=h(Hg-gH),

the frequency law (3) and energy conservation,

(10) Ii = o.
In the last section of the paper, Jordan noted Heisenberg's implicit assumption
that / q(nm) /2 determined the transition probabilities (Le., that Heisenberg's
IR {q(n, n - a) eiv(n, n - at)t} did in fact correspond to the observable radia-
tion), and proceeded "to see in what way this assumption can be based upon
general considerations" .25 Applying the matrix mechanics to the electro-
magnetic field, he found that the mean radiation, identified as the diagonal
sum of the radiation matrix, was indeed determined by the / q(nm) /2.
72 CHAPTER 6

THE MATRIX TRANSFORMATION THEORY OF BORN,

HEISENBERG AND JORDAN

Born and Jordan noted in their paper that the canonical equations of motion
(8) and Heisenberg's quantum condition (6) could be replaced as foundations
for the theory by the equivalent assumptions of energy conservation (10)
and the commutation relationship (7). But such a fundamental role for the
commutation relationship is first apparent in a letter from Heisenberg to Pauli
of 18 September, and this letter also contains the first attempt to express
the new theory in terms of a theory of transformations, an attempt which
was to have far-reaching consequences. 26 Heisenberg had received details
of the new matrix formulation by 13 September and in his reply to Jordan
he had set out immediately to extend its applicability, drawing for this
purpose on classical perturbation theory.27 By the time he wrote to Pauli
a few days later he had based this treatment on the supposition that any
p , q satisfying the commutation relationship and for which the Hamiltonian
was diagonal would represent a solution to the problem, deriving such a
p , q for the perturbed problem by a transformation of known unperturbed
solutions. 28 Heisenberg's transformation theory, derived from that used in
the classical case, was somewhat unwealdy, but on receiving the idea Born
quickly suggested a simpler form of transformation,29
(11)
This was the standard form for a matrix transformation leaving the matrix
equations invariant, and had indeed appeared as such in the introduction to
the Born-Jordan paper. 30 It preserved the commutation relationship, and
together with a natural extension of this relationship to several degrees of
freedom it formed the core of the definitive formulation of matrix mechanics
in the 'three-man-paper' of Born, Heisenberg and Jordan. 31 The problem,
first stated by Heisenberg and reformulated by Born, could now be expressed
as follows: 32

Given the canonical equations of motion for a known Hamiltonian,

(12) . aH
q=a,;'
and any Hermitian matrix quantities Po, q 0 satisfying the commutation relations,
 THE NEW KINEMATICS AND ITS EXPLORATION 73

to fmd a transformation matrix S such that P = SPoS-I, q = Sq oS-I gave

H(pq)=SH(Po,qopr l = W,
a diagonal matrix.

For the case of small perturbations, H =Ho + AlII + ",2H2 + ... , P =Po +
).pI + ... ,q = qo + I\ql + ... , the solution S = 1 + AS I + ... could be found
by calculating the terms successively.

PHYSICISTS AND MATHEMATICIANS: DISAGREEMENT AND

DIVISION

The three-man-paper, first drafted by Born and Jordan and fmally completed
on 26 November, after Born had left on 28 October for America, by Jordan
and Heisenberg, was presented as a joint effort. But there were in fact strong
differences of opinion between the authors, and these were manifest both in
their paper and in their ensuing research. Born recalled that he had originally
asked Pauli to help with the matrix mechanics, but that he had been given
a "cold and sarcastic refusal" on the lines of: "Yes, I know you are fond of
tedious and complicated formalisms. You are only going to spoil Heisenberg's
physical ideas by your futile mathematics." 33 Whether accurate or not, this
recollection conveys a true impression. Pauli had past experience of working
with Born,34 and about this time he noted of that earlier work, which he had
not found very satisfying, that "the effort expended did not correspond to
the results achieved, especially as these were chiefly negative."3s He wrote
to Kronig in October that "one must next attempt to free Heisenberg's
mechanics from the GOttingen Gelehrsamkeitsschwall [literally, "torrent of
erudition"] and expose still further its physical crux," 36 and in November
Heisenberg wrote to him that 37
I'm still pretty unhappy about the whole theory, and was thus glad that you were so
completely on my side in your views on mathematics and physics. Here I'm in an environ-
ment that thinks and feels the exact opposite, and I do not know if I'm not just too
stupid to understand mathematics.

A major factor in the success of GOttingen as a centre for theoretical physics

lay in its unique composition: the th,ree departments of experimental physics,
theoretical physics and mathematics were run by three men, Franck, Born
and Hilbert, all of whom were outstanding in their respective fields, and who
worked in extremely close collaboration with each other. Much of Born's
work on the quantum theory seems to have been prompted by the influence
74 CHAPTER 6

of Franck (who himself worked closely with Bohr),38 while the idea of his
earlier difference equation formulation of this theory had most probably
stemmed from a seminar in Hilbert's department. 39 Later, the extremely
close links between Hilbert and Born were to be of crucial importance for
the formulation of quantum mechanics. But while bringing great advantages,
such a situation also led naturally to tensions. Now, as Heisenberg wrote
to Pauli,40

GOttingen splits into two camps, one which with Hilbert ... speaks of great success,
achieved through the introduction of matrix rules to physics, the other which with
Franck says that we still cannot understand the matrices.

In the Hilbert camp, Heisenberg placed Weyl and, by implication, Born, who
had been carried away by the mathematics of the new theory. Jordan, as far
as may be gathered from his subsequent research, took something of a middle
path. But Heisenberg was firmly in the physical camp. The problems resulting
from his fundamental innovations, problems as to observability and the
physical significance of the new kinematics and commutation relations,
seemed to play no part in Born's attempts to elaborate on the mathematical
theory. To Heisenberg, indeed, the whole concept of a "matrix mechanics"
represented a rejection of the physical problem, and he expressed to Pauli
a serious intention of replacing the terminology by something physically
meaningful, such as "mechanics of quantum-theoretical quantities" .41 He
also expressed serious doubts about the fundamental role that Born and
Jordan attributed to the transformation theory, and wrote to Pauli on this
point that "you have shown in Hydrogen how one actually integrates" (in
a mathematical tour de force Pauli had obtained a direct solution from the
theory for the hydrogen atom) "and so that rest is just formal rubbish."42
Pauli, mathematically more experienced,43 was less upset by the new devel-
opments; but he seems nevertheless to have agreed wholeheartedly with
Heisenberg about the need for physical comprehension, and once again to
have found himself on the opposite side of the fence from Weyl.
Given the above division, it is not surprising that within the context of
the three-man-paper, the discussion surrounding it and the work consequent
upon it, there were several quite distinct and divergent lines of development.
In the three-man-paper itself, Heisenberg and Jordan derived the laws of con-
servation of linear and angular momentum and thence, for a non-degenerate
system, the standard selection rules and normal Zeeman intensities.44 In a
paper completed in January, Pauli applied the theory successfully to the
hydrogen atom,45 and those applications between them consolidated the
 THE NEW KINEMATICS AND ITS EXPLORATION 75

prospects for the theory's success. In two papers completed the following
Spring, Jordan set the transformation theory upon a rigorous footing, pro-
viding that every canonical transformation could be put in the form p =
SpOS-l, q = SqoS-l, and that every point transformation was canonica1. 46
Although the transformation theory remained practically restricted to the
case of small perturbations where the unperturbed solution was known, it
was thus established as a theoretically generally valid approach to the solu-
tion of a general physical problem, and this was to be important for the
further development of the theory. Of more immediate importance in this
respect, however, were those developments reflecting the particular concerns
of the authors: Jordan's application of matrix mechanics to radiation fluctua-
tions and quantum statistics, Born's development of the mathematics of the
theory so as to generalise its applicability, and the discussion by Heisenberg
and Pauli of some of the theory's physical implications.

PAULI, HEISENBERG, AND THE PHYSICAL SIGNIFICANCE OF

THE NEW KINEMATICS

Heisenberg had written to Pauli on 16 November that "I have taken great
pains to make the work more physical than it was" ,47 and his efforts had
born some fruit. He had persuaded Born to include a physically based
form of matrix differentiation alongside the physically incomprehensible
"symbolic differentiation" of the Born-Jordan paper,48 and he had written a
wholly physical introduction to the three-man-paper in which the word
"matrix" was completely avoided. 49 He here layed emphasis upon both the
observability criterion and the new kinematics, and posed the fundamental
problem of establishing a general relationship between the symbolic quantum
theory and the classical conceptions of the observed world. But the physical
insight to the new theory that he sought continued to elude him. On 17
November, however, Pauli wrote a letter to Bohr, for general consumption,
in which he included something "on the principal questions that are still
left open in the new theory": 50
TItis theory is so far cut out only for those cases in which all points remain in fInite
spaces. In its present formulation, it is still not fItted for example, either to include
impact phenomena or to include the problem of understanding simultaneously coupling
and interference. Generally we still have no logically uniform theory that includes all
the applications of classical theory in the borderline case of high quantum numbers.

So far, Pauli was simply voicing generally recognised facts. But he then pro-
ceeded to offer some ideas of his own:
76 CHAPTER 6

Perhaps the following hits the right direction for possible further progress. In the new
theory, all physically observable quantities still don't really occur. Absent, namely, are
the time instants of transition processes, which are certainly in principle observable
(as, for example, are the instants of the emission of photoelectrons). It is now my fum
conviction that a really satisfying physical theory must not only involve no unobservable
quantities, but must also connect all observable quantities with each other. Also, I
remain convinced that the concept of 'probability' should not occur in the fundamental
laws of a satisfying physical theory. I am prepared to pay as high a price as you like for
the fulftlment of this desire, but unfortunately I still do not know the price for which
it is to be had.

Expanding on his statement that the time instants of transition processes

were absent from the new theory, he added a short footnote:

Instead of e21rivt one could write just as well e 21riv or something like that, and define
time derivatives of a quantity q as 21ri(Wq - qW)/h. The general problem is then to set
up wider concepts that embrace the actual applications one has made of the classical
space-time picture. The position is now that the concept 'duration' [zeitlicher AblaufJ
of a process, and in particular the concept of 'time period' or 'frequency of oscillation'
has become wholly formal. The formal character of the frequency condition ... is a
consequence of the formal character of time. and not of the formal character of energy.

On where this led, Pauli was still unclear, but he thought that one might try
defining time in terms of energy and proceed from there: s1 since all time
measurements depended upon periodic processes, this was also possible
within the existing quantum mechanics, and consistent with his general
operationalist ideas. 52
In due course Heisenberg read Pauli's letter,53 and writing to him on 24
November he offered sincere thanks for such a clear and helpful exposition
of Pauli's views, together with some thoughts of his own prompted by these
views: 54

Your problem of the 'duration' naturally plays a fundamental role, and I've thought
over several matters for domestic use. First, I believe that one can distinguish between
a 'coarse' and a 'fme' duration. When, as in the new theory, a point in space has no
longer a fIXed place, or when this place is still only defmed formally and symbolically,
then the same is true also of the time-point of an event. But there is always given a rough
duration, as also a rough place in space: with our geometric picture we shall still be
able to achieve a rough description of the phenomena. I think it is possible that this
rough description is perhaps the only one we may ask for from a formalism. Now the
beautiful thing is that for purely periodic motion evidently not even such a coarse lapse
can be defmed; it seems to me that the formulae do not admit of such an interpretation
(i.e. one knows of the electron only that it is somewhere close to the core). But if one
has an aperiodic orbit, i.e. a Fourier integral, then the - let us say - infrared part of
 THE NEW KINEMATICS AND ITS EXPLORATION 77

the spectrum agrees with the classical theory, the usual calculating rules being valid eo
ipso in good approximation (the better the longer the wave) - and just this infrared
part indeed gives the coarse duration! A motion sufficiently like uniform rectilinear
motion would thus be as classical as ever possible. But as soon as purely periodic motions
are superposed our space-time presentation again fails completely (Compton effect).

In the earlier derivation of the new kinematics it had been Pauli who had
provided the (operationally based) conceptual innovations; and Heisenberg
who had put these into a practically viable form. Now, in their struggle to
understand the physical implications of this kinematics, a similar procedure
was repeated. Pauli raised the fundamental problem of the defmition of
time in the new theory, and Heisenberg linked this problem with that of
space, and with Bohr's correspondence principle. The latter was no longer
an integral part of the new theory, which had its own foundations built
upon an altogether different type of correspondence. But it had played a
crucial part in the development of this th~ory, and in Heisenberg's search
for an understanding of the new type of correspondence, and of the new
kinematics, it was to be a valuable guide. Looking at the new theory in the
light of Pauli's remarks, Heisenberg saw that it included no such things as
space and time in the classical sense. But in order to correspond with the
classical theory and classical visualisations it had to provide for approximate
space-time specifications, and for time-dependent quantities these took the
form of time averages, given by the diagonal sum of the matrix. These quan-
tities were thus specified the better, classically, the smaller the contribution
of off-diagonal terms to the whole matrix. In particular the quantity position,
whose off-diagonal coefficients corresponded to transition probabilities, was
most accurately specified when only small transitions were possible, i.e.,
for long wavelengths (low frequencies) and especially for largely aperiodic
motions. The closer one got to uniform rectilinear motion, the closer one got
to zero radiation and a purely diagonal position matrix, allowing in principle
(though the theory could not yet cope with it) a precise position specification.
As to the specification of time, however, Heisenberg still seems to have
been confused; for while defending the overall classical nature of uniform
rectilinear motion, he followed Pauli's suggestion of identifying long wave-
lengths (large oscillating periods) with coarse time specifications.
The full relationships between space, time, momentum and energy were
not yet apparent in matrix mechanics, and it would be over a year before
Heisenberg could see things clearly enough to formulate his uncertainty
principle. But some of the ideas behind that principle were already present
in his letter to Pauli. In particular, the idea that a rough classical description
78 CHAPTER 6

was all that could be expected from the formalism, combined with the basic
hypothesis that this formalism was concerned with a theory of observables,
was to be of crucial importance.

JORDAN AND HEISENBERG ON QUANTUM STATISTICS

Before he was interrupted by the arrival of Heisenberg's new theory, Jordan

had been largely concerned with the problem at the centre of the wave-
particle paradox of light in quantum theory, that of the statistics of black-
body radiation. In a joint paper with Born he had derived Planck's law
(using Bose's form of the statistics) from the virtual oscillator theory,55
and he had followed this up with a paper drawing on Einstein's theory
of the ideal gas. 56 Now, in the three-man-paper, he re-examined Planck's
law and Einstein's energy fluctuation formula from the point of view of
matrix mechanics. 57
One of the most striking features of black-body radiation theory was that
it had remained, over the years, almost entirely unconnected with the atomic
theory on which it should, ideally, have depended. Whether formulated as
a theory of waves or of light-quanta it was dependent upon a non-classical,
physically incomprehensible, and apparently arbitrary assignment of a priori
probabilities, and this was as true for the recent derivation proposed by Bose
as for all the earlier attempts. 58 Only in de Broglie's theory was the assign-
ment of probabilities given any physical justification, and even this was
vague, and dependent upon the highly speculative hypothesis of matter
waves. 59 In the light-quantum terminology of Bose's theory, one had to
assume for the distribution of light-quanta over unit cells, of volume h,
in phase space a formula that was based upon the non-classical assumption
that the light-quanta be treated as indistinguishable (and thus somehow
non-independent) for the calculation of probabilities. Working independently,
de Broglie had associated each light-quantum with a phase-wave, and had
calculated the distribution of quanta from the requirement that the resulting
wave pattern should be stationary. Since the number of waves thus associated
with a region in phase space was equal to the number of unit cells into which
that region could be divided, de Broglie's derivation closely paralleled Bose's,
but with the advantage that the non-independence of light-quanta required
could be given some sort of physical explanation: 60

If two or more atoms or light-quanta have exactly superposing phase waves so that one
can say in consequence that they are transported by the same wave, their movements
 THE NEW KINEMATICS AND ITS EXPLORATION 79

cannot be considered as entirely independent and the atoms can no longer be treated
as distinct unities in the calculation of probabilities.

Following Einstein's extension, inspired by de Broglie's work, of the Bose

statistics to the distribution of an ideal gas, de Broglie's theory generally
and his derivation of Planck's law in particular had received close attention
from Born and Jordan in GOttingen. 61 Now, wishing to investigate the
physical significance of the new kinematics and its implications for the
fundamental wave-particle problem of light, it was natural that Jordan
and Heisenberg should tum to the black-body context and to the question
as to how de Broglie's conceptualisation of the Bose statistics carried over
into matrix mechanics: might these statistics even be derivable from the
new kinematics? The derivation of Planck's law itself, being fundamentally
an exercise in statistical mechanics, did not prove amenable to this path
of attack, but the derivation of Einsteins's fluctuation formula,
£2 _
(13) (E-E)2 = - - +hvE,
zv V

with the bars signifying mean values, V a volume of phase space and Zv =
81TV 2 dv/c 2 ,
did. The quantum-theoretical formula for the mean square
energy fluctuation of a radiation field had been derived by Einstein through
consideration of Planck's law, but it was fundamentally, in the terminology
of classical theory, a formula for the fluctuation due to interferences in a
wave-field. Generalising the identification between de Broglie's phase waves
and Bose's cells in phase space, Jordan noted that "if waves are propagated
with a phase velocity v in an s-dimensional isotropic part of space, V = LS,
the number of eigenvibrations for the frequency range dv is equal to the
number of [unit] cells ... , and this in fact holds for arbitrary s, hence, e.g.,
also for vibrating ... strings." 62 The number of quanta in an appropriate
cell corresponded to the quantum number of an oscillator. If one could derive
Einstein's fluctuation formula from the interferences in a vibrating string,
Jordan reasoned, one could in principle extend the calculation to more
general cases of wave fields, and thence to black-body radiation, eventually
deriving from the quantum theory of the wave field all the phenomena
associated with light-quanta. 63 And in the three-man-paper he was indeed
able to derive the required fluctuation for a vibrating string.
In the light of its possible implications, Jordan's result was one of the most
important achievements of matrix mechanics to date. It brought black-body
radiation phenomena in principle within the scope of the new theory, and
80 CHAPTER 6

opened the way for the future quantum field theory. But it did not as yet
lead to the Bose statistics and it did not as yet answer the conceptual ques-
tions that were bothering Heisenberg. It offered no insight to the extension
of the wave-particle problem to matter and the ideal gas, and did not even
approach the physical significance of this problem in respect of the new
kinematics. Heisenberg wrote to Pauli in November 1925 expressing the hope
that from an analysis of the grounds on which the 'new kinematics led to
Einstein's fluctuation formula all the essential features of the light-quantum
theory might be rediscovered. 64 But the conceptual problems entailed were
too much for him, and besides, as he had written earlier when first reporting
Jordan's work, he wished he knew more about statistics. 65
Given this dual obstacle to the development of his understanding, Heisen-
berg could apparently make little of Jordan's achievements, but he was
determined to make more progress toward comprehending the quantum
statistics. The following Spring, at the end of a paper in which he extended
the matrix mechanical transformation theory to the realm of many-body
problems, he came back to the statistical problem in the material context
of the distribution of electrons in an atom - the 'statistical weights' riddle
of Bohr's atomic theory.66 This riddle had previously been solved by Pauli's
exclusion principle, but Heisenberg noted that neither this principle nor the
Bose-Einstein statistics had yet been derived from matrix mechanics. And
from the consideration of a characteristic resonance phenomenon he deduced
that both were in fact compatible with it and indeed - applied to the same
particles - with each other. Considering two identical particles, he argued
that for any solution to a given problem there must be a second solution,
obtained by switching the bodies round, and that 67

If only one of the two systems occurs in nature, then on the one hand this admits a
reduction of the statistical weights [as in the Bose-Einstein statistics] ; ... [but] on the
other hand Pauli's exclusion of equivalent orbits is of itself fulfded.

Generalising to a system of n identical particles, he noted that there was

again a reduction of statistical weights, from n! to 1, and he again found the
Bose-Einstein statistics to be "in harmony with Pauli's exclusion".68
Though Heisenberg was apparently unaware of the fact, Fermi had already
demonstrated that the exclusion principle led to the completely different set
of statistics associated with his name,69 and Heisenberg's effort emphasises
that, as he later admitted,70 he was still very confused indeed about statis-
tics. 71 But it also emphasises his continuing concern with their interpretation,
which was later to feature in his struggles toward the uncertainty principle.
 THE NEW KINEMATICS AND ITS EXPLORATION 81

BORN'S MATHEMATICAL APPROACH AND THE OPERATOR

FORMALISM

While Heisenberg concerned himself with the meaning of the new theory
and Jordan with its applications, Born concentrated upon extending its
applicability and rigour through a generalisation of the mathematics. In the
three-man-paper he replaced the matrices a(nm) with bilinear forms, A (x, y) =
L L a(nm) xnx:-Z. This presentation had the advantage of bringing the theory
rnt'S the realm of a more familiar algebra, and it enabled Born to prove the
uniqueness and - for finite variables or for bounded forms of infinitely many
variables - the existence of an energy solution to the matrix mechanical
transformation problem.72 Through the introduction of a traditional eigen-
value terminology it also pointed the way toward a treatment of degenerate
cases, associating these with the familiar problem of the occurrence of mul-
tiple eigenvalues Wn in the solution, W = (onm Wn ). Continuous spectra,
corresponding to aperiodic phenomena, still caused problems, for the theory
was still based upon the periodic behaviour of electrons in atoms; uniform
rectilinear motion, for example, was completely outside its framework.
But cases in which a Fourier integral representation was possible could be
handled by the simple replacement of the summation sign of the bilinear
form with an integral,
L Wnxnx~ - jW(¢)y(¢)y*(¢) d¢.
n
Apart from the problem of relating transition probabilities to amplitUde
densities instead of to discrete amplitudes, all went through as in the discrete
case. In both cases the mathematical theory needed for the existence proof
had been developed, by Hellinger, only for bounded forms. But Born felt
justified in carrying the theory over to unbounded forms also, arguing that
"Hellinger's methods obviously conform exactly to the physical content of
the problem posed." 73
At the end of October, Born went as a guest lecturer to MIT, where
Wiener, who had visited GOttingen the previous year, was professor of mathe-
matics. Wiener recalled that Born had arrived very excited and searching
for a further generalisation of matrix mechanics. 74 By divine plan, coincidence
or whatever, Wiener had only six months previously completed a paper
on "the operator calculus",75 and this paper, which was very much in the
Gottingen tradition of the fusion of rigorous mathematics with the established
techniques of physics, provided just the generalisation required. Operators
shared the non-commutative property of matrices but not their restrictions,
82 CHAPTER 6

and Born and Wiener were able to build up a general formulation of the new
quantum mechanics in which even those aperiodic phenomena that had not
been covered by Born's integral formulation could now be treated. 76 They
defined an operator q in a completely general way as "A rule in accordance
with which we may obtain from a function x(t) another function yet),
which we symbolise by yet) =qx(t)." 77 Using the function
(14) x(t) =r, e21riW nt/h x n ,
n

they showed that an operator q could be derived from any matrix or Her-
mitian form (qnm) by putting
(15) 1 J.T ds . q(t, s),
q = lim
T ..... OO
-2
T -7

where
(16) q(t, s) = r, r, qnm xm(t)x~ (s) = r, r, qnm e2 7ri(Wnt - Wms)/ h.
nm nm
If Ym = ~ qmnxn, then yet) = qx(t). To derive a matrix from the operator
q, they applied the operator e-27riWt/h q to the function e27riWt/h, generating
(17) q(t, W)=e-2rriWt/hqe27riWt/h.
They then defined

(18) qvw = lim

T ..... OO
1
2T
r
'-7
q(t, W)e-27ri(V - W)t/h dt

= lim _1_ JT e-21riVt //l q e27riWt/h dt,

T ..... OO 2T -7

which was the reverse process of the above. Not all such integrals converged,
and they showed later that uniform rectilinear motion in fact gave divergent
(oscillating) integrals and so had no matrix representation. But since the
operators were themselves quite generally defined, this motion was as answer-
able to the operator mechanics as was any other. 78
Born and Wiener showed that their operators obeyed exactly the same
rules as did matrices for multiplication, etc., and that since the operators
D = a/at and 2rriW/h acted equally on the functions used to connect matrix
and operator representations one could replace the time derivative of a matrix,
hi
q = 2 (Wq - q W), by that of an operator, q = Dq - qD. The canonical
equations of motion and commutation relationship were also reinterpreted
as operator equations. 79
 CHAPTER 7

WA VE MECHANICS AND THE PROBLEM

OF INTERPRETATION

SCHRODINGER'S WA VE MECHANICS AND ITS RECEPTION

Born's operator fonnulation opened up a host of new avenues, which he

clearly intended to explore in further publications. 1 But before he could do
this, events took a dramatic tum with the advent of Schrodinger's wave
mechanics. 2 In his first communication of this new theory, submitted at the
end of January, Schrodinger transfonned the time-independent Hamiltonian
partial differential equation of motion,

as
(1) H(q, P = aq ) = E,
into a fonn that could (at least for the one-electron problem with constant
mass) be expressed as a quadratic form of a new unknown 1/1, S =k log l/I, and
of its derivative a1/l/aq, set equal to zero:
k a1/l
(2) F=H(q,~, aq )-E=O.

Seeking solutions in which l/I was a product (S a sum) of functions of individual

coordinates, he then treated the problem as a variational one, asking for real
finite continuous unique-valued and twice differentiable eigenfunction
solutions 1/1, and corresponding eigenvalues E, for which the integral of the
quadratic form was an extremum,

(3) f
5 F dq =O.
Applied to the hydrogen atom, this gave a continuous spectrum of possible
solutions for all E > 0 (hyperbolic orbits), but only a discrete finite set of
solutions, corresponding to the Bohr energy levels, for E < O.
In his second communication, a month later, Schrodinger expanded on his
treatment, placing it in the context of Hamilton's and de Broglie's work, and
demonstrating for the general (time-dependent) case the previously assumed
connection between the Hamiltonian problem (1) of particle mechanics and
the variational problem (3). He noted that the variational problem had to be
conducted in multi-dimensional configuration space rather than in ordinary
83
84 CHAPTER 7

Euclidean space, and the solutions l/J, uninterpreted in his first communication,
corresponded to a wave form in configuration space. Showing that, as for the
de Broglie phase waves, the group velocity of his configuration-space waves
corresponded to the velocity in the particle picture, he adopted de Broglie's
conclusions from the analogy with geometric and wave optics and suggested
that the 'true' mechanical phenomenon should be seen as represented by
wave processes in configuration space, the particle picture being merely an
approximation in cases of well-defined wave groups. In his first communica-
tion he had not mentioned that, since his solutions l/J were physically unde-
fined, he had in fact abandoned the orbital model of the atom and replaced it
as in matrix mechanics by a purely symbolic scheme. But he now stated this
fact explicitly and explained that in his view the abandonment of the orbital
model was necessitated by the fundamental wave nature of phenomena, to
which a particle approximation could not in this case be given. And while
the matrix-mechanical oscillators remained symbolic and physically incom-
prehensible, his own wave oscillations constituted a perfectly viable physical
picture. He then built up the new wave mechanics from scratch, deducing
the form of wave equation he had previously derived for hydrogen from the
periodic solutions of the simple second order wave equation,

(4)

Finally, he explained that in this approach the discreteness of quantum theory

arose naturally from the imposition, as in classical vibration theory, of
boundary conditions, and without any need for a discrete quantisation
condition.
In a paper completed in March, SchrOdinger established a formal connec-
tion between his theory and matrix mechanics,3 obtained by identifying the
matrix P with the operator -2 h. oloq; a general function F(p, q) led to an
71"1
operator [F), and he associated this operator with a matrix Fkl,

(5) f
Fkl = p(x) Uk (x) [F) ut(x) dx,

where the functions Ui (x) y p(x) constituted a complete normalised orthogo-

nal system in configuration space. In this paper SchrOdinger also pursued the
physical interpretation of his theory, and suggested that IR (l/Jol/J* lot) might
correspond to the charge density, the idea being to derive the intensities of
emitted radiation (Heisenberg's transition amplitudes) from the wave equa-
tion, and this from a modified electromagnetic theory.
 WAVE MECHANICS AND THE INTERPRETATION PROBLEM 85

In two further papers completed in May and June SchrOdinger pursued

the application of the wave mechanics, including the time-dependent case,
and in the latter paper he returned to the question of the interpretation of I/J.
He now identified the charge density as p =el/Jl/J* and explained that4
'" '" * is a kind of weigh t {unction in the configuration space of the system. The wave-
mechanical coniJgUIation of the system is a superposition of several - strictly all -
kinematically possible point-mechanical configurations. If you like paradoxes, you can
say that the system is, as it were, simultaneously in all kinematically conceivable posi-
tions, but not 'equally strongly' in all of them. In the case of macroscopic motions,
the weight function contracts in practice to a small region of practically indistinguishable
positions the centre of gravity of which in the configuration space covers macroscopically
detectable distances.

In this "reinterpretation" ,5 the fundamental conceptual role of the I/J them-

selves as the ultimate physical reality was explicitly played down, and the
wave motions were reinterpreted in terms of possible particle motions. But,
reflecting an ambiguity already present in de Broglie's work,6 SchrOdinger
nevertheless held fast to the priority of the wave over the particle picture. 7
The reception of SchrOdinger's wave mechanics was mixed. Many physiCists
who, like Einstein or Wien, had shared SchrOdinger's repugnance of the
physically incomprehensible "transcendental algebra" of matrix mechanics,
were thoroughly delighted. 8 But even Einstein and Lorentz soon became
critical of SchrOdinger's interpretation as practically untenable,9 and among
the phYSicists who had been responsible for matrix mechanics the reaction
was more immediately and more strongly ambivalent. Even before
SchrOdinger had drawn the explicit connection between the two theories it
had been clear to all concerned that they must be closely related. SchrOdinger's
replacement of the classical electron particle motion in the atom by the
motion of a complex wave form in configuration space constituted a clear
and explicit rejection of classical kinematics parallel to that in matrix
mechanics. The necessary complexity of his wave form corresponded to that
of the matrix-mechanical oscillators as noted by Born and Wiener. And the
preliminary results of the theories were the same. Moreover the formal con-
nection between the theories was implicit in the Born-Wiener operator
formulation of matrix mechanics, for once the identification p - a/aq had
been made, then the ordinary Hamiltonian equation of motion,

H=L..+ V(q)=E
2m
if viewed as an operator equation, led directly to SchrOdinger's wave equation.
86 CHAPTER 7

The identification itself could be inferred by analogy from the results H - a/at,
pq - qp - 1, and Ht - tH "'" 1, the latter not explicit in the Born-Wiener
paper but easily derivable from the formula for time differentiation first
established by Born and Jordan,
i=Hg-gH.
Born later cursed himself for having missed the p . . . . a/aq identification, claim-
ing that it would have led him to the wave mechanics before SchrOdinger. 10
But this seems unlikely in view of the conceptual gulf separating the two
approaches, and there is nothing whatsoever to suggest that Born even con-
nected the two theories in this way before SchrOdinger's own demonstration
of the connection. Early in April, before SchrOdinger's demonstration had
been published, Pauli independently produced a fuller and more careful
analysis of this connection. l l But although he had adopted the operator
formulation in January, and had then derived explicitly the crucial energy-
time commutation rule, he appears, like Schrodinger, to have begun his
analysis from the wave mechanics end. 12
There could be little doubt in the minds of the matrix-mechanical physicists
that the wave mechanics was both relevant and, since it attained the generality
of the Born-Wiener formulation without recourse to abstract mathematics
and through the portrayal of the physical problem in a classically familiar
and apparently visualisable form, important. Pauli rated Schrodinger's work
as "among the most important ... of recent times"13 and Heisenberg wrote
that "Schro :!L'lger's mathematics clearly signifies a great advance."14 From a
conceptual viewpoint, however, they were less impressed. Pauli was troubled
by the conviction that had dominated his physical thinking since he was a
student, that a pure continuum field theory of physics was impossible on
operational grounds. 1s Writing to SchrOdinger in May, he first praised the
latter's theory, but then added that "I have generally the strongest doubt in
the feasibility of a consistent wholly continuous field theory of the de
Broglie rays. One must probably still introduce into the description of quantum
phenomena essentially discontinuous elements as well. "16 Writing again the
following November he expressed a "real conviction that the quantum
phenomena in nature show facets that cannot be covered by the concepts of
continuum (field) physics alone",17 and in December he insisted yet again
that "quantum phenomena can never be explained" - whatever success
Schrodinger might achieve with his formulation - "in terms of continuum
physics. "18
Concerned with more immediate problems, Heisenberg had meanwhile
 WAVE MECHANICS AND THE INTERPRET A TION PROBLEM 87

written to Dirac in May that "I quite agree with your criticism of SchrOdinger's
theory with regard to a wave theory of matter. This theory must be incon-
sistent, just like the wave theory of light."19 And while Dirac does not seem
to have been too incensed by Schrodinger's interpretative claims - which he
later dismissed as "metaphysics" - Heisenberg was. As a mathematical
generalisation of matrix mechanics Schrodinger's theory was fine, but other-
wise it was a misleading delusion. It did not lead to a consistent wave mechanics
in de Broglie's sense,2° and its treatment of the wave function 1J;, complex and
defined in multi-dimensional configuration space, as somehow constituting a
visualisation [Anschaulichkeit] or classical physical representation was simply
"rubbish".21 Mocking the idea of a "rotating electron, its charge distributed
over the whole of space with axes in a fourth and fifth dimension", he wrote
to Pauli in June that he found this part of the theory the more detestable the
more he thought about it. 22 Upon meeting Schrodinger personally in July,
Heisenberg came to the conclusion that any element of deception was quite
unconscious, and that Schrodinger was simply a nice chap twenty-six years out
of date,23 and this opinion seems to have been widely shared. Sommerfeld, at
whose institute the meeting had taken place, drew the "general conclusion
that the wave mechanics is truly an admirably worthy micro-mechanics, but
that the fund~mental quantum riddles are not remotely solved by it."24
Some months later, after Schrodinger had lectured also in Copenhagen, Bohr
wrote to Kronig that Schrodinger seemed to think that he had ridded physics
of the quantum hypothesis altogether, but that "this appears, however, to be
a misunderstanding, as it would seem that Schrodinger's results so far can
only be given a physical application when interpreted in the sense of the usual
postulates."25 Jordan, whose "slightly tactless" behaviour had to be followed
by apologies from Bom,26 wrote to SchrOdinger that "all of the quantum-
mechanical people known to me are convinced that the basic concepts of
Bohr are still to be retained."27

BORN'S STATISTICAL WAVE MECHANICS

Born himself wrote to SchrOdinger in 1927 that, although he had "meanwhile

returned again to Heisenberg's standpoint", he had originally disagreed with
Heisenberg and had thought that "your wave mechanics signified more
physically than our quantum mechanics. "28 He could not follow Schrodinger's
interpretation of the theory but he was, he wrote, convinced of its superiority
to matrix mechanics by its simple handling of aperiodic processes, and he had
therefore adopted it for his own researches. 29
88 CHAPTER 7

After visiting MIT, Born had spent January through March 1926 on a
lecture tour of the United States, and on his return to Germany he had gone
straight to Frankfurt where his wife was convalescing after an illness. By the
time he got down to work again in Gottingen the first installments of
Schrodinger's wave mechanics had already been published and the connec-
tion with matrix mechanics derived. Born was naturally attracted to the new
theory. It had been developed from the wave theory of matter to which he
had himself been attracted the previous year, and it was coextensive with the
theory he had just formulated with Wiener. But it was physically more
suggestive than that theory (albeit misleadingly so), and was expressed in
terms of familiar classical mathematical physics. As he wrote later in the year,
wave mechanics also had the advantages of permitting "the retention of the
conventional ideas of space and time in which events take place in a completely
normal manner",30 a statement with which Heisenberg would not have
agreed, and of offering a natural origin for the quantum behaviour: the
commutation relations arose naturally as operator equations in this theory,
but had to be imposed as axiomatic upon matrix mechanics.31 But it seems
to have been the familiar classical form and wide applicability that appealed
most immediately. In his first paper based upon wave mechanics, completed
in June 1926, Born considered its application to collision processes, a subject
with which he had been concerned before the arrival of Heisenberg's new
kinematiCS, which combined the main problems - aperiodic effects and
transition processes - with which quantum theory was faced, and which he
had presumably intended already to attack on the basis of his own operator
formulation. 32 He concluded that 33

Of all the different forms of this theory only Schrodinger's has proved suitable here, and
I may directly on these grounds take it as the most profound comprehension of the
quantum problem

Applying wave mechanics through a first order perturbation theory of the

collision between an atom and a free electron, Born obtained an asymptotic
solution at infinity ('Born approximation'),

(6) I/I~T (x, y, z; qk) =~ II dw CPnm (a, 13, 'Y)

electron atom m ax+/3Y+-yz>O T

• sin {k nm (ax + l3y + 'YZ + o)} I/I~ (qk),

where I/I~ (qk) was the mth eigenstate of the unperturbed atom, CPnm (a,l3, 'Y)
T
 WAVE MECHANICS AND THE INTERPRETATION PROBLEM 89

the wave function for the electron (originally incident from the z-direction
with energy T), and

with vg m the transition frequencies of the unperturbed atom, the solution

for the final energy of the electron.
The problem was how to interpret this physically. Born claimed that 34

If one is to reinterpret this result in terms of particles, only one interpretation is possible:
cl»nm [corrected in proof to Icl»nm 12 I indicates the probability that an electron incident
frdm the z-direction will have b6en sent in the a, (3, 'Y direction (and with phase 6), its
energy T having been increased in the process by a quantum hv~m' at the expense of the
atomic energy ....
Schrodinger's quantum mechanics thus gives a complete answer to the question as to
the effect of a collision, but there is no question of a causal relationship. One cannot
answer the question "what is the state after the collision" but only the question ''what is
the probability of a given effect of the collision" (in which quantum-mechanical energy
levels must naturally be preserved).
Here the whole problem of determinism presents itself. From the standpoint of our
quantum mechanics there is no quantity that remains causal in the case of an individual
collision effect; but also in practice we have no grounds to believe that there are inner
eigenstates of the atom which stipulate a determined collision path. Should we hope to
discover such eigenstates later (such as phases of internal atomic motions), and to
determine them for the individual case? Or should we believe that the agreement of
theory and experiment on the impossibility of giving a stipulation of the causal lapse is a
preestablished harmony, resting on the non-existence of such stipulations. My own
inclination is that determinism is abandoned in the atomic world. But that is a phil-
osophical question, for the physical arguments are not conclusive.

Coming from someone who had shown no previous inclination toward such
a philosophical stance (but had on the contrary been viewed as a staunch
determinist), and in a context where his generally more philosophically-
minded colleagues were asking questions as to the meaning of space and
time that threatened to render the concept of causality meaningless, Born's
declaration was remarkable. More remarkable still was his subsequent
insistence upon acausality. In a second paper, completed in July, he returned
to the possibility of hidden phases providing a causal description of individual
events and argued that 3S
It appears to me a priori improbable that quantities corresponding to these phases can
easily be introduced into the new theory, but Mr. Frenkel has told me that this may
perhaps be the case. However this may be, this possibility would not alter anything
90 CHAPTER 7

relating to the practical indeterminacy of collision processes, since it is in fact impossible

to give the values of the phases; it must in fact lead to the same formulae as the 'phase-
less' theory proposed here.

In a subsequent short note to Nature, Born again stressed that for practical
purposes microscopic coordinates did not exist. 36 Classical theory, he claimed,
introduced such coordinates (for example those relating to the motions of
individual molecules) only to ignore them and take their statistical aggregate.
Quantum theory, on the other hand, did not bother with this charade: one
could not dismiss the possibility of microscopic coordinates existing, but they
were of no significance unless one could measure them, which one could not.
We shall return to these arguments. But first we must look more closely at
the statistical interpretation itself. In his second paper Born noted that
whereas matrix mechanics "started from the idea that an exact representa-
tion of the processes in space and time is quite impossible and is therefore
statisfied with the establishment of relations between observable quantities" ,
SchrOdinger attempted to assign to his wave process "a reality of the same
type as light waves possess".37 Neither conception appeared to him satis-
factory, however, and he preferred "to adhere to an observation of Einstein,
. . . that the [light] waves are present only to show the corpuscular light
quanta the way".38 He interpreted the Schrodinger waves in this way as a
"ghost field" (which term he attributed to Einstein) "or, better, guiding field",
the amplitude corresponding to a (determined) probability as in the first
paper. 39 As might be expected, Einstein in fact rejected Born's acausal
interpretation,40 but Born was clearly upset by this rejection and seems
genuinely to have assumed that his and Einstein's interpretations were the
same. 41 How, then, did this confusion arise, and how did Born reach his own
interpretation?
The first stated requirement of Born's interpretation was that it should
retain the corpuscular nature of the electron, and in his recollections he
linked this with Franck's experiments. 42 The most recent of these experi-
ments had in fact demonstrated the wave nature of the electron, and Born
himself had been strongly attracted to de Broglie's wave theory of matter. 43
But there could be no doubt in his mind that beyond collision phenomena
electrons (as also light) regained a particulate nature, and it must have already
been clear from his analysis of collision effects that the Schrodinger wave
packets would not remain sufficiently localised for this to be possible. The
particles could therefore not be seen simply as constructed from wave packets,
the interpretation given them by Schrodinger. Nor, in view of their complex
 WAVE MECHANICS AND THE INTERPRETATION PROBLEM 91

and multi-dimensional nature, as well as of experiments conducted with single

quanta or electrons in the apparatus at any time, could the wave be treated
simply as the statistical result of many individual particles. 44 Given his rejec-
tion of these possibilities, Born was virtually forced by his acceptance of the
physical significance of the Schrodinger waves into the interpretation he
adopted. The waves had to be seen as somehow guiding the motion of particles,
and the wave amplitudes had therefore to be identified with probabilities in
the way that he said. But what was not forced, and what Einstein disagreed
with, was the abandonment of any hope that the guidance of the particles by
the waves might ultimately be causally determined. The idea of a guiding
field had indeed been mooted by Einstein in the course of his efforts to
understand the wave-particle duality in the early part of the decade,45 and his
failure to construct a causal model on these lines may well have been a factor
in the subsequent rejection of causality by Schottky and others. 46 But
Einstein had been concerned only with real waves in 3-dimensional space, and
he himself had clung to the hope of a causal theory - in fact a causal field
theory with the corpuscular singularities arising from boundary restrictions
much as Schrodinger had suggested. 47 At the time, Born appears to have
shared this hope, but he now rejected it with such conviction that he expected
Einstein to follow him. Why?
There are possibilities for many influences here, including those ofWeyl-
who, having taken an active part in both matrix mechanics and wave
mechanics,48 may reasonably be supposed to have been in communication
with Born - and, in view of Born's emphasis on the fundamental "fusion of
mechanics and statistics" in his theory ,49 of Spengler or Reichenbach. These
influences, which have been considered elsewhere, may well have played a
part in Born's thinking. so But there were also internal factors operating.
Firstly there was Heisenberg's earlier suggestion that the quantum-mechanical
formalism might necessarily offer only limited information. 51 Secondly, there
was the fact that the theory appeared in some way to generate statistics.
Quantum theory had long involved an element of uncertainty, in the location
of an orbit, the moment of transition, etc., but this had always been a case of
uncertain conclusions following from uncertain data. In Born's analysis of the
electron-atom collision, however, uncertain conclusions (the atom in a
superposition of states, the electron spread throughout space) appeared to
follow from definite data on the initial motion of the electron and state of
the atom. Finally, Born's denial of the existence of further observable
microscopic coordinates raises a particularly interesting consideration. For
the most striking features of his presentation were the identification of utility
92 CHAPTER 7

and profundity, the confident assertion that the theory was final and immune
to further experimental advance, and the related assertion that it was com-
plete: 52

Many take it that the problem of the transitions of quantum mechanics ... cannot be
comprehended, but that new concepts will be needed. I myself came, through the
impression of the completeness of the logical foundations of quantum mechanics, to the
opinion that the theory is complete and that the transition problem must be contained
in it.

Born's insistence upon acausality, resting as it did on the non-existence and

non-observability of further microscopic coordinates, was equivalent to his
insistence that his theory was final, and the above remarks tie in this insistence
with that of the superiority of a mathematical to a physical approach, and
thus with the divisions that had arisen on this score the previous Autumn. In
concentrating on the practical applicability and mathematical formulation of
the theory (his second paper incorporated some of the mathematics worked
out with Wiener), Born had arrived at a theory the predictive range of which
coincided perfectly with that of existing experimental observations. If this
coincidence was taken as final - but only in this case - then his theory could
be taken independent of any underlying epistemological considerations, and
would constitute a full justification of his formal approach.

REACTIONS TO BORN'S THEOR Y

Born was, as Heisenberg noted, 53 a "mathematical methods man". His basic

approach ,to physics was to take a physically clear problem and to seek the
mathematically rigorous solution with the widest applicability, not to concern
himself with the philosophical subtleties of problem definition. In his view a
useful formulation of a problem was indeed a profound one, and he must
have considered acausality a small price to pay for the combination of
physical clarity and wide applicability in SchrMinger's theory. But to
Heisenberg and Pauli, who felt that through a reexamination of the kinematical
concepts they were at last getting to the core of the fundamental problems of
quantum theory, Born's attitude seemed retrogressive. His theory contained
distinct advances. His probabilistic interpretation of the wave function was
clearly an important extension of the concept of transition probabilities, and
combined with the wide-reaching mathematical theory it allowed the predic-
tion of non-classical electron collision phenomena (barrier penetration, etc.)
previously outside the scope of the new quantum mechanics. But of the
 WAVE MECHANICS AND THE INTERPRETATION PROBLEM 93

observability criterion and new kinematics there was no longer a trace. They
were indeed explicitly rejected and in Pauli's eyes at least Born's theory must
have shared the same faults as had Bohr's virtual oscillator theory. The need
for a redefinition of the fundamental concepts was again shelved for the sake
of utility, and the failure of the existing kinematics once again expressed
through a concept of acausaIity that could only be defined in terms of those
kinematics. 54 Pauli himself was apparently too interested in the advances of
the new theory, or too busy with his monumental encyclopedia article on
quantum theory, to waste words in criticism,55 but Heisenberg was less
reticent. He recalled that he was very angry with the relapse into pseudo-
classical terminology, 56 and writing to Pauli in July he mocked Born's use of
the wave terminology, likening his statement of the wave theory to that of
the apostolic creed. 57 But he too had to recognise the advances of the theory.
The problem now to be faced was not that of a choice between matrix
mechanics and wave mechanics, but rather that of bringing together the
conceptual innovations of the one with the practical achievements of the
other and with the variety of additional ideas to which matrix mechanics had
already led.

QUANTUM MECHANICS IN THE SUMMER OF 1926

By the Summer of 1926, the innovations of Heisenberg, Pauli and Schrodinger

had already led to a wide range of theories, results and ideas, and many of
these had to be brought together before a single fully-fledged 'quantum
mechanics' could be evolved. Among developments of particular importance
we have already noted the reflections of Heisenberg and Pauli on kinematics
and the thoughts of Pauli on observability, the development of matrix
mechanics as a theory of transformations, and Born's probabilistic interpreta-
tion of Schrodinger's wave function. There had also been other developments
due to Lanczos, who had given a continuous integral equation formulation of
matrix mechanics, 58 to Fermi, and especially to Dirac.
Working independently, Dirac had derived from Heisenberg's new kine-
matics many of the results of matrix mechanics. But he had done so through
a particularly elegant, and conceptually clear, formulation. His approach had
been to take the notion of "quantum-mechanical quantities", as defined by
Heisenberg, and to develop a consistent calculus of them. This had led him
in late 1925, through a comparison with the classical theory of action and
angle variables, to the conclusion that the quantum-mechanical equations
could be derived from the classical by substituting the quantum quantities
94 CHAPTER 7

[x, y] = (xy - yx) 21f/ih for the classical commutators, (x, y) = { :~ ~~­
ay ax }. For classical canonical variables,p, q, this gave him the commuta-
awaJ
tion relations, frequency condition, and indeed all the formulae of matrix
mechanics. 59 In a second paper, completed in January 1926, he had clarified
the relationship between the classical and quantum theories by introducing
the concepts of "q-numbers" (quantum-mechanical quantities obeying
quantum-mechanical rules) and "c-numbers" (ordinary classical numbers).6o
He admitted that "at present we can form no picture of what a q-number is
like" ,61 but the terminology helped him toward a clear expression of the
physical problem within the new theory: having solved the equations of
motion in their quantum-mechanical form, one could only get results com-
parable with experiment if one could represent the q-numbers by c-numbers
corresponding to experimentally observable values.
Dirac's formulation had the great virtue of preserving the significance of
the observability criterion and new kinematics. And in August 1926 he
pursued the former concept into the field of quantum statistics that had so
intrigued Jordan and so confused Heisenberg. He argued that 62

Heisenberg's theory ... enables one to calculate just those quantities that are of physical
importance, and gives no information about quantities such as orbital frequencies that
one can never hope to measure experimentally. We should expect this very satisfying
characteristic to persist in all future developments of the theory.

The criterion of observability became relevant when, for an atom of several

electrons, "the positions of two of the electrons are interchanged" , and "the
new state of the atom is physically indistinguishable from the original one."63
Labelling the two states (m, n) and (n, m), Dirac argued that one could not
observe the individual transition intensities from a third state, (m' , n') ~ (m, n)
and (m', n') ~ (n, m), but only the sum of the intensities of the two transi-
tions. The two states (m, n) and(n, m)thereforehad to be treated in quantum
mechanics as interchangeable, and this requirement led to two possible
solutions for the whole system,

(7) I/Imn -I/Im (l) I/In (2) ± I/I n (l) I/I m (2).

One solution corresponded to I/Im m - 0, or Pauli's exclusion principle. The

other gave I/I mn = I/Inm, corresponding to (m, n) and (n, m) referring to one
and the same state: extended to many particles this gave the Bose-Einstein
statistics. Fermi had demonstrated earlier in the year that Pauli's exclusion
 WAVE MECHANICS AND THE INTERPRETATION PROBLEM 9S

principle led directly to a different set of statistics,64 and Dirac now derived
the same conclusion independently, showing that if the Bose-Einstein statistics
were expressed in the form

(8) Ns =As/(eOt eftEs - 1),

then those following from the exclusion principle gave

(9) Ns =As/(eOt eftEs + 1).

He noted that the theory did not as yet allow one to determine which set of
statistics applied to the ideal gas, but suggested that this was more likely to
behave like electrons (as Fermi had in fact assumed) than like light-quanta.
In the same paper Dirac also drew on the matrix formulation of quantum
mechanics to generalise SchrOdinger's theory. SchrOdinger had restricted
himself to the classical problem of finding eigenfunctions I/In corresponding
to energy eigenvalues En. Dirac now pointed out that in matrix mechanics
there was nothing special about energy, and that one could equally well ask
for eigenfunctions 'ltn corresponding to eigenvalues of any function of the
space-time and momentum-energy coordinates. 65

PAULI'S IDEAS ON QUANTUM MECHANICS

Dirac's generalisation of wave mechanics was a crucially important first step

in the fusion of the various aspects of quantum mechanics into a coherent
whole. This fusion was largely completed in the Autumn and Winter of 1926-
27, during which time Dirac was in Copenhagen with Heisenberg, whose ideas
on quantum statistics he quickly put straight, and Bohr. It was catalysed by a
letter from Pauli to Heisenberg written on 19 October, in which Pauli, stimu-
lated by Dirac's considerations, tried to bring together some of the work of
recent months under the two headings of quantum statistics and collision
theory.66 Having thanked Heisenberg for a letter, no longer extant, Pauli
remarked that he had been thinking about the Fermi-Dirac statistics, and he
raised the question of zero-point energy, the existence of which Jordan had
taken the previous year to be one of the most fundamental features of
quantum theory ,67 and of which of the two sets of quantum statistics should
be applied to the ideal gas: 68

So far there exists a distinction between a crystal lattice and radiation, that, namely, in
t
respect of the zero-point energy hv. Now there are a priori reasons for supposing that a
96 CHAPTER 7

zero-point energy is also present in the ideal gas. [Pauli noted that he had discussed this
with Stem.) But the Einstein-Bose theory only allows it to be plugged on artificially
(c.f. the relevant statements of Schrodinger in Berliner Berichte and Physikali8che
Zeit8chrift), and that speaks from the fust against this theory and for Fermi-Dirac. To
make matters clearer for myself I have carried through the fluctuation considerations
of Einstein's work, Berlin Academy S8, §8, from the standpoint of the Fermi-Dirac
theory ....

Pauli continued to give a lengthy discussion of this problem and also to

consider the question, again raised by Jordan in the wake of an analysis by
Ehrenfest,69 of additive entropy. Neither discussion led to any really signifi-
cant new insight, but in a letter of 15 November Heisenberg, following his
instruction in quantum statistics by Dirac, extended the latter's analysis to
establish the usual characterisation of particle statistics. 7o Following up
Dirac's characterisation of the statistics as relating to symmetrical or anti-
symmetrical solutions he showed that an overall symmetry requirement led
to an association of the statistics with spin: half-integral spins gave anti-
symmetric space functions and Fermi-Dirac statistics, while integral spins
gave symmetric space functions and Einstein-Bose statistics.
The second part of Pauli's letter, concerning collision theory, was more
provocative and deserves extensive quotation. 71 What he had to say was, he
admitted, something of an "undigested dumpling", but it was certainly a very
rich dumpling. He began by considering a one-dimensional collision:

A mass point runs over an obstacle, characterised by a potential eigenfunction V(x),

falling off sharply to zero on both sides from a flXed point xo. The maximum of V is
fmite. Further, the incident energy E of the mass point is large compared with the
maximum value of V, so the unperturbed unifonn rectilinear motion must be taken as a
zeroth approximation, and successive perturbation theories can then be applied. Naturally
according to classical mechanics, if E > I Vmax I, the mass point always runs over the
obstacle. But according to Born's quantum mechanics ... it sometimes happens even in
the fust approximation that the mass point is reflected back, i.e. reverses its velocity
direction at the obstacle. And indeed there arises according to Born as a standard for the
probability of reflection the square of the amplitude of the wave for x < x o , represented

i:
by

(A) 1/1 1 =~ t e ikx e zikt V(O dt·

Pauli's next step was to treat a one-dimensional rotator perturbed by a force

field characterised by a potential V ( {} ) :

The case is periodic and can be treated by means of ordinary matrix mechanics.... The
essential thing now is that the system is degenerate, in that the given (quantised) energy
 WAVE MECHANICS AND THE INTERPRETATION PROBLEM 97

of the mass point, En, can rotate to the left (quantum number +n) as much as to the
right (quantum number -n). A time-constant quantity fwould have in its matrix repre-
sentation not only diagonal elements f+n+n' f-n- n but also elements of the form
f+n-n' f -n+n· According to Schrodinger's recipe, the matrix elements are now given in
the unperturbed rotation by some function F ({}), real and periodic in the angle {}:
F({})e in {} = :E~ FnmeimfJ
m=-~

or
F
nm
=~
21r
10
21r
F({}) ei(n-m) fJ d{}.

That the numerical values Fnm here depend only on the difference n - m arises from the
fact that the system is exactly force-free (everything goes the same in the three-dimen-
sional case).
The time average of the perturbation energy V({}) in the state ±n is a Hermitian
matrix of the form
V+n+n V+n-n
V-n+n V-n- n

wherein, according to (B),

1 (21r
V+n+n = V-n-n = 2- ),
1r 0
V(fJ) dfJ,

V+n-n * = -2
= V-n+n 1 121r V(fJ) e zinfJ dfJ.
1r 0

Putting V+n+n = Co, IV+n-n I = C,' V+n-n =C, e i6 , V-n+n =C, e- i6 , one gets the
'secular' equation for energy E,

IcCO, e--Ei6
E = Co ± C, , so the original energy values split into two. This results further in the fact
that in both states the perturbed rotator corresponds to standing waves:
l/J = cos (n fJ - 6/2) or sin (n {} - 6/2).

That is naturally your favourite unadulterated resonance phenomenon (released moreover

from the mess of equivalent electrons; here there is only one particle). The energy swings
to and fro between the left- and the right-turning oscillators. And in each of the 'secular'
quantised states of the perturbed rotator, the particle must rotate as often in the positive
as in the negative sense. But how is that possible? It is only possible, as Born has rightly
shown in his treatment (in defIance of all classical mechanical presentations) if the
particle is reflected back from the obstacle despite its greater kinetic energy E.

Once again, the same typically quantum behaviour was manifest. The next step,
98 CHAPTER 7

"and now comes the beauty of it", was to connect the wave-mechanical
formulation of the one case with the matrix-mechanical formulation of
the other:

The integral, which according to Born's formula (A) gives the probability of reflection,
is none other than the matrix element V+n -n , ... , when one goes to the limit of
an infinitely great radius of the rotator. 1V+n-n 12 is (to within the factor 11k) pro-
portional to the probability, where 2 1V+n _ n 1 is the energy splitting of the secular
disturbance.
Naturally this connection can also be extended to higher approximations. The
essential thing is this: one carries out the matrix perturbation calculation, but not
that for the secular perturbation, so that the energy is derived from a non-diagonal
time-constant matrix of the type

E+n-n)
E- n - n
Then IE+n -n 12 gives the collision probability.

As Pauli explained, the same recipe could also be generalised to the three-
dimensional case:

The mass point now runs through a central field and the potential energy V(r) falls
off quickly from the centre. Now each such space-function falling off quickly with
x, y, z can be represented classically in the unperturbed rectilinear motion as a time-
dependent Fourier integral - corresponding to a continuous matrix. One must only
decide which canonical variables to take for the unperturbed (rectilinear) motion.
The ps (analogous to the action variables 1) must be constant in time and the energy
must always be a function of the ps; the qs (analogous to the angle variables w) must
be constant or a linear function of time. When the system is degenerate they are not
unequivocally determined.
Example 1: the ps are the ordinary cartesian momentum components, the qs the
cartesian coordinates.
Example 2: the ps are E, angular momentum P, angular momentum Q parallel to z.
The qs are t, perihelion angle (3, node angle 'Y.
These are also defmeable for rectilinear motion.
Now comes the dim point. The ps must be taken as controlled, the qs as uncontrolled.
That is to say that one can always calculate only the probabilities of fixed variations of
the ps from given initial values and averaged over all possibe values of q ....
We remain now in classical coordinates. Each space-function F(x, y, z), which falls
of sharply from the centre, would then correspond (according to Schrodinger's proce-
dure for the calculation of matrices) to the continuous matrix
 WAVE MECHANICS AND THE INTERPRETATION PROBLEM 99

Then IV:Z:-r:~ 12 corresponds again to the probability of the deflection from Px ..•
to P~ ... , ? ., where the boundary condition of energy conservation, ~ pi
= 1: P~ 2
holds.
In respect of the higher approximations, these too presumably give no essential
difficulties. One can also proceed from the matrix corresponding to the acceleration
dp!dt and then observe that classically the Fourier coefficient corresponding to the
zero value of the frequency determines the deviation. But quantum-mechanically the
square of the Fourier coefficient of dp!dt that corresponds to zero energy change
determines (to within a harmless factor) the probability of the deflection. In inelastic
collisions it is in principle the same.

So much, wrote Pauli, for the mathematics. But what of the physics?

The physics of it still continues to be unclear to me. The first question is, why may only
the ps, and in any case not both the ps and also the qs be described with any accuracy.
This is the old question that occurred if the velocity direction and asymptotic distance
of the orbit from the core were given (at least with a certain accuracy). On this I know
nothing that I have not already long known. It is always the same: there is not on
account of the bending [of the orbit I any weak radiation in the wave optics of 1/1 fields,
and one may not at the same time relate both the 'p-numbers' and the 'q-numbers' to
ordinary 'c-numbers'. One can see the world with p-eyes and one can see it with q-eyes,
but if one opens both eyes together then one goes astray.
The second question is how the above matrix elements come to determine the
collision probabilities. The direction in which one must here steer, I believe to be the
following. The historical development has involved the connection of matrix elements
with the observation of accessible data being undertaken in a roundabout way, through
the emitted radiation. But I am now convinced with the whole fervour of my heart that
the matrix elements must be connected with in principle observable kinematic (perhaps
statistical) data of the particles concerned in the stationary states. This is quite apart
from whether anything in general and what (electromagnetically) is radiated (the
velocity of light may be set to 00). Also I do not doubt that behind it is hidden the key
to the treatment of unperiodic motions. Now we have that all diagonal elements of the
matrices (at least of functions of p alone or of q alone) can already generally be inter-
preted kinematically. So one can indeed ask first for the probability that in a fixed
stationary state of the system the coordinates qk of a particle lie between qk and qk +
dqk. The answer is then II/I(ql ... q~ 12 dql ... dqf with 1/1 the Schrodinger eigen-
function. (From the corpuscular standpoint it thus already makes sense for it to lie in
multi-dimensional space). We must look at this probability as in principle observable,
just as the light intensity as a space-function in standing light waves. It is thus clear that
the diagonal elements of the matrix of each q-function must be

physically interpreted as the 'mean value of F in the nth state'. Here one can make
100 CHAPTER 7

a mathematical quip: there is also a corresponding probability density in p-space: for

this one puts (formulated in one dimension for simplicity)

Pik =f P I/>i (P) I/>k (P) dp

21ri
y q i k -- - f I/>i-dp-+
al/>k - f -a-I/>k
al/>i * dp , ...
ap p

You see that I have switched to the opposite of the usual prescription for the construc-
tion of the matrix elements Pik and qik from the eigenfunction. From the matrix equa-
tion of energy conservation,
2
~m +V(q)=E,

one gets

fL
L2m + V (- ~~\l</>=EI/>;
2m apJj

V is thought of as an operator, say in powers of a/ap. In the harmonic oscillator, where

the Hamiltonian is symmetric in p and q, I/> is also Hermite's polynomial. One can also
operate a perturbation theory with the 1/>.
Similarly in the Hydrogen atom, I/> must be a simple function, but I have not yet
worked it out. In any case, there is thus also a probability that in the nth state Pk lies
betweenPk andpk + dpk' and it is given by II/>n (P • .. . Pt) 12 dp • .. , dpf' so that

Fnn(P) =f F(P) II/>n (Pk) 12 dp, ... dpf"

When, accordingly, the diagonal elements of the matrices follow physically from the
kinematic statements contained in the I/>k(P., ... ,Pt) and "'k(q., ... , qt), I do not
believe that your fluctuation treatment can say anything new and important for the
intepretation of unperiodic motions beyond this. For there it is always a question of
time averages. In the collision phenomena it is likewise a question of time-constant
elements, but of non-diagonal elements of the matrix in degenerate systems (the physical
significance of which is still not clear to me).
But my chief question is what the other matrix elements mean purely point-kine-
matically, and wholly unconnected with electromagnetic radiation. I have proposed a
statement, in which is asked: I know that at time t the particle has position coordinates
qo. What is then the probability that it has the coordinate q at time t + T? But I am no
longer certain whether that is a reasonable question. I think in any case in kinematics
of statistical data, which convey the time development of the behaviour of particles.
These data could throughout be of such a kind that one cannot speak of a definite
'path' of the particles. Also one again hits here on the fact that, as elsewhere, one may
not ask after P and q together.

This second part of Pauli's letter does not have the coherence of a published
paper, is not easily split up, and needs reading several times. It seems to have
 WAVE MECHANICS AND THE INTERPRETATION PROBLEM 101

been stimulated, as was the first part, by the work of Dirac. Pauli also drew
on the thoughts of Kramers,72 who had been developing what later became
known as the Wentzel-Kramers-Brillouin approximation in the context of
Schrodinger's theory, and he referred to some work of Heisenberg's on
fluctuations. But when Heisenberg published this work he himself drew
heavily on Pauli's thoughts,73 and these would appear to have been largely
original. They were also still somewhat confused, but Pauli had nevertheless
been able to make distinct progress toward a fusion of Born's theory with
the results - and problems - of matrix mechanics. Following Dirac's com-
ments on the extension of Born's theory to include predictions of quantities
other than energy, Pauli had shown, among other things, that this theory
could be expressed in momentum space as well as in configuration space.
He had suggested how the theory might lead to probability predictions of
p, q, and functions of these, and he had stressed the identification, previously
assumed by Jordan, of the diagonal elements of a matrix with the time mean
of the quantity represented. He had stressed repeatedly the impossibility
of simultaneous predictions of p and q and had linked this with the neces-
sarily statistical nature of classical kinematic pictures. All this constituted
substantial progress.
 CHAPTER 8

TRANSFORMATION THEORY AND THE DEVELOPMENT

OF THE PROBABILISTIC INTERPRETATION

TRANSFORMATION THEORY

On 28 October, Heisenberg replied to Pauli with profuse thanks. 1 The letter

had been handed round to Bohr, Dirac and Hund, and had been generally
discussed. Above all, Heisenberg wrote, he had been inspired by the dis-
cussion of collision processes, and he now understood much better the
significance of Born's formulation. In particular, Pauli's discussion of the
rotator indicated generally that "wherever in classical mechanics one type
of motion changes discontinuously into another, quantum mechanics supplies
a continuous transition which, so far as it may be thought of graphically,
signifies a probability dictum".2 In the wake of SchrOdinger's controversial
lecture, 3 and of the arrival of Dirac, attention at Copenhagen was focussed
very much on the problem of relating Born's formulation to matrix mechan-
ics, and on that of demonstrating that Schrodinger's theory could not be
continuously interpreted but must share the essential discreteness of matrix
mechanics. Pauli's letter clearly contributed much to both problems, and on
4 November Heisenberg wrote to him again, declaring himself "more and
more inspired by the content of your last letter every time I reflect on it."4
Two days later, he submitted a paper on the energy fluctuations of a gas,
in which both problems were brought together. 5 Drawing on his earlier
application of matrix mechanics to many-body problems, Heisenberg now
looked at the energy interchange between two particles and showed that
for observable quantities (such as the time means of energy and of the energy
fluctuation squared) the new quantum mechanics agreed with the discrete
conclusions of the light-quantum theory. The solutions for a system of two
particles were derived, on the new theory, through a matrix transformation
S of the solutions to the 'unperturbed' problem of! independent particles.
In place of the continuous oscillations of classical theory this led to the
states of interacting atoms being superpositions of their unperturbed states,
which could only be interpreted probabilistically. The discrete states were
thus maintained, the atoms oscillating between them as in Pauli's discussion
of the rotator and spending a determinable proportion of time in each.
Following Pauli's recipe, the probability that a system initially in state a
102
TRANSFORMATION THEORY/PROBABILISTIC INTERPRETATION 103

should be found in a new state (j was given by the appropriate element of the
transformation matrix. The time means of observable quantities (and thus the
proportion of time for which they took each allowed value) followed as the
diagonal sums of their respective matrices.
Heisenberg's main conclusion, as stressed in his paper, was that "a con-
tinuous interpretation of the quantum-mechanical formalism, including that
of the de Broglie-SchrOdinger waves, would not correspond to the essence of
the known formal interaction."6 But of greater importance for the progress
of the theory were the identification of the transformation matrix with a
set of probabilities, which brought together Born's interpretation and the
transformation theory; the explicit identification of the diagonal sum of a
matrix with its time mean; and the application of this to the prediction of
the proportion of time for which a system occupied a given state. In addition
to these achievements, Heisenberg had written to Pauli in his letter of 4
November that he had realised that "in general, every scheme that satisfies
pq - qp = h/27ri is correct and physically useful, so one has a completely
free choice as to how to fulfil this equation, with matrices, operators or
anything else." 7 He had moreover recognised the wave function ¢ of Pauli's
p-space representation as the Laplace transform of Schrodinger's wave I/J in
q-space, and had concluded that "the problem of canonical transformations
in the wave representation is thereby as good as solved."8 The problem was
not yet actually solved, as Heisenberg, like Pauli, was still too confused about
the various possible formulations to make the necessary generalisation. But
this generalisation was provided a few weeks later by Dirac.
Dirac had already been looking, at the end of October 1926 at the matrix
interpretation of SchrOdinger's charge density 1/J1/J*.9 In early November he
appears to have been somewhat diverted by the appearance of Klein's exten-
sion of Kaluza's five-dimensional unification of general relativity theory and
Maxwellian electromagnetic theory.1O Klein was working in Copenhagen
under Bohr, and Dirac, who was interested in relativity theory as well as in
quantum theory, naturally explored the possibilities of his work. But by 23
November Heisenberg could write again to Pauli informing him that "Dirac
has managed an extremely broad generalisation of my fluctuation paper"
- which is to say of the ideas Heisenberg had developed through considera-
tion of Pauli's earlier letter: 11
The fundamental idea of Dirac is perhaps this: let p, q, be any canonical conjugate
quantities, !(P, q) a function of the same.
Question: what can one say about!physically, in quantum mechanics?
Answer: one chooses, for example, q as a c-number (e.g. q = 10). Then it is classically
104 CHAPTER 8

possible to calculate the function [(P, 10). Quantum-mechanically this cannot be done.
But one can specify how great the range of p is for which [lies between the c-numbers
[ and [+ df If one puts for p and q the variables E and t, with [= WZ, you see already
that the Dirac problem amounts to my fluctuation note. But one can also, for example,
choo se E and t and put [= q. So one can deduce the fraction of time for which q lies
between q and q + dq. The solutions of this problem are the SchrOdinger I/I(q) I/I*(q).
Dirac has generally succeeded in solving the above problem mathematically as well.
As the probability function there appears a matrix S, of which the indices in the general
problem (f. p, q) are Sf,q (in the special case E, t, Wa, then as for me, SW a, E). The
Sf,q are, as always, solutions of a principal axes transformation of a Hermitian matrix.
The principal axes transformation can be reduced to a differential equation ala Schro-
dinger. One therefore has the means to actually determine S quite generally (this last
point corresponds very closely with your idea of p-waves). In the special case E, t, q,
the matrix has indices q and E, so Sq,E is the Schrodinger function, Sq,E = I/IE(q).
The Schrodinger function is thus identical to the matrix S in Born's principal axes
transformation for the determination of the eigenvalue W. But the above-named physical
interpretation of S contains at the same time all the physical statements that may
actually be made at present: e.g. Born's collision processes, Jordan's canonical trans-
formations, etc. Dirac's work certainly has many points of contact with yours and
mine. But it is nevertheless very general, as it functions, for example, just as easily and
surely in continuous variations as in periodic ones. I hold Dirac's work to be an extra-
ordinary advance.

Dirac's theory was indeed a great advance, and it did indeed have much in
common with the ideas of Heisenberg and Pauli, from whose combined
considerations it seems to have arisen. Heisenberg and Pauli had long been
asking just what measurements the new quantum mechanics could predict,
and in their recent correspondence it had become clear that, as Dirac noted
explicitly, a joint specification of p and q, or of E and t, was impossible.
They had concluded that what could be predicted were probabilistic results,
and they had specified these results for a variety of individual cases: Pauli
had specified the probability distributions of functions in a given energy
state, Heisenberg had specified the probability distribution of energies result-
ing from a given perturbation, and both had linked the probability functions
with transition matrices. Dirac had always shared their interest in the obser-
vability criterion and the associated applicability of the theory, and all he
had to do now was to apply his much more abstract mathematical view of
the theory and generalise. The question he asked was how, "when one has
performed all the calculations with the q-numbers and obtained all the
matrices one wants, ... one is to get physical results from the theory, i.e.
how can one obtain c-numbers from the theory that one can compare with
experimental values." 12 Generalising the viewpoint of Pauli and Heisenberg
by treating p, q and E, t simply as pairs of canonical coordinates, he deduced
TRANSFORMATION THEOR Y{PROBABILISTIC INTERPRETATION 105

that the only questions that could be posed in quantum mechanics were of
the form:

Given ~r' what do we know about any variable g as a function of 71r, the canonical
conjugate of ~r?

To answer these questions one had to transform from one scheme of matrices
(e.g. that in which the ~r were determined, i.e. diagonal) to another scheme (in
which the required function was diagonal). From consideration of Lanczos's
theory Dirac had developed for himself a continuous matrix formulation in
which to work, and using this he developed the theory of transformations
required, deducing the existing formulations as special cases. Following the
recipe of Born, Pauli and Heisenberg, he identified the probability distri-
bution of g as a function of 11 for given ~ with the diagonal sum of the matrix
transforming from a scheme in which ~ were diagonal to one in which g were
diagonal.
A fortnight after Dirac's paper was completed, a similar treatment was
reached independently by Jordan, who had also been in communication
with Pauli, and who had also, like his colleagues, been searching for a fusion
of wave mechanics and matrix mechanics. 13 To base a generalised quantum
mechanics upon transformation theory was natural to Jordan, who had
been primarily responsible for the development of that theory in matrix
mechanics. Working under Born and generally favourable to a continuous
formulation of quantum theory - he had devoted the larger part of his
research to date on deriving the light-quantum phenomena from oscillator
theories 14 - he naturally approached the problem, however, from the wave
mechanical formulation rather than from the matrix mechanics. He thus
based his treatment on Pauli's form of the probabilistic interpretation rather
than Heisenberg's and on the transformation from one set of variables to
another (as discussed in Pauli's letter) rather than from one scheme to another
representing the same variables. He set himself the problem: 15

[If) in place of p, q, new variables, p. Q, may be introduced by a canonical transfor-

mation such that H(p. q) = ii(P. Q) •.• we wish to construct the new wave equation
withii.

If the old equation were {H~~~, Y) -W}</>(y) =0, and the new {Ht~!, x)
- w} I/I(x) =0, Jordan asked how the new function 1/1 was related to the
original function </>, and he based his answer of Pauli's suggestion that
106 CHAPTER 8

If cf>n(q) is normalised, then 1cf>n(q) 12 dq is the probability that when the system finds
itself in the state n, the coordinate q takes a value in the interval (q, q + dq). If q. (3,
are two Hermitian quantum-mechanical quantities, which we shall here take for con-
venience as both constantly varying, then there will always exist a function cf>(q 0 , (30)'
such that 1</>(q 0' (30) 12 dq gives the (relative) probability that for a given value (30 the
quantity q will take a value in the interval (q 0' q 0 + dq). The function </>(q. (3) of Pauli
denotes the probability amplitude.

From this suggestion, Jordan deduced that two postulates should be expected:
the function if>(q, 13) should be independent of the mechanical nature of the
system and dependent only on the kinematic relations between q and 13;
and the functions should combine as <I> (1/1 0 , 130) = J 1/1 (Qo, q) if> (q, 130) dq. He
noted that it was the probability amplitude if> (and not the probability, \ if> \2)
that followed the usual combination law for probabilities, and he related
this to the interference of the probability waves.
Guided by the above considerations, Jordan postulated that for any two
quantum-mechanical quantities, q, 13, standing in a determined kinematic
relationship with each other, there should exist a probability amplitude
if>(x, y) as ~d'ined by Pauli, together with a complementary amplitude I/I(x, y),
such that rf>(x, y) = I/I*(Y, x), et cetera. He postulated further that the ampli-
tudes should combine interferingly, in a way that led to the relationship
above. He then defined two variables as canonically conjugate if their prob-
ability amplitudes were given by p(x, y) = e- xy / €, from which he deduced
that for a given value of one variable, all values of the conjugate variable
were equally probable. He postulated that for any given variable a canonical
conjugate variable should always exist, and proceeded to develop the theory
of operators transforming from one set of canonical coordinates to another.
The Schrodinger and momentum space wave equations followed as special
cases, as did the other formulations so far developed.

THE FORMULATION OF QUANTUM MECHANICS

The theories of Dirac and Jordan were developed independently, but they
shared the same conceptual roots and were mathematically equivalent. On
24 December, Dirac wrote to Jordan, giving a detailed account of his own
theory, and apologising for once again having duplicated Jordan's results.
On the relationship between the two theories he wrote that 16

Dr Heisenberg has shown me the work you sent him, and as far as I can see it is equivalent
to my own work in all essential points. The way of obtaining the results may be rather
different though .... In your work I believe you consider transformations from one
TRANSFORMATION THEORY/PROBABILISTIC INTERPRETATION 107

set of dynamical variables to another, instead of a transformation from one scheme of

matrices representing the dynamical variables to another scheme representing the same
dynamical variables, which is the point of view adopted throughout my paper. The
mathematics would appear to be the same in the two cases however.

Despite their intimate connection, the two theories did in fact differ in
some respects, and in particular in respect of the relationship proposed
between formalism and interpretation, an aspect of quantum mechanics of
which there was as yet no clear understanding. Born had claimed that his
interpretation was a necessary consequence and integral part of the wave-
mechanical formalism, but this attitude does not appear to have been generally
shared, and was clearly rejected by both Heisenberg and Pauli as glossing over
the fundamental problems involved. But neither of these physicists was
clear as to what to substitute for it. Pauli was committed to a philosophy
according to which all terms in the theory should be operationally dermed,
but the radical conceptual revision that this would entail was clearly beyond
him, and in his letter of 28 October 1926 he concentrated on the lower
order requirement that all the terms be observationally (but if necessary
statistically) interpreted. Heisenberg too had concentrated on the limits of
applicability of existing concepts, which he had suggested might be funda-
mentally statistical, and he too had thus committed himself, though he would
clearly have liked to see the theory physically founded, to working from the
formalism toward the physical interpretation. Dirac seems to have viewed
the foundations of the theory as essentially mathematical; but he took
great care to kept formalism and interpretation distinct, and emphasised
that the probabilistic interpretation did not follow from the formalism, as
Born had suggested, but must rather follow from a separate association of
theoretical and physical terms that included probabilistic assumptions. In
his letter to Jordan he stressed that Heisenberg's form of the probability
interpretation followed "if all points in l1-space are equally probable (and
only when this is SO)";17 and in his published paper he again emphasised that,
as Pauli had in fact required,18

The notion of probabilities does not enter into the ultimate description of mechanical
processes; only when one is given some information that involves a probability (e.g.
that all points in 1j-space are equally probable for representing a system) can one deduce
results that involve probabilities.

Jordan, on the other hand, incorporated his statistical postulates into the
foundations of his theory, which was built up explicitly of functions satisfy-
ing these postulates. At first sight this had the great advantage that as in
108 CHAPTER 8

classical theory a physical interpretation of the symbols preceded their

mathematical theory, but in fact the physics and mathematics were hope-
lessly confused. The statistical postulates were first required in respect of
general mathematical functions, the physical interpretation of which was
only deduced later.
The problem of the relationship between mathematics and physics is of
course a philosophical one, and the issue could never be settled to everyone's
liking. Through the earlier debate on the relative importance of mathematics
and physics in matrix mechanics, and through a continuing struggle between
the operationally conditioned corpuscular approach of Pauli and Heisenberg
and the field-theoretical preferences of Jordan and Bohr,19 the old disagree-
ment of 1918-1919 between Weyl, Pauli and Einstein on the construction
of physical theories had remained very much alive and unresolved. 20 But a
generally acceptable and workable formulation of quantum mechanics was
badly needed, and since a full understanding of the observability criterion
and new kinematics was still wanting this had necessarily to take some sort
of axiomatic form. In the Spring of 1927 Hilbert, the doyen of the axiomatic
method in physics, recovered sufficiently from a severe illness of the previous
year to take up once again his early interest in the new quantum mechanics.
And with his assistants, Nordheim and von Neumann, he provided a clear
axiomatic formulation of transformation theory.21 In the interests of clarity
and in order to allow the maximum freedom for the development of the
formalism, they followed Dirac in keeping formalism and interpretation quite
distinct. But developing their theory primarily from Jordan's (Jordan was
of course in GoUingen) they based the interpretation upon a set of physical
axioms akin to Jordan's statistical postulates. For any two mechanical quan-
tities, F.(p, q), F 2 (p, q), it was required that there should exist a function
cp(x, y; F l , F 2 ) such that qxp* = w(x, y; F l , F 2 ) was the relative probability
that, given F2 = y, Fl was in the range (x, x + dx). The probabilities were
required to be independent of the mechanical system and coordinates, to
be reflexive, ¢(x, y; x, y) = fjJ(y, x; y, x), and to combine as fjJ(x, z; F l , F 3 ) =
fCP(x, y; F l , F 2 ) fjJ (y, z; F 2 , F 3 ) dy. To obtain a quantum mechanics, it
was postulated that the physically defined cp(x, y; q, F) should be associated
with the kernel of the canonical transformation: q ~ F(q).
A second difference between the theories of Dirac and Jordan, and one
that was not alleviated by the Hilbert-von Neumann-Nordheim exposition,
was in respect of the relationship between discrete and continuous formu-
lations. Jordan's theory, and in particular his statistical postulates, were
expressed in the continuous wave-mechanical formulation. Dirac's theory, on
TRANSFORMATION THEORY/PROBABILISTIC INTERPRETATION 109

the other hand, was based on the discrete formulation of matrix mechanics. At
first sight this difference appeared to be relatively unimportant, for Jordan
was able to derive the matrix mechanics, and Dirac the wave mechanics.
But Dirac's derivation rested upon the use of the delta function

ll(x - y) =0, x fy; rY + CIt ll(x - y) dx

Jy -CIt
= 1,
which, as he well knew, was not a true mathematical function at all. 22 Jordan,
and Hilbert, von Neumann and Nordheim, did not use the delta function as
such. But their derivations depended upon the existence of transformations
which could only in fact be completed using the delta function. 23 Jordan
did not explore this problem. Dirac recognised it but appears to have thought
it insignificant given what he saw as a natural eqUivalence between continuous
and discrete treatments. 24 Von Neumann, however, combined an insistence
on mathematical rigour with a liking for mathematical generalisation, and
in 1927-28 he quickly achieved both.2s
Rather than relating the continuous and discrete functions themselves
von Neumann considered the spaces on which these functions were defined.
He was then able to draw on a famous theorem by Fischer and Riesz, to the
effect that the space of all Lebesgue integrable complex valued functions I(q),
with finite norm N = f 1112 dq, was isomorphic to the space of all complex
valued sequences {q;}, with finite ~ Iq; 12.26 Defining a more general Hilbert
space as any space isomorphic to these he then reformulated the transforma-
tion theory in this more general space, and so overcame the delta function
problem.
At the same time, von Neumann also generalised Dirac's probability
interpretation to its definitive form.27 Given that the system was in a state
..p(q) he defined the probability that the measurement of a quantity corres-
ponding to the operator A should give a value between a and b as the norm
of the projection corresponding to the spectral resolution of A. If

A = L: A dP(A), 1= L: dP(A),

then the projection was defined as

p[a, b) =P(b)-p(a),
and the required probability was
110 CHAPTER 8

From this von Neumann could derive the earlier expressions of the probability
interpretation by appropriate choices of the space over which the functions
were defined. Making further use of the spectral resolutions of operators
he was also able to generalise the interpretation to give the probabilities that,
given that the values of quantities corresponding to operators Hi lay in the
ranges Ji, those corresponding to the operators Ai should lie in the ranges
/ .28
z·
 CHAPTER 9

THE UNCERTAINTY PRINCIPLE AND THE COPENHAGEN

INTERPRETA TION

THE UNCERTAINTY PRINCIPLE

The problem of founding quantum mechanics upon a new conceptual frame-

work, replacing that of classical kinematics, was never fully solved. But
concurrent with the development of transformation theory, and working
from the same set of conceptual insights, Heisenberg was able to make some
progress toward understanding the limits of applicability of the existing
kinematical concepts in the new theory. Much of Pauli's October letter had
been devoted to this latter problem, and in his reply of 28 October Heisenberg
took up some of Pauli's leads: 1

But the most interesting of your observations is naturally the so-called 'dim point'. I
should like to believe that your p-waves have just as great a physical reality as the q-
waves; only naturally not so great a practical significance. But I am very sympathetic
to the equivalence on principle of p and q. The equation pq ~ qp =hi thus corresponds
always in the wave presentation to the fact that it is impossible to speak of a monochro-
matic wave at a fixed point in time (or in a very short time interval). But if one makes
the line less sharp, the time interval less short, then that very truly does have a meaning.
[Pauli noted here that in the contrary case, given a short time interval, it was meaningless
to speak of a precise energy value.] Analogously, it is meaningless to talk of the position
of a particle of fixed velocity. But if one accepts a less accurate position and velocity,
that does indeed have a meaning. So one understands very well that it is macroscopically
meaningful to speak more approximately of the position and velocity of a body.

Confirming that his analysis so far drew on Pauli's ideas as well as his own,
Heisenberg added in parentheses at this pOint that "all this is throughout
naturally nothing new to you." Both Pauli and Heisenberg had recognised
the impossibility of a joint specification of a pair of canonical coordinates,
and in particular of p and q or E and t, and they had linked this with the
commutation relations for such coordinates. From this letter of Heisenberg's
it is clear that they had also recognised that the degree to which one could
specify one coordinate of a pair was inverse to the degree of specification of
the other coordinate, and that by specifying neither of them too precisely
one could get an approximate, macroscopic, joint determination. But
what did this mean? The previous year Heisenberg had speculated on the
111
112 CHAPTER 9

possibility of "coarse" determinations of space and time, and he now

suggested 2
that space and time are actually only statistical concepts, as, perhaps, are temperature,
pressure, etc. in a gas. I mean, that space-like and time-like concepts are meaningless for
one particle, and that they become more and more meaningful the more particles are
treated.

But this could only be the first step to a new kinematic understanding, and
Heisenberg admitted that although Pauli's comments had raised his hopes
somewhat he had not, despite frequent attempts, got any further in this
direction. He had also got no further than Pauli on the kinematical significance
of the off-diagonal matrix elements. 3
The next significant step in Heisenberg's thought appears to have arisen in
response to some considerations of Pauli's, now lost, on ferromagnetism. 4
Writing to Pauli on 15 November, Heisenberg approved the general tendency
of his work, but suggested that the application of the ideas was "incautious": 5
The general division of phase space into cells the volume of any of which is the quantity
h is certainly a correct principle. But, and now comes my objection, if you specify the
cell walls sharply, and can then determine how many particles are in each cell, can you
not then, through the choice of neighbouring cell walls, [md the number of atoms
in as small a cell as you like? ... I mean, is the choice of fixed cell walls physically
meaningful?

He continued, illustrating his argument with a diagram (see Figure 1), to

suggest that
2 2

1,2 -------+~r-------~+_------~rr-------

1,2 -~----f':7"I~~I----} .
I ,Z ------io,.L..,r-----r-f'-""--'~:.....t:.+__t_------

Fig. 1.
 UNCERTAINTY PRINCIPLE/COPENHAGEN INTERPRETATION 113

Perhaps it is so that one can only specify, for example, the ratio of the two cell walls,
alb, but not the position of a flXed cell wall.
One could in other words specify the shape of the cells, but not their precise
location, for the latter would allow the consideration of overlapping cells of
arbitrarily small volume, and would thus be equivalent to a precise joint
specification of P and q. Heisenberg continued:
The same objection also holds now for the E, t distribution. But for the special case of
fixed t, everything is again in order. I hold it to be very reasonable to study even this
special case more closely, out of which will perhaps come something of the kinematic
meaning of the matrices.
In conclusion, I am also of your opinion that at the end of the dim point will be a
very clear point. I mean: if once space-time is already somehow discontinuous, it is then
very satisfactory that it makes no sense to speak, for example, of the velocity x at a
fixed point x. For to defme velocity one always needs at least two points, which can lie
in a discontinuous relationship but not infmitely close.

Heisenberg'S "special case" is not in fact special in the way suggested, but
Pauli, who had noted as much upon Heisenberg's previous letter, would no
doubt have put him right on that. 6 What is important is that by taking the
problem of the limitations of the classical kinematical concepts back into the
foundations of the statistical theory of gases, on which Pauli was now work-
ing, Heisenberg had reached a new picture of this problem in which the
degree of possible joint specification of P and q was precisely defined in terms
of the cells in phase space. For an atom in a given cell, the joint specification
of p and q was subject to the restriction PI q I '" h, where PI, q I, corresponded
to the "gauge" of the measurements, or to the walls of the cell.
A second feature of Heisenberg's letter, and one that was to be of great
importance for the further development of his ideas was his association of the
problem of specifying P given q with the problem of measuring or defining P
in a discrete quantum theory. In his last paper, and in his current discussions
with Bohr, he had emphasised repeatedly the essential discreteness of quantum
theory, and he now saw that if this discreteness operated on the space-time
metric itself, then this would lead directly to the impossibility of a joint
position-momentum specification. 7 To defme a momentum or velocity
required two position measurements, which could classically be taken infinitely
close to each other; if in quantum theory they could not, then a precise
velocity measurement would leave an uncertainty as to where between the
two positions it applied. In the letter to Pauli reporting Dirac's work, a week
later, Heisenberg followed up this insight with a related but more general
reflection: 8
114 CHAPTER 9

I often reflect on the true meaning of this whole formal connection, but it is horribly
difficult to be clear about it. That the world might be .continuous I am more than ever
convinced is completely out of the question. But so long as it is discontinuous then all
the words that we apply to the description of an event have too many c-numbers. One
no longer knows what the words 'wave' or 'particle' mean.

As we shall see, this viewpoint was also to be important, both for Heisenberg's
ideas and also for Bohr's.9
At about the time he wrote this last letter, Heisenberg was also writing
a non-technical paper on quantum mechanics for the scientific journal
Die Naturwissenscha!ten, and in this he again stressed the discreteness of
the quantum-mechanical world and the need to free quantum mechanics
from the constraint of visualisability (A nschaulichkeit) , or visual pictures
(anschauliche Bilder) , if contradictions were to be avoided. In late November,
however, Heisenberg's thoughts were diverted by the development of the
Dirac-Jordan transformation theory, and he does not seem to have returned
to the kinematic problem until February 1927 when, he recalled, Bohr's
absence on a holiday gave him the opportunity to sort out his ideas in peace
and quiet. lO He may also have been stimulated in response to a new paper on
the interpretation of quantum mechanics by Jordan,u On 5 February he
wrote to Pauli that he was "again occupied all the time with the logical
foundations of the whole pq - qp swindle", 12 and criticised Jordan for
talking of such things as "the probability of an electron being at a determined
point" (as in the classically-oriented interpretation of Born), when "the
concept 'path of an electron' is not properly defined." Heisenberg had already
established the previous Autumn a rough theoretical uncertainty in terms of
the phase space cells, and with Pauli and Dirac he had always viewed the
impossibility of a joint theoretical specification of p and q as corresponding
to an operational problem of their joint observability. He had also begun to
relate the latter problem to a possible discreteness of space-time. But as yet,
despite all the indications of a perfect correspondence between theory and
observation, he had not investigated closely the problem of the measurement
of the kinematic properties of an electron, and had not made the connection
between theory and measurement a precise one. These he now did, and
writing to Pauli on 27 February he askedY

What does one understand by the words 'position of the electron'? This question can be
replaced according to the well-known model by the other: 'How does one determine the
position of the electron'? One takes perhaps a microscope with sufficiently good resolving
power and looks at the electron. The accuracy depends on the wavelength of the light.
For a sufficiently short wave-length of light the position of the electron at a fixed time
 UNCERTAINTY PRINCIPLE/COPENHAGEN INTERPRETATION 115

(and if necessary its size) can be ascertained to any accuracy; that could equally be
obtained through the impacts of very fast particles with the electron. It thus has thill
meaning also, if we designate the electron as a particle. According to experience we
completely disturb the electron's mechanical behaviour through such an observation of
its position, through the Compton effect or collision effect respectively. At the moment
when the position is 'q', the momentum is wholly undertermined:
pq - qp = h/21ri.

. . . Analogous considerations may be applied to the velocity of the electron. The

following experiment might perhaps be applied to the defmition of the words 'velocity
of the electron': at a known time one suddenly makes all forces on the electron zero,
then the electron runs on linearly, and one deduces the velocity perhaps from the
Doppler effect of the reddest possible light. The accuracy would be the greater the
redder the light concerned, but then the electron must run correspondingly longer with·
out external forces. Then one switches on the forces again. The [position] accuracy
depends on the distance for which the electron remains without forces:
pq - qp = h/21ri •

. . . Such considerations may be repeated in the same way for all canonical coordinates.
One will always fmd that all thought experiments have this property: when a quantity p
is pinned down to within an accuracy characterised by the average error p, , then the
canonical coordinate q can only be given at the same time to within an accuracy charac-
terised by the average error
q, "" h/21rp,.

Heisenberg's derivation of observational uncertainty could not be called

rigorous. Even when, writing it up in his paper on the subject,14 he filled out
the discussion of the position measurement to include explicitly the
momentum uncertainty resulting from the Compton effect, Bohr could still
object that the finite aperture of the microscope had been ignored. 15 And
even when this and other points had been taken into account the derivation
remained philosophically unsatisfactory: on the one hand it was operationally
based, but on the other hand it was phrased entirely in terms of concepts that
were operationally undefined. As a demonstration of the limited applicability
of the classical concepts in practice, and as an order of magnitude estimation
of the resulting uncertainty, it was however convincing and effectively valid.
Moreover, it was accompanied by a derivation of the quantum-theoretical
uncertainty with which it was in complete agreement. Working from the
Dirac-Jordan theory, Heisenberg interpreted the assumption that q should
be determined to an accuracy q 1 as q' through the requirement that the
probability amplitude of a function 1/, IS(1/, q)l2, fell at q = q' ± q 1 to e- 1
of its value at q = q' . Working out the probability 1S(1/,p) 12 he found that
116 CHAPTER 9

the corresponding points p' ± PI were then given by P1Q1 = h/2rr, in agree-
ment with the thought experiments.
In his derivation of the uncertainty relations, which he extended in his
paper to the observation of energy and time, Heisenberg showed that in
terms of the usual kinematics the essential quantum discontinuity imposed a
restriction upon our observations, which restriction was accurately reflected
by quantum mechanics. But while this specified the limits within which the
classical concepts could be consistently applied, it did nothing to define the
relationship between theory, observation and concepts. In another part of
his letter, however, and again in his paper, Heisenberg did approach this more
fundamental problem. 16 Arguing from his thought experiments on the
measurement of position, velocity, etc., he asserted that these concepts were
operationally well defmed, and therefore perfectly validY The problem lay
in combined conceptions, such as those he had earlier characterised as having
"too many c-numbers": "particle", equivalent to particle path, a series of
joint position-momentum specifications, or "wave". In the letter to Pauli he
wrote that 18

It is therefore meaningless to talk, for example, of the l-S 'orbit' of the electron in the
Hydrogen atom.For if we wish to actually determine the position of the electron
essentially more accurately than to within 10-8 em, the atom will already be destroyed
by a single observation. The word I-S 'orbit' is thus, as it were, already purely experi-
mentally, i.e. without knowledge of the theory, meaningless. On the other hand the
imagined position determination can be repeated in many I-S Hydrogen atoms. So there
must be an exactly determined probability function (the well-known l/J\s(q) l/Jrs (q»
for the given energy I-S. The probability function corresponds to the classical 'orbit'
over all phases. One can say, with Jordan, that the laws of Nature are statistical. But one
can say with Dirac, and this seems to me essentially more profound, that all statistics are
flrst introduced through our experiments. The fact that we do not know in which
position the electron will be at the moment of our experiment arises only, so to speak,
from the fact that we do not know the phase in advance if we know the energy: (Jw -
wJ = h/2m) and in the classical theory it was in this respect no different. That we
cannot know the phase, without once again destroying the atom, is characteristic of
quantum mechanics.

One could not define or measure precisely the phase of an electron orbit of
given energy, and similarly, Heisenberg noted in his paper, one could not
specify the continuous path Goint position and velocity) of an effectively
free electron. 19 Born's interpretation of the spreading wave packet in terms
of possible continuous paths of particles was operationally unsound. But one
could use a series of measurements to achieve an approximate specification of
the electron orbit in an atom, and one could similarly measure and define
 UNCERTAINTY PRINCIPLE/COPENHAGEN INTERPRET A nON 117

an approximate, discontinuous, path of an electron as a time sequence of

distinct position observations. Whether Dirac had anything like this in mind
when asserting that statistics were introduced through the experiments is
extremely doubtful - he tended as we shall see to avoid all questions of
interpretation, and was probably only concerned, as in his transformation
theory, to keep these questions and their statistical overtones out of the
mathematical formalism. But turning this assertion to his own use, Heisen-
berg now claimed, on the basis of his analysis, that "the electron path comes
into existence only when we observe it."20
Heisenberg also disagreed with the approach of Born and Jordan on the
related problem of causality. Born had declared his belief in the ultimate
rejection of causality.21 Jordan had argued that the existing quantum laws
could only be interpreted statistically (a fact that he associated with the
inclusion of the imaginary constant in the fundamental equations), refusing
to commit himself as to whether an underlying causal description might or
might not exist. 22 But both had argued, and were arguing at this time,23
entirely within the assumption of the classical kinematical conceptions. To
Heisenberg, Dirac and Pauli these conceptions themselves were invalid, or at
least restricted in their validity, and this meant that the classical concept of
causality was effectively undefined. In a discussion strongly reminiscent of
the earlier analysis of Senftleben,24 Heisenberg now insisted that "we cannot
know, as a matter of principle, the present in all its details" and argued that
"since physics has to confine itself to the formal description of the relations
between perceptions", the causality concept was simply irrelevant. 2s
In his letter to Pauli, Heisenberg explained that he was still very unclear
on many points, and that he was writing to clarify his own thoughts, and he
asked for Pauli's "relentless criticism" .26 A fortnight later, having written up
the first draft of his paper, he repeated the request. 27 But Pauli apparently
had little criticism to offer, and the final version of the paper, submitted on
23 March, differed little in essential content from the letter of 27 February.
But bringing all his earlier ideas together, Heisenberg did introduce as a foun-
dation for his argument the essential discreteness of quantum mechanics, and
the implications for the measurement and definition of velocity of a discrete
space-time. 28 He also compared, in the introduction of this paper, the uncer-
tainty relations and their place in quantum mechanics with Einstein's principle
of the constant velocity of light and its place in special relativity theory. 29
And in later recollections he associated the origins of the uncertainty principle
with Einstein's observation that, to quote Heisenberg, "it is the theory that
decides what we can observe.,,3o But the close relationship between theory
118 CHAPTER 9

and observation was hardly new to Heisenberg, and the model of special
relativity seems to have been introduced after Heisenberg had completed his
paper rather than before. There is nothing of it in the February letters to
Pauli, and having been introduced to the first draft of the paper it was sub-
sequently omitted. On 9 March Heisenberg wrote to Pauli on the subject: 31

I do not believe at all that one can somehow make the quantitative laws plausible through
the equations PI ql ~ h/21ri. But that is no different in quantum mechanics from any-
where else. For example in relativity theory, the principle of constant light velocity is
also unfounded. Why should not the light velocity depend ultimately on the masses at
infinity? The statement of constant light velocity is only the simplest if one accepts
Einstein's dermition of simultaneity. So I believe also: if one once knows that P and
q are not simple numbers, but that PI q I - h/21ri, then the statement that P and q are
matrices is the simplest conceivable. I am, naturally, well aware that this formulation
might appear unsatisfactory, but is it not the same arbitrariness as is met in all physical
theories? I have written something on this in the conclusion. But this conclusion is
generally still very dubious, and I can imagine that I might still change it completely or
leave it out completely.

COMPLEMENT ARITY

In the past, Heisenberg had often agreed with Pauli on the need for firm
physical foundations of a theory, but had often been carried away by the
success of his own unfounded speculations. Since the introduction of his
new kinematics he had held firmly by Pauli's belief that quantum mechanics
must ultimately receive a clear physical foundation based on the operational
definition of all the terms involved. But the magnitude of this task had
forced both physicists to concentrate on the limits of the old concepts
rather than on new foundations, and successful in his investigation of these
limits Heisenberg was once again carried away. The title of his paper may
be translated as "On the perceptual (anschaulichen) content of quantum-
theoretical kinematics and mechanics", and in it he claimed that with the
addition of the uncertainty principle quantum mechanics became in some
sense visually, perceptually or intuitively - his use of the word anschaulich
was now shifting slightly - consistent. He claimed in other words that the
fundamental problem of establishing a new conceptual foundation appropriate
to the theory was now more or less solved. In fact, of course, Heisenberg's
interpretation did not constitute a proper foundation and he, and the pre-
sumably critical Pauli, appear from the last letter cited to have realised this.
Although the fundamental quantum discontinuity was used in the paper as
a foundation of the uncertainty principle there was no logical connection
 UNCERTAINTY PRINCIPLE/COPENHAGEN INTERPRETATION 119

between the two, and the latter was more of an excuse than an explanation.
But Heisenherg was content with what he had got. The defensive tone of
Heisenberg's letter suggests that Pauli was not so happy, but he had no
foundation to offer himself, and was besides deeply involved with technical
aspects of the theory. The task of putting Heisenberg's ideas onto some sort
of a physical basis thus fell on Bohr, his severest critic of the moment, who
had yet to publish anything on the new quantum mechanics.
Of all those actively involved in the search for a new quantum mechanics
in the 1920s, Bohr was at once the most radical and the most conservative.
He had been initially responsible for the idea that classical mechanical and
kinematical concepts were incapable of describing quantum phenomena,
and he had continued to believe this throughout. But he had also held fast to
the belief that these concepts, and especially those of the classical wave theory
of light, could not be replaced. 32 While most of his colleagues kept an open
mind on the issue of wave-particle duality, this belief had led Bohr to take a
firm stance in support of the classical wave theory and against the light-
quantum concept and, in late 1923, against Heisenberg and Born's programme
of a discrete physics. 33 It had also led him, in early 1924, to the virtual
oscillator model of the interaction between matter and radiation, and while
Born and Heisenberg accepted this merely as a heuristic device, and Pauli
rejected it outright, Bohr appears to have held to it as to a genuine physical
theory.34 Early in 1925, however, an accumulation of new experimental
evidence had forced him to reconsider. First the results of Einstein on gas
statistics, Ramsauer on barrier penetration by slow electrons, and Davisson
and Kunsman on electron scattering appeared to extend the wave-particle
duality to material collisions, for which energy-momentum conservation
could not be rejected as it had been in the virtual oscillator theory. 35 Einstein's
study of the statistical mechanics of a monatomic ideal gas showed that such
a gas had to be treated as having elements of both wave and particle behaviour,
just as had black-body radiation. In Ramsauer's experiments a slow electron
incident upon an atom was shown to have a definite probability of passing
right through the atom without any change in its motion. This, the first
example of a quantum barrier penetration phenomenon, suggested that the
electron might partake of wave as well as particle properties. Finally the
results of Davisson and Kunsman on the scattering of slow electrons through
crystals showed a periodic variation of scattering intensity with scattering
angle, interpreted in Born's Gottingen department in the Spring of 1925 as
electron diffraction, again a wave phenomenon of matter. Then, on top of all
this, the results of Geiger and Bothe showed that, as most physicists by then
120 CHAPTER 9

expected, the rejection of energy-momentum conservation was non-viable

even for radiative processes. 36
Bohr's conviction was still sufficient for him to speak out against energy
conservation in a talk delivered on 20 February 1925,37 but on 18 April he
wrote to Heisenberg that he was prepared for all eventualities, "even for an
acceptance of a coupling process in distant atoms. The costs of this accept-
ance are of course so great", he wrote, "that they cannot be measured in the
usual space-time description." 38 The results of Geiger and Bothe were in fact
already in the post, and three days later Bohr replied to Geiger that 39

I was completely prepared that our proposed point of view on the independence of the
quantum processes in separated atoms should tum out to be incorrect .... Not only
were the objections of Einstein [to the implications of the virtual oscillator theory J very
unsettling; but recently I have also felt that an explanation of collision phenomena,
especially of Ramsauer's results on the penetration of slow electrons through atoms,
presents difficulties for our usual space-time description of a kind similar to those
presented by a simultaneous understanding of interference phenomena and a coupling
of the change of state in separated atoms through radiation. In general I believe that
these difficulties so far exclude the maintenance of the ordinary space-time description
of phenomena that in spite of the existence of coupling conclusions concerning an
eventual corpuscular nature of radiation lack a satisfactory basis.

In this letter Bohr appeared calm, but he seems to have been struggling hard
to adapt to the new situation and to comprehend and perhaps accept some-
thing of the more radical ideas of Pauli. He had written to Heisenberg that
"particularly stimulated by discussions with Pauli, I am these days working
slavishly to the best of my power to accustom myself to the mysteries of
nature and to attempt to prepare myself for all eventualities."40 And on the
same day that he wrote to Geiger, he also wrote to Franck: 41

I have long been intending to write to you again, for the uncertainty about the correct-
ness of my reflections on collision processes, to which I gave expression in the postscript
to my last letter, has since been more and more strengthened. In particular there are the
Ramsauer results on the penetration of slow electrons through atoms, which do no~
appear to tie in with the accepted point of view. Indeed these results must offer difficulties
for our usual space-time description of nature of a similar kind to the coupling of transi-
tions in distant atoms through radiation. But then there are no longer any grounds for
doubting such a coupling, or for doubting conservation laws in general .... Also the
thermodynamic considerations of Einstein were very disturbing indeed .... Moreover,
I've just now heard from Geiger that his experiments have decided in favour of coupling
and there is really nothing to do but take our attempted revolution as painlessly as
possible into oblivion. Nevertheless, we cannot forget our goals that easily, and in the
last few days I've been tormented by all sorts of wild speculations, in order to fmd an
 UNCERTAINTY PRINCIPLE/COPENHAGEN INTERPRETATION 121

adequate foundation for the description of radiation phenomena. On this I have talked
a lot with Pauli, who is still here ....

On 1 May, in reply to Born's advocacy of the theory of matter waves, which

he wholly rejected, Bohr again indicated the direction of his thoughts: 42
Quite apart from the question of the correctness of such objections against your theory,
I should like to stress that I am of the opinion that the assumption of coupling between
stationary states in different atoms through radiation excludes any possibility of describ-
ing the physical situation by visual pictures. By my utterances about coupling in my
letter to Franck I meant only that I had already come to suspect that for collision
processes such models had shown a more inferior application than usual. In essence, this
is indeed purely negative, but I feel, especially if the coupling should really be a fact, that
one must then take refuge to an even higher degree than before in symbolic analogies.

Finally, in a postscript written in July to a paper based on the virtual oscillator

theory, Bohr noted that the coupling issue did not involve a decision between
the wave and particle pictures but 43
rather the problem as to how far the space-time pictures in terms of which physical
phenomena have so far been described can be applied to atomic processes. . .. In view
of the recent results one should not be surprised if the required extension of classical
electrodynamics should lead to a far-reaching revolution of the concepts on which the
description of nature has so far been based.

Bohr could have made the last point ten years earlier. The rejection of visual
pictures and restriction to symbolic analogies suggested in his letter to Born
were nothing new, but just brought him back into line with Pauli and (by
this time) Heisenberg, and indeed with his own earlier views. The difference
was that whereas before his peSSimistic statements as to the impossibility of a
space-time description of quantum phenomena had always gone hand in
hand with a more optimistic approach, according to which a wave-theoretical
approach might be saved at the expense of the laws of conservation and
causality, now the pessimism dominated. He could not forget his hopes, and
must still have retained something of a wave-theoretical bias, but he could
think of no new escape and as the new quantum mechanics developed in
1925-26 he remained silent and apart. According to Heisenberg's recollection
it was only after Schrodinger's Copenhagen lecture that, in the provocative
company of Heisenberg and Dirac, Bohr was once again stirred into action,
and then it was apparently in opposition to the approach being adopted by
Heisenberg and Pauli.
We have little evidence of Bohr's arguments during the Autumn of 1926,
only Heisenberg's recollection that while he himself took part optimistically
122 CHAPTER 9

in the development and elucidation of quantum mechanics Bohr remained

pessimistic, and that the relationship between them grew very tense. 44 But
Bohr seems to have had two main objections to Heisenberg's work. First, he
seems to have criticised it, as Heisenberg had earlier criticised Born's, for
being too concerned with the formalism. Unable to develop a new conceptual
foundation for the theory both Heisenberg and Pauli were forced to concen-
trate on the limitations of the old conceptions, and to approach this entirely
- at least until Heisenberg sorted out his ideas in February - from the study
of the formalism. Bohr, though unable to do any better himself, insisted that
the limitations of the classical concepts should be founded directly on
physical and philosophical principles, and the formalism then derived from
the physics. Heisenberg would not have disagreed with this on principle, but
as often before he preferred to go for results first. Bohr's second objection
seems to have been to Heisenberg's emphasis upon the particle picture at the
expense of the wave picture, and this was again related to earlier disagree-
ments. Heisenberg had always emphasised the discreteness of quantum
theory ,45 and now that he was won over to Pauli's views he presumably
shared Pauli's operational preference for a particle theory and objection to
a pure field theory.46 Moreover, by concentrating on the problem of the
observationally possible precision, or degree of localisation, in terms of the
classical concepts he naturally adopted particle terminology. Since an observa-
tion always corresponds to an apparently localised event (observations of
wave properties being second order deductions), any discussion of the mea-
surement problem must emphasise the particle as reality, and even when he
discussed the theoretical problem of the orbit concept in terms of spreading
wave packets, in his uncertainty principle paper, Heisenberg interpreted all
observations, as localisations or reductions of the wave packet, in terms of the
particle conceptions. Operationally this view-point was perfectly reasonable,
but it clearly did not do sufficient justice to the wave concept for Bohr's
liking. Bohr had always preferred a field theoretical approach, and even now
he appears to have supported his student Klein in his advocacy of an interpre-
tation of quantum mechanics based upon Kaluza's analogue to Weyl's unified
field theory - the theory in opposition to which Pauli had first developed
his operationalist preference for the particle theories. 47 At the very least he
seems to have insisted upon the symmetry of the wave-particle duality being
maintained. 48
In February, Bohr left Copenhagen for a holiday, and when he returned
Heisenberg had formulated the considerations of his uncertainty principle
paper. 49 Here the priority of the particle picture was maintained, and although
 UNCERT AINTY PRINCIPLE/COPENHAGEN INTERPRETATION 123

his analysis of observations was lacking in rigour and divorced from any
general physical principles (the vaguely supposed derivation from the quan~'!m
discreteness was a perfect example of what Pauli elsewhere called a "specifi-
cally Gottingen bad habit", into which Heisenberg relapsed when released
from Bohr's supervision),50 Heisenberg seemed perfectly happy with it. Bohr
was not, and the arguments recommenced with renewed vigour,51 the first
manifestation being in an appendix to Heisenberg's paper in which he noted
some objections raised by Bohr, including that 52

Above all the uncertainty in the observation does not depend exclusively upon the
occurrence of discontinuities, but is directly connected with the requirement that justice
be done at the same time to the different experiences, expressed on the one hand by
the corpuscular theory and on the other hand by the wave theory. For example, the
necessary divergence of the wave-packet due to fInite aperture has to be taken into
account in the use of an imaginary ,),-ray microscope. This fIrst leads to the observation
of the electron position in the direction of the Compton recoil being known only within
an uncertainty, which leads to the relation [p 1 q 1 - h). Next, it is not suffIciently
stressed that the simple theory of the Compton effect is strictly applicable only to the
free electron. The resulting caution in the application of the uncertainty relation is
essential, as Professor Bohr has stated clearly, among other things for an overall discus-
sion of the transition from micro- to macro-mechanics and for the discussion of the
generation of the orbital path by observation. Finally the considerations on resonance
fluorescence are not quite correct, as the connection between the phase of the light and
that of the electron is not so simple as assumed.

Bohr's technical objections, though valid, did not affect Heisenberg's conclu-
sions, and Heisenberg wrote to Pauli on 14 March that, in respect of the
position experiment, "I believe that everything I wrote is correct." S3 But far
more important were Bohr's criticisms on the lack of foundation of the
analysis - it did not follow from the occurrence of discontinuities - and
on the insufficient justice done to the wave picture. On 4 April Heisenberg
wrote again to Pauli, reporting among other things on friendly discussions
with Bohr: s4

Otherwise there continues to be general harmony here, and constant discussion of

thought experiments. I am arguing with Bohr as to how far the relation has its founda-
tions in the wave or discontinuity sides of quantum mechanics. Bohr stresses that in the
-y-ray microscope, for example, the inflexion of the wave is essential, I stress that the
light"<luantum theory and even the Geiger-Bothe experiment are essential. By over-
emphasising one the one side and one the other we can discuss much without anything
new.

In another letter of 16 May, Heisenberg reported on another error in his 'Y-ray

124 CHAPTER 9

microscope experiment, which was not however crucial, and on his continued
differences with Bohr: 55

Since my return from the Easter holiday ... we have talked a lot here about the quantum
theory. Bohr will write a general work on the 'conceptual structure', from the viewpoint
that 'there are waves and particles' - if one begins at once with that one can naturally
also make everything contradiction free. Prompted by this work Bohr has called my
attention to the fact that there was still something essenful overlooked in my work
(Dirac also asked me about it subsequently): in the -y-ray microscope one could first
of all imagine: one determines the direction of a falling light-quantum and of the reflected
light-quantum, then one knows according to the Compton effect both the position and
velocity very precisely (more so than PI ql - h). But one can not in fact do this on
account of the bending of the light (wave theory!) To give the accuracy A the micro-
scope must have an aperture of order 1. So sure enough the relationship PI q I ...; h comes
out naturally, only not quite as naturally as I had thought. Besides this several points
could also be better put and better discussed in all particulars if one began a quantitative
discussion immediately with the waves. Nevertheless, I am naturally now as before of the
opinion that the discontinuities are the only interesting things in quantum theory and
that one can never stress them enough. For this reason I am also, now as before, very
happy about my last work - despite the known defects all the results of the work are
indeed correct, and on them I am also in agreement with Bohr. Otherwise there are at
present between Bohr and I essential differences of taste over the word anschaulich.

As we have seen, Heisenberg had maintained that with the uncertainty

relations quantum mechanics was anschaulich. 56 But to Bohr the require-
ments of Anschaulichkeit were much stricter, and could be satisfied only by
a fully consistent classical visualisation. Quantum mechanics was not anschau-
lich in this sense, and Bohr's complementarity principle was to be based on
the insistence that it was not, and could not be SO.57 Unfortunately Bohr,
emphasising the importance of the classical conceptions, seems to have
emphasised especially, as he had so often before, the importance of the wave
concepts. For writing again on 31 May Heisenberg explained to Pauli that
there had in effect been a serious division between the adherents of the
wave mechanics and those of the matrix-mechanical Dirac-Jordan theory. 58
In earlier correspondence, Heisenberg and Pauli had joined up in opposi-
tion to a movement in support of the wave mechanics, led by Ehrenfest,
Darwin and Klein.59 More recently a misunderstanding between Heisenberg
and Klein had lead to personal animosity, with Bohr weighing in on Klein's
side, and calm discussion had given way to trenchant argument. 60 Pauli,
however, had resolved the personal problem, everything was now calm
again, and Heisenberg could look at things more dispassionately:61
So I came into a battle for the matrices and against the waves: in the heat of this battIe I
 UNCERTAINTY PRINCIPLE/COPENHAGEN INTERPRETATION 125

often criticised Bohr's objections against my work too sharply and thus, without knowing
it or wishing it, offended him personally. If I now reflect on those discussions I can
understand very well that Bohr was annoyed by them.

What led Heisenberg to take such a humble position as he now did is unclear,
but it could be that in the wake of the personal reconciliation brought about
with Pauli's help Bohr told him something of the content of his proposed
paper on the "conceptual structure" of quantum mechanics. For in this paper,
a version of which was first delivered at Como the following September, it
became apparent that Bohr's insistence on the wave-particle duality did not
directly threaten Heisenberg's stress on the particle picture for observations,
and did not prevent Bohr from treating the quantum discontinuity as being
of prime importance.62 In the heat of the argument, Heisenberg seems to have
confused Bohr's attack on the particle picture with an attack, such as Bohr
would never in fact have made, on the fundamental quantum postulate, and
realising this he may have felt a little ashamed.
What Bohr did was to fill in the logical gap in Heisenberg's analysis between
the quantum postulate of discontinuity and the derivation of the uncertainty
relations. He did this in a way that preserved the symmetry of the wave-
particle duality, replacing Heisenberg's identification of observation and
defmition by a contrast between these two concepts. Heisenberg's thought
experiments showed the impossibility of a joint specification and observation
of position and momentum, and whereas Heisenberg had taken the funda-
mental contradiction as that between position and momentum Bohr took it
as that between specification and observation. Looked at this way the
problem was not a specific one relating to conjugate coordinates, but the
general philosophical one of the observer and the observed, with which Bohr
was familiar, to which he was sympathetic, and through which he could at
last understand and derive from a physical basis the non-applicability of
classical concepts and lack of a visualisable theory that had long bothered
him. The argument as presented in his paper was that 63
Our usual description of physical phenomena is based entirely on the idea that the
phenomena concerned may be observed without disturbing them appreciably .... [But]
the quantum postulate implies that any observation of atomic phenomena will involve an
interaction with the agency of observation not to be neglected.

The outcome of this was that 64

On one hand, the definition of a state of a physical system, as ordinarily understood,
claims the elimination of all extemal disturbances. But in that case ... any observation
will be impossible, and, above all, the concepts of space and time lose their immediate
126 CHAPTER 9

sense. On the other hand, if in order to make observation possible we posit certain
interactions with suitable agencies of measurement, not belonging to the system, an
unambiguous defmition of the state of the system is naturally no longer possible, and
there can be no question of causality in the ordinary sense of the word.

In one sense at least Bohr's argument was not wholly satisfactory, for while
he claimed that the quantum postulate "forced" the impossibility of joint
observation and definition this is far from self-evident. 65 But he did place
Heisenberg's analysis on much sounder and more general grounds. Heisenberg
had adopted Pauli's operationalist creed that observability and defineability
should be equated in a consistent theory. But unable to create such a theory
he carried this creed into his analysis of a theory that was still, in terms of
its conceptual foundations, inconsistent. Bohr's analysis showed that this
inconsistency lay in the very fact that an operational definition of the kine-
matic concepts needed was impossible. The ideals of observation and
definition, both necessary to any physical theory, were in fact incompatible.
Bohr defined this combination of joint necessity and mutual incompatibility
through the notion of "complementarity", and from the complementarity of
observation and definition he derived that of space-time description and
causality, and that of the wave and particle pictures.66 He also gave two
derivations of the uncertainty principle, one from the wave-particle duality
that reflected his earlier arguments with Heisenberg, and one directly from
complementarity.67
Pauli, as may be expected, accepted Bohr's new ideas enthusiastically,
criticising the presentation on a few minor points but declaring that "in
general I am very much in agreement, both with the overall thrust of your
paper and with most of its details."68 The main force of the comple-
mentarity principle lay in its demonstration that for quantum phenomena
any operationally defined system of concepts was impossible, the processes
of operation and definition being themselves incompatible, and in this sense
the principle represented a victory for Bohr over Pauli. But Bohr had also
now admitted that the classical conceptions in particular were incapable of
consistent application, and that this limitation arose, as Pauli had always
insisted, from their being operationally ill-defined. In a few years the dialogue
between the two men was to be reopened in the new context of nuclear
physics, but for the moment they were in agreement, a situation that must
have given great satisfaction to them both. Of the other physicists, Jordan
later became one of the chief advocates of complementarity,69 and for the
present he, Heisenberg, Dirac and Born at least refrained from criticism.
 UNCER T A-INTY PRINCIPLE/COPENHAG EN INTERPRET AnON 127

A COMMON INTERPRET A nON

By the Summer of 1927 there was still a large measure of disagreement

between the quantum physicists. Bohr and Heisenberg still disagreed as to
how much emphasis should be placed on the particulate nature of measure-
ment, and Pauli could still write to Bohr in August in condemnation of Jordan
and Klein's continued use of the "phenomenological" wave-mechanical
approach.:ro But Bohr, could 'now write back in agreement with Pauli, and at
the fifth Solvay congress in October Born and Heisenberg could sink their
differences of the past few years in a joint paper.71 By December, another
joint paper had bridged the gap between Pauli and Jordan. 72
In the papers and comments delivered to the Solvay congress by Born and
Heisenberg, Bohr, Pauli and Dirac, there were still significant differences of
emphasis. Bohr reiterated his notion of complementarity, Born and Heisen-
berg concentrated on the interpretation of the theory within the classical
concepts, and Dirac, in discussion, insisted on the restriction of the theory's
application to our knowledge of a system, and on its lack of ontological
content. 73 But these differences were now treated more as matters of personal
taste and metaphysics than as anything physically serious, and it was clear
that all the physicists concerned, together with Pauli who was also present,
effectively shared a common interpretation. While their opponents, Einstein,
SchrOdinger, Lorentz and, for the time being, de Broglie, continued to argue
heatedly about the interpretation,74 they treated it rather as a problem of the
past, and one that had now been more or less settled. What mattered now was
the problem of extending quantum mechanics to general electromagnetic
phenomena, of developing a relativistic quantum electrodynamics. 7s This
problem had already dominated the thoughts of Jordan, Pauli and Dirac since
the previous February,76 and before long Heisenberg was also fully engaged
upon it.77
Opposition to what became known as the Copenhagen interpretation has
never ceased, and this interpretation has developed, particularly through
analysis of the measurement problem, since the 1927 Solvay congress. 78 But
as the least common denominator of the leading quantum physicists' views it
was already established at that congress as a commonly held - and, since its
opponents were unable to agree among themselves, the only commonly held
- interpretation. On this interpretation it was agreed that, as Dirac explained,
the wave functions represented our knowledge of the system, and the reduced
wave packets our more precise knowledge after measurement. Within the
classical conceptions of waves and particles the waves could be interpreted,
128 CHAPTER 9

as Born and Heisenberg wrote, as probability functions, corresponding to our

knowledge of the system, of localised measurements. These measurements
corresponded to kinematic and mechanical concepts associated with particles,
and effectively created a situation that corresponded, within the limits of
uncertainty, to the particle concept. Similarly, successive measurements could
create a situation corresponding to the notion "path of a particle", just as
Heisenberg had earlier explained. Within the classical conceptions, causality
was effectively abandoned as irrelevant (after some toing and froing in
individual papers by Born and Heisenberg, in which each advocated the posi-
tion earlier adopted by the other, it was replaced by the concept of weak
causality or causation).79 And it was agreed that the quantum-mechanical
theory corresponded precisely to the limits of observation, that within the
context of the theory anything beyond these limits was practically irrelevant,
and that in this sense the theory was "complete" and, given the interpretation,
"without contradictions". 80 Finally, the limits of the classical concepts
could be explained, if desired, through Bohr's notion of complementarity.
 CHAPTER 10

CONCLUDING REMARKS

Bohr's principle of complementarity and the ensuing Copenhagen interpreta-

tion brought both the creation of quantum mechanics and, for the time
being, the Bohr-Pauli dialogue to an end. The long sought for new system of
operationally defined concepts upon which quantum mechanics was to have
been built upwar.ds had remained elusive, and in this sense Bohr's views had
finally prevailed. But the inadequacy of the existing concepts, and in particular
of visualisable models, had now been established and given a foundation. As
Pauli had always maintained this foundation lay in their operational in-
adequacy, expressed in terms of a complementarity between definition and
observation which effectively prohibited any operationally based definition.
In philosophical terms the compromise was somewhat unfortunate. By
stopping short of any assertions as to the nature and location of any reality
underlying the defined limits of knowledge the Copenhagen interpretation
masked the very real differences of philosophical viewpoint between its main
creators and proponents, and in fact ran generally counter to their views. For
all their insistence on the role of observation, both Pauli and Heisenberg were
for example, and were to remain, philosophical realists. Scientifically, however,
this limitation of the Copenhagen interpretation provided it with a great
strength, wedding it closely to the theory and rendering it effectively immune
to changes of philosophical opinion. Only when identified with an anti-
realist philosophy has it ever come under serious attack, and such an identifi-
cation is, as our analysis has shown, a misleading one.
The close relationship between the formalism of quantum mechanics and
its Copenhagen interpretation suggests that their roots too were closely
related, and our analysis has confirmed that this was in fact so. The analysis
is of course far from complete. In portraying the creation and development of
quantum mechanics primarily in terms of the Bohr-Pauli dialogue, we have
inevitably overstressed some and understressed other aspects of the history.
The most significant omission has, of course, been that of the detailed back-
ground to SchrOdinger's wave mechanics, especially in respect of its relation-
ship to Einstein's work on quantum gas statistics.! Other areas in which we
have done less than full justice to the available historical literature include
the technical details of Heisenberg's work in the early middle 1920s;2 the
129
130 CHAPTER 10

contributions to quantum mechanics of Dirac;3 and the technical details of

Bohr's struggles with the problems of atomic structure in the early 1920s.4 In
stressing the importance of Pauli's views and their opposition to those of
Bohr we may also have painted rather too brilliant a picture of the young
Pauli. Brilliant he certainly was, and arguably one of the most brilliant
physicists of this century; but he was also young. His youthful criticisms of
Weyl, made we must remember when he was still in his teens, were certainly
pointed. But he had not yet acquired- the impressive soundness of his mature
years and the criticisms were not altogether coherent in their details. More-
over, the maturing process took place largely under Bohr's supervision during
the year that Pauli spent in Copenhagen, and to the relationship between
these two physicists that we have portrayed, that of friendly adversaries, must
be added that of pupil and teacher. 5
Pauli's prominence in our story must also not be mistaken for dominance,
which characteristic belongs rather to Bohr. But despite this qualification our
analysis has indicated that Pauli's creative influence was a crucial factor in the
origins and development of quantum mechanics. This is not to take away
anything from Heisenberg's abilities and achievements. Whereas Heisenberg's
ability to get to the technical heart of a problem might still conceivably have
led him in due course to something like the new kinematics, even without
Pauli's aid, it is most unlikely that Pauli on his own would have got there. It
is however clear that the creation of the new kinematics, though Heisenberg's
in the end, was the result of a very close collaboration between the two
young physicists and of a methodological programme, based on a demand for
operational consistency and for an operationally defined system of concepts,
laid down by Pauli.
Elsewhere too Pauli's views, standing in opposition to Bohr's insistence
that the classical concepts were somehow rooted in our perceptions and thus
irreplaceable, again provided the starting point for Heisenberg's brilliant
analyses. How far Bohr and Pauli engaged in an active and conscious dialogue
concerning their beliefs is far from clear, but looking at the history of the
creation of quantum mechanics in terms of such a dialogue we find that the
key events fall naturally into place. The peculiar history of the virtual oscilla-
tor theory, the reactions to which cut right across the traditional quantum-
theoretical camps, finds a straight-forward explanation and interpretation.
Heisenberg's new kinematics appears not only as a development of his own
work on fluorescent polarisation, in terms of which alone its pedigree
seems incomplete, but also as a direct consequence of his acceptance, as a
result of his own studies, of Pauli's arguments as to what the long sought for
 CONCLUDING REMARKS 131

new mechanics should set out to achieve. Continuing analysis of the

applicability of the classical kinematic concepts also prepared the way for the
development of the transformation theory, and led directly to the considera-
tion of position and momentum measurement and thus to Heisenberg's
uncertainty principle. Finally, Bohr's principle of complementarity also fits
naturally into the story as a further development of Heisenberg's analysis and
as a considered response to Pauli, producing an agreed conclusion, for the
time being at least, to their long debate.
This debate was clearly not the only focal point in the development of
quantum mechanics. Other physicists viewed the problems from other
perspectives, and the polarisation between Bohr and Pauli, conditioned by
their teacher-pupil relationship and by their common stance on topics such as
the core theory of the atom, was not always in emphasis. But the connections
we have made are nevertheless genuine ones, and although the presentation
of the historical creation of quantum mechanics in terms of the Bohr-Pauli
dialogue may well be somewhat artificial in purely historical terms it does
offer an insight to the underlying themes and currents of thought, and does
add to our understanding of the events that took place. Heisenberg did set
out in 1925 after discussions with Pauli to do precisely what Pauli had
prescribed a few months earlier. Bohr did present his principle of comple-
mentarity primarily in terms of the relationship between observation and
definition, and not in terms of what was in fact the deduced relationship
between wave and particle pictures. And there can be little doubt that
throughout the period we have discussed, and throughout the creation of
quantum mechanics, the relationship between observation and definition was
in fact crucial to the creative enterprise.
Another feature to have emerged from our treatment, and one that seems
to be of some historical importance, concerns the relationship between
different branches of physics. Perhaps because historians of physics have
often themselves been specialists in one branch of physics or another, the
way in which the fundamental problems transcended the boundaries within
the discipline seems to have been rather overlooked. Some authors have tried
to relate different areas of speciality with differing views on the nature of the
quantum problem and the interpretation of quantum mechanics. 6 But while
such lines may perhaps be drawn for Einstein and Bohr, the most remarkable
feature of the work of the key creative figures, Pauli and SchrOdinger, is
its lack of sub-disciplinary barriers. Both were not only informed but also
active in all the key areas of theoretical physics of the period, including the
theoretical spectroscopy of the Bohr atomic theory, quantum statistical
132 CHAPTER 10

thermodynamics, and general relativity theory. Other key figures such as

Born and Dirac shared this catholicism, and while it is difficult to assess just
how crucial the cross-fertilisation of ideas between the different subject areas
may have been, there can be no doubt that the fundamental problems of
physics not only cut across the internal boundaries but were also widely
perceived as so doing. In purely technical terms, the new kinematics emerged
from the problem complex of the Bohr atomic theory;but its emergence was
founded on a crystallisation' of ideas derived from much wider considerations.
In particular our treatment has suggested that, as one would indeed expect,
the two great physical achievements of the inter-war period, quantum
mechanics and general relativity, were far from being the isolated and inde-
pendent developments they have previously been portrayed as. In this case as
in others the scientific creativity operated primarily through the consideration
of fundamental physical, methodological and epistemological problems, and
only secondarily through that of their manifestations in any specific area.
Once our attention has been drawn to the organic relationship between
developments in different areas of physics it is natural to enquire how the
creation of quantum mechanics fits in to the much wider context of Western
intellectual history. That some connection exists is not to be doubted, but
the partially progressive nature of physical science and the technical language
in which it is expressed make it no easy matter to determine what that
connection might be. The fact that quantum mechanics emerged from a
debate between scientists of very different philosophical persuasions, with the
Copenhagen interpretation itself being something of a philosophically neutral
compromise, complicates the issue further. The perception of quantum
mechanics by those not responsible for its creation, in terms of indeterminism
and anti-realist instrumentalism, may certainly be linked with other, non-
scientific, trains of thought. Paul Forman has demonstrated the romantic
and anti-determinist nature of the prevailing intellectual milieu of the period,
and has shown that in Germany at least physicists recognised and were to
some extent influenced by this milieu. 7 Stephen Brush has connected the
instrumentalism associated with quantum mechanics with a general tendency
towards romanticism and anti-realism in inter-war intellectual and artistic
activity.8 But causality does not seem to have been a central issue in the
creation of quantum mechanics, and with a few exceptions the physicists
responsible for its creation and interpretation were not themselves instrumen-
talists. The line they did take was indeed much closer to the philosophy
of Kant, 9 and their considerations were in many ways closer to those of their
mid-nineteenth century predecessors, Helmholtz and Maxwell, than they were
 CONCLUDING REMARKS 133

to the philosophies of the twentieth century, by when Kantian epistemology

was distinctly out of fashion. 1o The question as to whether the quantum
mechanics itself, rather than just the way in which it was perceived, may
be directly related to other cultural, philosophical or artistic developments
thus remains an open one. Perhaps it may, but if any worthwhile attempt to
portray such a relationship is to be made it will be necessary first to reach a
very clear picture indeed of the very complex origins of quantum mechanics,
and it is towards this more proximate end that our analysis here has been
conducted.
 NOTES

Short titles are used for works cited in the bibliography, or for repeat citations within
the same chapter. The abbreviations SHQP, BSC, Bohr MSS and ETHZ are used for
the primary source archives given in Section A of the bibliography, PB for the Pauli
Briefwechsel, Bd. 1 (Section B of the bibliography) and SQM for Waerden's Sources
(Section C of the bibliography).

CHAPTER 2. WOLFGANG PAULI AND THE SEARCH FOR

A UNIFIED THEOR Y

1 Stuewer, Compton Effect: but see Chapter 4 below. The Compton effect seems to
have been a turning point for American physicists, but less important to the Europeans.
In Cambridge, for example, a talk on Compton's theory was given to the Kapitza Club
in August 1923 by Herbert Skinner, and the members present signed a note in the
minute book: "Compton is wrong. PK. HWBS. PMB. DRH. J. E. Jones." (Kapitza,
Skinner, Blackett, Hartree were the first four signatories.) Only after a discussion of
Pauli's theory of radiative equilibrium, also based upon light'<}uantum collisions, was
one member, Blackett, won over: "Compton right we hope. ECS. PMB. We hope wrong.
P. Kapitza. DRH. EGD. MHAN. HWBS." (The additional signatory in favour of light-
quanta was Stoner, those against Dymond and Newman.) See Minutes of Kapitza Club,
3 August 1923 and 29 January 1924, SHQP, 38, 2.
2 One possible prominent exception is Sommerfeld, for whom see Chapters 3, 4 below.
3 Hendry, 'Wave-particle duality'. We shall see that it was the rejection of the require-
ment that a description be visualisable that characterised the new quantum mechanics.
Although the criterion of visualisability was firmly established, however, this was far from
being the first time it had been challenged. In the early nineteenth century Coleridge
and other opponents of Laplacian mechanics had already attacked the "despotism of
the eye" that made people judge a theory by its visual properties rather than by more
fundamental criteria.
4 This is reflected in their work, much of which is discussed below. See also Klein,
'First phase', and Raman and Forman, 'Why Schrodinger'.
5 See for example Jammer, Conceptual Development, Waerden, Sources, Serwer,
'Unmechanischer Zwang', and MacKinnon, 'Heisenberg'.
6 A. Einstein, 'Die Grundlagen der allgemeinen Relativitatstheorie', Ann. der Phys. 49
(1916), 769-822, translated in Lorentz, Principle of Relativity, 111-164. The best
account of general relativity theory is probably still W. Pauli, 'Relativitatstheorie',
Encykl. Math. Wiss. 19 (1921), translated as Theory of Relativity (Oxford, 1958). For
a recent historical account see Mehra, Einstein and Hilbert.
7 The covariant tensor formulation was introduced by Einstein and Grossmann in 1913,
abandoned by Einstein in 1914, and reintroduced by him in 1915: A. Einstein and
134
 NOTES 135

M. Grossmann, 'Entwurf einer verallgemeinerten Relativitatstheorie', Zeit. Math. u.

Phys. 62 (1913), 225 -261, and 'Kovarianzeigenschaften der Feldgleichungen der auf die
veraligemeinerte Relativitatstheorie gegriindeten Gravitationstheorie', ibid. 63 (1914),
215 -225; A. Einstein, 'Die formale Grundlage der allgemeinen Relativitatstheorie',
S.-B. Preuss. Akad. Wiss. (1915), 778-786 and 799-801, and 'Die Feldgleichungen
der Gravitation', ibid., 844-847. For a more complete set of references, see Mehra,
Einstein and Hilbert. For the principle of equivalence, see A. Einstein, 'tiber das Rela-
tivitatsprinzip und die aus demselben gezogenen Folgrungen', lahrb .. der Radioaktivitiit
u. Elektronik 4 (1907), 411-461, and 'tiber den Einfluss der Schwerkraft auf die
Ausbreitung des Lichtes', Ann. der Phys. 35 (1911), 898-908, translated in Lorentz,
Principle of Relativity, 99-108. The postulate of general covariance corresponds to the
requirement that the tensor field equations be covariant with respect to the arbitrary
coordinate transformations; it is required only that the coordinate systems be unique
and continuous (Gaussian).
8 They were also concerned with the rival theories of Nordstrom, Abraham, Ishiwara
and especially Mie, but Hilbert's work in particular closely parallelled that of Einstein, and
in 1915 he derived independently the field equations of Einstein's theory: D. Hilbert,
'Grundlagen der Physik', Nach. k6nigliche Ges. Wiss. G6ttingen, Math.-Phys. KI. (1915),
395-407, and see also ibid. (1917), 477-480. The relationship between Einstein's
work and Hilbert's is discussed by Mehra, Einstein and Hilbert, Earman and Glymour,
'Einstein and Hilbert', and Pyenson, 'Gottingen reception'. Born worked with Hilbert
from 1908 to 1914, when he published a work based on Mie's theory: M. Born, 'Der
Impuls-Energie-Satz in der Elektrodynamik von Gustav Mie', Nach. k6nigliche Ges.
Wiss. G6ttingen, Math.-Phys. Kl (1914), 23-37. Weyl worked with Hilbert from 1906
to 1913. His work on Mie's theory was published in his book: H. Weyl, Raum-Zeit-
Materie (Berlin, 1918), Section 26. The fourth edition (1920) of this book was trans-
lated as Space-Time-Matter (London, 1922).
9 Einstein to Weyl, 23 December 1916, ETHZ 91, 536. See J. Dieudonne, 'Weyl',
Dict. Sci. Biog. 14 (1976), 281-285. The Einstein-Weyl correspondence deserves pub-
lication at length, but permission to do this has so far been refused, no reason having
been given, by the Einstein Estate.
10 Weyl, Raum-Zeit-Materie, Sections 34 to 36; H. Weyl, 'Gravitation und Elektrizitat',
S.-B. Preuss. Akad. Wiss. (1918), 465-480, translated in part in Lorentz, Principle of
Relativity, 201-216; 'Reine Infinitesimalgeometrie', Math. Zeit. 2 (1918), 384-411;
'Eine neue Erweiterung der Relativitatstheorie', Ann. der Phys. 59 (1919),101-133.
11 Weyl, 'Grav. u. Elek', 477-478, translation, 215-216.
12 Mie to Weyl, 26 October 1918, ETHZ 91, 674.
13 Sommerfeld to Weyl, 3 July 1918, ETHZ 91, 751. Sommerfeld also criticised the
theory, however, and his criticisms are to be found in Sommerfeld to Weyl, 7 November
1919,11 December 1919 and 6 January 1920, ETHZ 91,752-754.
14 Eddington to Weyl, 18 August 1920, ETHZ 91, 523. Eddington's first enthusiastic
response was in a letter Eddington to Weyl, 16 December 1918, ETHZ 91, 522.
15 Einstein to Weyl, 8 March 1918, ETHZ 91, 539, including two pages of extravagant
praise.
16 W. Pauli, 'Zur Theorie der Gravitation und der Elektrizitat von Hermann Weyl', Phys.
Zeit. 20 (1919), 457-467, and 'Mehrkurperihelbewegung und Strahlenableitung in
Weyl's Gravitationstheorie', Verh. Deut. Phys. Ges. 21 (1919),742-750.
136 NOTES

17 E. T. Whittaker, A History of the Theories of Aether and Electricity 2, (London,

1953),214-217.
18 Einstein to Weyl, 23 December 1916, ETHZ 91,536.
19 Einstein to Weyl, 8 March 1918, in response to Weyl to Einstein, 1 March 1918,
ETHZ 91, 539, 538a.
20 Einstein to Weyl, 8 April 1918, ETHZ 91, 540.
21 Einstein to Weyl, 15 April 1918, ETHZ 91, 541: "So schon Ihre Gedanke ist, muss
ich doch offen sagen, dass er nach meiner Ansicht ausgeschl@ssen ist, dass die Theorie
der Natur entspricht." Einstein had already communicated this paper, though Weyl had
meanwhile become dissatisfied with it, feeling that it did not go far enough: Weyl to
Einstein, 15 April 1918, ETHZ 91, 540a.
22 Einstein to Weyl, 15 April 1918, ETHZ 91, 541, and see also Einstein to Besso,
20 August 1918 and 26 July 1920 in Einstein and Besso, Correspondance, 132-134,
155-158 (Items 46, 52.1).
23 Einstein to Weyl, 19 April 1918, ETHZ 91, 543, with repercussions in Weyl to
Einstein, 27 and 28 April 1918, and Einstein to Weyl, 1 May 1918, ETHZ 91, 543a,
543b, 544; Weyl, 'Grav. u. Elek.', 478-480, not in translation.
24 Weyl, ibid.
25 Weyl, Raum-Zeit-Materie, 1st edition, 226-227, not in 4th edition or translation.
26 Weyl to Einstein, 19 May 1918, ETHZ 91, 545a: "Weiss ich doch nur zu gut, in
einem wie viel Lauteren Verhiiltnis Sie zur Wirchlichkeit stehen als ich."
27 Ibid.: "Behalten Sie flir die wirkliche Welt recht, so bedaure ich, den liebern Gott
einer mathematische Inkonsequenz zeihen zu miissen."
28 Einstein to Weyl, 31 June 1918, ETHZ 91, 546. Writing again on 27 September
1918, ETHZ 91, 548, Einstein lamented that God had not made it easy for them.
This resort to the highest authority was partly light-hearted banter (all the protagonists
in this particular controversy remained on the friendliest of terms), but it also reflects a
traditional Platonic approach to theoretical physics, namely that of asking how a creator
might reasonably have designed things.
29 Einstein to Weyl, 27 September 1918, ETHZ 91, 548, and see also Einstein to
Besso, 26 July 1920, Einstein and Besso, Correspondance, 155-158 (Item 52.1). Ein-
stein and Weyl repeated their respective positions ih papers of 1920 and in a number of
letters between 1918 and 1923, when the extant correspondence breaks off for a few
years: H. Weyl, 'Elektrizitiit und Gravitation', Phys. Zeit. 21 (1920), 649-650; A.
Einstein, ibid., 651; correspondence between Weyl and Einstein, ETHZ 91, 548a-556.
30 For Einstein's continuing conviction see Einstein, ibid., but in September 1918
he wrote that he was certain both of them had no other aim than to fmd the truth, and
suggested that this would be established one way or the other within a couple of years:
Einstein to Weyl, 27 September 1918, ETHZ 91, 548.
31 Pauli, Relativity, 196.
32 A. S. Eddington, 'Relativity of field and matter', Phil. Mag. 42 (1921), 800-806,
and 'A generalisation of Weyt's theory of the electromagnetic and gravitational fields',
Proc. Roy. Soc. A99 (1921), 104-122. ,See also A. S. Eddington, Space, Time and
Gravitation (Cambridge, 1920), Chapter 11, where Weyl's theory is enthusiastically
reviewed, and Mathematical Theory of Relativity (Cambridge, 1923).
33 Eddington to Weyl, 10 July 1921, ETHZ 91, 525. Eddington wrote here in a reply
to a lost letter from Weyl that his own theory was so directly inspired by Weyl's that
 NOTES 137

he had not seen it as a rival until he had received Weyl's objections. Then he had rec-
ognised that they had started from opposite ends, he from the pure geometry and Weyl
from the action principle.
34 Ibid.
35 Ibid.; Einstein saw Eddington's theory as even further removed from reality than
Weyl's, describing it as "beautiful but physically meaningless": Einstein to Weyl, 5
September 1921, ETHZ 91, 551, and see also his letter 6 June 1922, ETHZ 91, 554.
36 G. Mie, 'Grundlagen einer Theorie der Materie', Ann. der Phys. 37 (1912), 511-
534, 39 (1912), 1-40, 40 (1913), 1-66. See Mehra, Einstein and Hilbert, Pyenson,
'Gottingen reception', and L. Pyenson, 'Mathematics, education, and the Gottingen
approach to physical reality, 1890-1914',Europa 2 (1979), 91-127.
37 For the background to this electromagnetic world view see R. McCormmach, 'H. A.
Lorentz and the electromagnetic view of nature' ,Isis 61(1970), 459-497.
38 Hilbert, 'Grundlagen'.
39 See Note 8 above and Einstein to Weyl, 23 December 1916 and 3 January 1917,
ETHZ 91, 536, 537.
40 Mie, 'Grundlagen'.
41 Weyl, Raum-Zeit-Materie, 1st edition, Section 35, and 'Neue Erweiterung'.
42 Pauli, 'Grav. u. Elek.' and Relativity, 202, 205-206. See also A. Einstein, 'Spielen
Gravitationsfelder in Aufbau der materiellen Elementarteilchen eine wesentliche Rolle?',
S.-B. Preuss. Akad. Wiss. (1919), 348-356, translated in Lorentz, Principle of Relativity,
191-198.
43 The positron was yet to be discovered.
44 Pauli, Relativity, 206.
45 I do not use terms such as 'operationalism' to define precise philosophical systems,
for the physicists in general and Pauli in particular did not work out or set down such
precise systems. The terms are rather used in a sense consistent with the views they are
used to describe, as indicating general classes and tendencies of belief only.
46 Mehra, Einstein and Hilbert, 56, dates Einstein's serious attempts at a unified theory
from 1928; but the general aim is clearly apparent in his work from 1907, as he tried
to place both gravitation and light-quanta upon a field-theoretical basis. His desire for
a unified theory is explicit in his letter to Weyl of 27 September 1918, ETHZ 91,
548.
47 Einstein to Born, 27 January 1920, in Born and Einstein, Letters, 20- 23 (Item 13).
48 Einstein, 'Spielen Grav.'.
49 Ibid., 356.
so Einstein to Born, 27 January 1920, and see also Einstein to Born, 3 March 1920,
Born and Einstein, Letters, 20-26 (Items 13, 14).
51 Einstein's programme was presented in A. Einstein, 'Bietet die Feldtheorie Mog-
lichen fUr die Losung des Quantenproblems?', S.-B. Preuss. Akad. Wiss. (1923), 359-
364. The programme was also outlined in Einstein to Besso, 4 January 1924, Einstein
and Besso, Co"espondance, 197 -199 (Item 72), partly translated in Mehra, Einstein
and Hilbert, 80; and in Einstein to Lorentz, 25 December 1923, quoted by Forman,
'Weimar culture', 96. He wrote to Besso that the programme represented a "logical
possibility", but that the mathematics was too difficult for him.
S2 See for example Besso to Einstein, 25 December 1923, Einstein and Besso, Co"es-
pondance, 192-194 (Item 71). When in late 1921 Einstein thought, wrongly, that some
138 NOTES

experiments by Geiger on canal rays offered conclusive proof of the particular nature
of light, he wrote gleefully on this point to Ehrenfest; but writing to Weyl at the same
time he was more concerned with the problems the experiment seemed to propose for
the field-theoretical programme. See Klein, 'First phase', and Einstein to Weyl, 16 and
22 December 1921, ETHZ 91, 552, 553.
S3 Eddington, 'Field and matter' and 'Generalisation ofWeyl's theory'.
54 Ibid.
55 Ibid.; writing to Weyl, however, he accepted that this was unlikely: Eddington
to Weyl, 10 July 1921, ETHZ 91, 525.
S6 Eddington to Weyl, 10 July 1921, ETHZ 91, 525.
57 Weyl to Pauli, 10 May 1919,PB, 3-5 (Item 1).
58 Ibid.
S9 Weyl to Pauli, 9 December 1919, PB, 5-8 (Item 2). This was the basis of Weyl's
acausal manifesto: H. Weyl, 'Das Verhiiltnis der kausalen zur statistischen Betrachtungs-
weise in der Physik', Schweizerische Medizinische Wochenschrift 1 (1920), 737-741.
In the letter Weyl, having discussed the connection between choice of sign of electric
charge and the direction of time, concluded: "In contrast with most physicists, I hold
the essential distinction [between past and future J to be a fact of even more funda-
mental significance than that between positive and negative electricity. Nevertheless,
modern physics may be right in finding no place for 'lawful' or 'field' physics. For I
am completely convinced that the statistics are in principle somewhat independent,
compared with the causality, which is 'lawful'; because it is in general paradoxical to
introduce a continuum as something ready made. I think that field physics actually
plays only the role of the 'world geometry'; in matter there is still something else,
something of reality, which is not causal but which is perhaps to be thought of in terms
of 'independent decisions'. which we take account of in physics through the medium of
statistics.
60 Weyl to Pauli, 9 December 1919, PB, 5-8 (Item 2). This idea was explored in H.
Weyl, 'Feld und Materie', Ann. der Phys. 65 (1921), 541-563, and 'Was ist Materie?',
Naturwissenschaften 12 (1924),561-568,585-593,604-611.
61 Weyl, Space· Time-Matter, 311.
62 Weyl to Pauli, 9 December 1919, PB, 5-8 (Item 2). The context of this observation
was Weyt's reply to Pauli's criticism of the unobservability of the field within an electron.
Weyl argued that internal motions of the electron might well have measurable conse-
quences elsewhere. This did not, however, answer Pauli's point, which was concerned
with the problem of the definition of the field concepts.
63 See Forman, 'Weimar culture' and Hendry, 'Weimar culture'.
64 Weyl, 'Das Verhiiltnis'.
6S See Weyl, Space-Time-Matter, 212-213, and Pauli, Relativity, 192.
66 Pauli, 'Grav. u. Elek.', and Relotivity, 205-206.
67 Einstein had also expressed doubts on the maintenance of energy conservation in
Weyl's theory: Einstein to Weyl, 27 September 1918, ETHZ 91,548.
68 A. Einstein, 'Rapport sur l'etat actuel du probleme des chaleurs specifiques', in P.
Langevin and M. de Broglie (eds.), La Theorie du Rayonnement et les Quanta (Paris,
1912),407-435, and 'Zur Quantentheorie der Strahlung', Phys. Zeit. 18 (1917), 121-
128.
69 Hendry, 'Weimar culture'. Einstein's derivation of Planck's law also appeared on the
 NOTES 139

surface to rest on the particulate nature of light, but it actually entailed several as-
sumptions (including that of the existence of stimulated or induced emission) that
could be understood only in terms of the wave theory. See A. S. Eddington, 'On the
derivation of Planck's law from Einstein's equation', Phil. Mag. 50 (1925), 803-808,
and J. Hendry, 'An investigation of the mathematical formulation of quantum theory
and its physical interpretation, 1900-1927', Ph.D. thesis, London University, 1978,
Appendix F.
70 Hendry, 'Weimar culture', but see also Forman, 'Weimar culture'.
71 W. Schottky, 'Das Kausalproblem der Quantentheorie als eine Grundfrage der
modernen Naturforschung iiberhaupt', Naturwissenschaften 9 (1921), 492-496, 506-
511.
72 They were not at the same institution, but this did not stop them working in close
liaison, as did Weyl and Pauli in later years.
73 K. Forsterling, 'Bohrsches Atommodell und Relativitatstheorie', Zeit. Phys. 3 (1920),
404-407, reviewed by Pauli inPhys. Ber. 2 (1921), 489.
74 E. Schrodinger, 'Dopple~orinzip und Bohrsche Frequenzbedingung', Phys. Zeit. 23
(1922), 301-303.
75 Schrodinger to Pauli, 8 November 1922, PB, 69-73 (Item 29).
76 Ibid.
77 E. Schrodinger, 'Was ist ein Naturgesetz?', Naturwissenschaften 17 (1929), 9-11,
inaugural lecture at Ziirich, December 1922.
78 SchrOdinger to Pauli, 8 November 1922,PB, 69-73 (Item 29).
79 Schrodinger, 'Naturgesetz'.
80 This decision of Schrodinger's remains obscure. He seems to have been working at
the time on both quantum atomic theory and classical relativity theory: E. Schrodinger,
'Die Wasserstoffahnlichen Spektren vom Standpunkte der Polarisierbarkeit des Atom-
rumpfes', Ann. der Phys. 72 (1925), 43-70, and 'Die Erflillbarkeit der Relativitats-
forderung in der klassischen Mechanik', Ann. der Phys. 77 (1925),325-336. The latter
paper was once again connected with Weyl's work, to which it referred. Schrodinger's
theme was that the cosmological problem might be solved through the reconciliation
of general relativity with the Mach principle.
8! E. SChrOdinger, 'Quantisierung als Eigenwertproblem', Ann. der Phys. 79 (1926),
489-527, esp. 489. The words were however omitted from the English translation of
the paper, perhaps as unsuitable for an English audience.
82 E. Schrodinger, 'Ober eine bemerkenswerte Eigenschaft der Quantenbahnen eines
einzelnen Elektrons', Zeit. Phys. 12 (1922), 13-23. See also Raman and Forman, 'Why
SchrOdinger', 305.
83 Schrodinger, ibid., and see Raman and Forman, 'Why SChrOdinger', 303-310.
84 Raman and Forman, ibid.
85 Eddington, Mathematical Theory. This was itself an extended version of the mathe-
matical appendix to the French edition of his more popular Space, Time and Gravitation,
the first edition of which contained the first extended presentation of his philosophical
ideas.
86 Pauli,Relativity, 206.
87 Pauli to Eddington, 20 September 1923, PB, 115-119 (Item 45). This letter was
sent in response to a copy of Eddington's Mathematical Theory.
140 NOTES

CHAPTER 3. NIELS BOHR AND THE PROBLEMS OF

ATOMIC THEORY

1 Pauli's work on atomic theory had been planned for some time. See Schrodinger to
Pauli, 12 July 1920, 13 February 1921, PB, 24-26 (Items 8, 9), and for the context
ibid., pp. 23-24.
2 Serwer, 'Unmechanischer Zwang', 23-24.
3 Interview with J. Franck, SHQP. See Jammer, Conceptual Development, 85, for the
work of Franck and Hertz. Recalled by Heisenberg (interview with Heisenberg, SHQP) as
a "mathematical methods" man, Born seemed more interested in the existence or
otherwise of solutions than in the solutions themselves, and was as happy with negative
as with positive results. He was the ideal man for the proposed investigation.
4 Born to Pauli, 23 December 1919, PB, 9-11 (Item 4): "I was especially interested
by your remark at the end, that you hold the application of the continuum theory to
the inside of the electron to be meaningless, because it is a case there of in principle
non-observable things. I have followed just this line of thought for some time, but so
far without any positive result, namely, that the way out of all the quantum difficulties
over wholly fundamental points must be sought thus: one must not carry over the
concept of space and time as a four-dimensional continuum from the macroscopic
world of experience to the atomic world."
5 Fierz, 'Pauli'.
6 In general Bohr used his assistants as sounding boards for his new ideas, but did not
actually work with them; Pauli was in this an exception. See Serwer, 'Unmechanischer
Zwang', 225.
7 N. Bohr, 'On the constitution of atoms and molecules', Phil. Mag. 26 (1913), 1-25,
476-502,857-875.
8 See Jammer, Conceptual Development, and Heilbron, 'Kossel-Sommerfeld theory'.
9 A. Einstein 'Zur Quantentheorie del Strahlung', Mitt. Phys. Ges. Zurich 18 (1916),
47-62;Phys. Zeit. 18 (1917),121-128, translated in SQM, 63-78.
10 N. Bohr, On the Quantum Theory 0/ Line Spectra (Copenhagen, 1918). The ter-
minology was introduced in N. Bohr, 'Ober die Linienspektren der Elemente', Zeit.
Phys. 2 (1920),423-469, translated in Bohr, Theory o/Spectra, 20-60. It is sometimes
aruged that the correspondence principle should be dated to 1913, but this is to miss
its whole point. Bohr did talk of a correspondence with the classical theory in 1913,
but this was no more than that correspondence which relates almost any theory to its
predecessors, in the domain in which they represent valid approximations. It was not
until 1918 that Bohr talked of a correspondence between quantum and classical theories
in which the latter formed an integral part of the foundations of the former, as discussed
below.
11 Bohr considered a system with one degree of freedom, and a transition between two
states of this system with high quantum numbers and close frequencies. For two states
(n', w') and (n", w") whose energies were E' = hn' w', E" = hn" w" (for the Planck
oscillator, w' = w", but this does not hold in general), he considered the limit "where
n is very large, and where the ratio between the frequencies of the motion in successive
stationary states differs very little from unity", in fact requiring n' - n" « n', with
w' /w" -+ 1 faster than n -+ infinity. In this case the frequency of emitted or absorbed
radiation was given by v = (n' - n") w, where w "" w', w", and Bohr compared this
 NOTES 141

with the classical Fourier expression for particle displacement, l:~ cos 2'/1'(Twt + cT).
He linked the classical harmonic given by To with the transitioIf To = (n' - n"), and
reasonably expected the probability of a quantum emission of frequency ToW to be the
same as the intensity of a classical emission of the same frequency, namely CT • Finally,
he suggested that this relationship should hold in some (vague) way for small quantum
numbers too, despite the fact that the frequencies could no longer be expected to
correlate there (w' /: wIt) as they had done, in practice, for the higher states.
12 See A. S. Eddington, 'On the derivation of Planck's law from Einstein's equation',
Phil. Mag. 50 (1925), 803-808.
13 Heilbron and Kuhn, 'Genesis'.
14 See Note 10 above.
15 Bohr, 'Linienspektren'.
16 Bohr to Darwin, catalogued July 1919, more likely 1920, BSC 1, 4.
17 N. Bohr, 'L'application de la theorie des quanta aux problemes atomiques', in Insti-
tut International de Physique Solvay, Atomes et Electrons (Paris, 1921), 228-247;
original English draft in Bohr, Works, Vol. 3,364-380.
18 N. Bohr, 'On the application of the quantum theory to atomic structure: Part I.
The fundamental postulates', Supplement to Proc. Camb. Phil. Soc. (1924), p. 35.
This paper is a translation from Zeit. Phys. 13 (1923), 117-165, submitted November
1922.
19 Sommerfeld to Bohr, 10 May 1918, BSC 7,3.
20 See for example Stuewer, Compton Effect, for some strong views on duality ex-
pressed by physicists not closely concerned with the problem. See also Hendry, 'Wave-
particle duality'.
21 Bohr, 'Fundamental postulates', 35, 1.
22 Bohr to H~ffding, 22 December 1923, BSC, as quoted by Honner, 'Transcendental
philosophy', 7.
23 A. Einstein, 'Rapport sur l'etat actuel du probleme des chaleurs specifiques', in P.
Langevin and M. de Broglie (eds.), La Theorie du Rayonnement et les Quanta (Paris,
1912), 407-435, esp. 429.
24 Ibid., 429.
25 Einstein to Ehrenfest, 31 May 1924 and see also 12 July 1924, and Einstein to
Born, 29 April 1924, all of which are discussed with extracts in Klein, 'First phase',
32-35.
26 Einstein, 'Rapport', 443.
27 See Hendry, 'Weimar culture'.
28 Darwin to Bohr, 20 July 1919, BSC I, 4.
29 Bohr to Darwin, catalogued July 1919, BSC 1, 4.
30 Ibid.
31 Bohr, 'L'application', 374.
32 N. Bohr, 'Application of the quantum theory to atomic problems in general', in his
Works, Vol. 3, 397-414, esp. 413.
33 Bohr, 'Fundamental postulates', 40.
34 See Klein, 'First phase', and Chapter 5 below. Einstein almost always linked the two
issues together.
35 M. Planck, 'Ober die Begriindung des Gesetzes der schwarzen Strahlung', Ann. der
Phys. 37 (1912), 642-656, esp. 644.
142 NOTES

36 Forman, 'Weimar culture'.

37 Bohr, Atomic Theory and the Description of Nature, 91. For a wide-ranging dis-
cussion see Meyer-Abich, Ko"espondenz.
38 See Jammer, Conceptual Development, 84.
39 For example, H. Hqlffding, History of Modern Philosophy, Vol. 2 (London, 1900),
286.
40 See for example O. W. Richardson, The Electron Theory of Matter (Cambridge,
1916),507.
41 Bohr, 'Fundamental postulates', 21.
42 N. Bohr, 'Problems of the atomic theory', in his Works, Vol. 3,569-574.
43 Darwin, manuscript draft of July 1919, SHQP 36, 3.
44 R. von Mises, 'Uber die gegenwlirtige Krise der Mechanik', Zeit. Ang. Math. u. Mech.
1 (1921), 425 -431, and Naturwissenschaft und Technik der Gegenwart (Leipzig, 1922).
See Hendry, 'Weimar culture', 81.
4S H. A. Senftle ben, 'Zur Grundlagen der Quantentheorie', Zeit. Phys. 22 (1923), 105-
156, esp. 127.
46 Bohr, 'Fundamental postulates', 42.
47 W. Pauli, 'Uber das thermische Gleichgewicht zwischen Strahlung und freien Elek-
tronen', Zeit. Phys. 18 (1923) 272-286; review of K. Forsterling, 'Bohrsches Atom-
modell und Relativitiitstheorie', Phys. Ber. 2 (1921), 489.
48 See Schrodinger to Pauli, 8 November 1922, PB, 69-73 (Item 29).

CHAPTER 4. THE TECHNICAL PROBLEM COMPLEX

1 Forman, 'Weimar culture'.
2 Hendry, 'Weimar culture'.
3 Ibid.
4 Interview with Heisenberg, SHQP. The reference was to physicists at Munich (under
Sommerfeld) and at Gottingen (under Born and Franck).
5 A. Sommerfeld, Atombau und Spektrallinien (Braunschweig, 1922), translated as
Atomic Structure and Spectral Lines (London, 1923), 253. See also Heisenberg to
Lande, 28 November 1921, SHQP 6, 2.
6 A. Einstein and P. Eluenfest, 'Quantentheoretische Bemerkung zum Experiment
von Stern und Gerlach', Zeit. Phys. 11 (1922),31-34.
7 M. Born and W. Heisenberg, 'Die Elektronenbahnen im angeregten Heliumatom',
Zeit. Phys. 16 (1923), 229-243.
8 Heisenberg to Bohr, 2 February 1923, BSC.
9 Heisenberg to Pauli,19 February 1923,PB, 79-81 (Item 31).
10 A. Lande, 'Schwierigkeiten in der Quantentheorie des Atombaues, besonders
magnetischer Art', Phys. Zeit. 24 (1923), 441-444. M. Born, 'Quantentheorie und
Storungstheorie', Naturwissenschaften 11 (1923), 537 -542. See also F. Paschen, 'Die
spektroscopische Erforschung des Atombaus', Phys. Zeit. 24 (1923), 401-407, who
wrote that " the present contradiction must be augmented by further incomprehensible
problems".
11 Born, 'Quantentheorie', 542.
12 W. Heisenberg, 'Zur Quantentheorie der Linienstruktur und der anomalen Zee-
maneffekte', Zeit. Phys. 8 (1922), 273-297, esp. 281.
 NOTES 143

13 Sommerfeld to Einstein, 11 January 1922, Einstein and Sommerfeld, Briefwechsel,

95-97 (Item 40).
14 Einstein did try to set up a decisive experiment, but failed: see Klein, 'First phase'.
15 A. H. Compton, 'Secondary radiation produced by X-rays, and some of their appli-
cations to physical problems', Bull. Nat. Res. Counc. 4 (1922), No. 20; 'A quantum
theory of the scattering of X-rays by light elements', Phys. Rev. 21 (1923),483-502;
'Wavelength measurements of scattered X-rays', ibid., 715; The scattering of X-rays',
J. Franklin [nst. 198 (1924), 61-72. In the Compton effect, X-ray scattering was
interpreted as a collision process between individual particulate light-quanta and in-
dividual electrons in the scattering substance. This gave the correct change in wavelength
as a function of scattering angle, which change could not easily be interpreted classically,
on the wave theory of light. For further details see Jammer, Conceptual Development,
157 ff., and Stuewer, Compton Effect. As has been noted the Compton effect does not
in general seem to have had a decisive effect on physicists' views of the nature of light,
but one exception may be Sommerfeld, whose response was reported by Compton,
'The scattering of X-rays', 69: "In a recent letter to me Sommerfeld has expressed the
opinion that the discovery of the change of wavelength of radiation, due to scattering,
sounds the death-knell for the wave theory of radiation." It is unclear whether Compton
noted the irony.
16 P. Debye, 'Zerstreuung von Rontgenstrahlen und Quantentheorie', Phys. Zeit. 24
(1923),161-166; W. Duane, 'The transfer in quanta of radiation momentum to matter',
Proc. Nat. Acad. Sci. 9 (1923), 158-164; A. H. Compton, 'The quantum integral and
diffraction by a crystal', ibid., 359-362.
17 W. Pauli, 'tiber das thermische Gleichgewicht zwischen Strahlung und freien Elek-
tronen', Zeit. Phys. 18 (1923), 272-286; see also A. Einstein and P. Ehrenfest, 'Zur
Quantentheorie des Strahlungsgleichgewichts', Zeit. Phys. 19 (1923), 301-306. While
disagreeing with Einstein's ultimate goal of a pure field theory, Pauli agreed with him
completely on the necessity of the light-quantum concept.
18 P. S. Epstein and P. Ehrenfest, 'The quantum theory of the Fraunhofer diffraction',
Proc. Nat. A cad. Sci. 10 (1924), 133-139, and 'Remarks on the quantum theory of
diffraction', ibid., 13 (1927),400-408; A. H. Compton, 'The total reflexion of X-rays',
Phil. Mag. 45 (1923), 1121-1131, esp. 1130.
19 Sommerfeld introduced the general conditions fPkdqk = nkh: A. Sommerfeld,
'Zur Quantentheorie der Spektrallinien', Ann. der Phys. 52 (1916), 1-94, 125-167.
These conditions were then expressed in terms of the classical Hamilton-Jacobi theory,
the action variable being defined as in the main text, by Epstein and Schwarzchild: K.
Schwarzchild, 'Zur Quantenhypothese', S.·B. Preuss. Akad. Wiss. (1916), 548-568,
and P. S. Epstein, 'Zur Quantentheorie',Ann. der Phys. 52 (1916), 168-188.
20 See N. Bohr, On the Quantum Theory of Line Spectra (Copenhagen, 1918); A.
Sommerfeld, Atombau und Spektrallinien (Braunschweig, 1919-1920). Also essential
to the theory was Ehrenfest's adiabatic principle: P. Ehrenfest, 'Adiabatic invariants
and the theory of quanta', Phil. Mag. 33 (1917), 500-513. See Jammer, Conceptual
Development, 98, and Klein, Ehrenfest, 264ff.
21 J. H. van Vleck, 'The normal helium atom and its relation to the quantum theory',
Phil. Mag. 44 (1923),842-869, and 'The dilemma of the helium atom', Phys. Rev. 19
(1922), 419-420; H. A. Kramers, 'Uber das Modell des Heliumatom', Zeit. Phys. 13
(1923), 312-341; I. Langmuir, 'The structure of the helium atom', Phys. Rev. 17
144 NOTES

(1921), 339-353; P. S. Epstein, 'Problems of the quantum theory in the light of the
theory of perturbations', Phys. Rev. 19 (1922), 578-608, completed September 1921.
For a detailed treatment of all this work see Small, 'The helium atom'.
22 M. Born and W. Pauli, 'tiber die Quantelung gestorten mechanischer Systeme', Zeit.
Phys. 10 (1922), 137-158. See also M. Born and E. Brody, 'tiber die Schwingungen
einer mechanischer System mit endlicher Amplitude und ihre Quantelung', ibid., 6
(1921), 140-152.
23 Born and Pauli, 'Quantelung'; Born to Einstein, 21 October 1921, Born and Einstein,
Letters, 57-59 (Item 33).
24 W. Pauli, 'tiber das Modell des Wasserstoffmolekulions', Ann. der Phys. 68 (1922),
177-240.
2S W. Heisenberg, 'Linienstruktur u. anomalen Zeemaneffekte'; A. Lande, 'Zur Theorie
der anomalen Zeeman- und magnetomechanischen Effekte', Zeit. Phys. 11 (1922),
353-363.
26 See Forman, 'Lande', and Cassidy, 'Werner Heisenberg'.
27 Ibid., and see also Serwer, 'Unmechanischer Zwang'. To comprehend the magnitude
of the problem faced we should remember that they were working without either the
spin or the parity concept.
28 Bohr to Lande, 15 May 1922, quoted in Serwer, 'Unmechanischer Zwang', 224-225.
29 Ibid.
30 M. Born and W. Heisenberg, 'tiber Phasenbeziehungen bei den Bohrschen Modellen
von Atomen und Molekeln', Zeit. Phys. 14 (1923),44-55.
31 Born and Heisenberg, 'Elektronenbahnen', 229.
32 Interview with Heisenberg, SHQP. See also Heisenberg to Pauli, 26 March 1923,
PB 85-86 (Item 34); "Basically we [Heisenberg and Born) are now both of the convic-
tion that all helium models so far are just as erroneous as the whole of atomic physics."
33 N. Bohr, 'On the application of the quantum theory to atomic structure. Part I.
The fundamental postulates', Supplement to Proc. Comb. Phil. Soc. (1924), especially
32-33.
34 Ibid., 33-34.
3S Bohr to Lande, 3 March 1923, SHQP 4, 1: "It was, as. you saw, a desperate attempt
to stick with integral quantum numbers, because we hoped to see, even in the paradoxes
themselves, a hint of the paths upon which one might seek the solution of the anomalous
Zeeman effect." ("Es war, wie Sie gesehen haben, ein Verzweiflungsversuch, den ganzen
Quarttenzahlen treu zu bleiben, indem wir hofften, eben in den Paradox en einen Finger-
zeig zu se~en f1ir die Wege, auf denen man die Losung des annomalen Zeemaneffektes
suchen diirfte.")
36 Heisenberg to Lande, 13 November 1922, quoted by Serwer, 'Unmechanischer
Zwang', 210-211.
37 Ibid.
38 Ibid.
39 A. Lande, 'Termstruktur und Zeemaneffekt der Multipletts', Zeit. Phys. 15 (1923),
189-205.
40 N. Bohr, 'Der Bau der Atome und die physikalischen und chemischen Eigenschaften
der Elemente', Zeit. Phys. 9 (1922), 1-67. See Kragh, 'Bohr's second theory'.
41 Bohr, 'Fundamental postulates', 16. See Note 20 above.
42 Bohr, Quantum Theory of Line Spectra.
 NOTES 145

43 Lande, 'Theorie der anomalen Zeeman- und magnetomechanischen Effekte', 361.

44 N. Bohr, manuscript printed in his Works, Vol. 3, 502-531, translation 532-565,
esp. 558. See also N. Bohr, 'Linien-spektren und Atombau', Ann. derPhys. 71 (1923),
228-288.
45 Pauli to Bohr, 11 February 1924,PB, 143-145 (Item 54).
46 See for example Bohr to Lande, 15 May 1922, quoted by Serwer, 'Unmechanischer
Zwang', 224-225.
47 Pauli to Lande, 23 May 1923,PB, 87-90 (Item 35).
48 Pauli to Sommerfeld, 6 June 1923, PB, 94-101 (Item 37).
49 Ibid.
50 Pauli to Bohr, 16 July 1923,PB, 102-105 (Item 39).
51 Pauli, 'Thermische Gleichgewicht'.
52 Pauli to Eddington, 20 September 1923, PB, 115-119 (Item 45).
53 Pauli to Bohr, 21 February 1924,PB, 147-149 (Item 56).
S4 Heisenberg to Pauli, 19 February 1923, PB, 79-81 (Item 31).
55 Heisenberg to Pauli, 9 October 1923, PB, 125-128 (Item 47). Heisenberg to Bohr,
22 December 1923, BSC. W. Heisenberg, 'Dber eine Abanderung der formalen Regeln
der Quantentheorie beim Problem der anomalen Zeemaneffekte', Zeit. Phys. 26 (1924),
291-307; and see Serwer, 'Unmechanischer Zwang', 213-218. The formalism as first
expressed in the letter to Pauli did not include the integration limits, but these could
be inferred.
S6 Serwer, 'Unmechanischer Zwang', 215-216.
57 Heisenberg to Pauli, 9 October 1923,PB, 125-128 (Item 47).
58 Ibid.
s9 Ibid.
60 Interview with Heisenberg, SHQP. The seminar was given by Courant and Siegel,
and would probably have been attended, as was normal, by all those in Born's depart-
ment as well as by those in Hilbert's.
61 Lande, "Termstruktur".
62 Bohr to Heisenberg, 31 January 1924, BSC.
63 Heisenberg to Pauli, 9 October 1923, PB, Item 47: ''The considerable negative aspect
of the theory is that we now understand the quantum theory no more whatsoever.
But that appeals very much to me. The proper aim must now be to reach the discrete
states unequivocally from the Symbol-model; whether the formulae thus attained will
approach a comprehensible meaning I doubt."
64 Pauli to Lande, 14 December 1923, PB, 134 (Item 51).
6S A. Lande and W. Heisenberg, 'Termstruktur der Multipletts hoherer Stufe', Zeit.
Phys. 25 (1924), 279-286. See Heisenberg to Bohr, 3 February 1924, BSC, for his
hopes, and Serwer, 'Unmechanischer Zwang', for discussion.
66 R. Ladenberg, 'Die quantentheoretische Zahl der Dispersionselektronen', Zeit. Phys.
4 (1921), 451-471, translated in SQM, 139-158. See also N. Bohr 'L'application de la
theorie des quanta aux problemes atomiques', in Institut International de Physique
Solvay, A tomes et Electrons (Paris, 1921), 228-247, and original English draft in Bohr,
Works, Vol. 3, 364-380.
67 R. Ladenburg and F. Reiche, 'Absorption, Zerstreuung und Dispersion in der Bohr-
sche Atomtheorie', Naturwissenschaften J 1 (1923), 584-598, esp. 597.
68 Bohr, 'Application to problems in general', 414.
146 NOTES

69 Bohr to Darwin, 21 December 1922,BSC.

70 Bohr, 'Fundamental postulates', 38.
71 H. A. Kramers, 'On the theory of X-ray absorption and the continuous X-ray spec-
trum', Phil. Mag. 46 (1923), 836-871, esp. 861. This work was partly a response to
Eddington's theory of X-ray capture as a means of absorption of radiation in stars. See
A. S. Eddington, 'On the absorption of radiation inside a star', Month. Not. Roy. Astr.
Soc. 83 (1922), 32-46, 84 (1923), 104-123; and 'Das Strahlungsgleichgewicht der
Sterne', Zeit. Phys. 7 (1921), 351.
72 Kramers, 'X-ray absorption', 843.
73 Ibid., 852.
74 See below, this chapter.
75 N. Bohr, 'Problems of the atomic theory', in his Works, Vol. 3,569-574.
76 Ibid., 571.
77 Pauli to Sommerfeld, 6 June 1923, PB, 94-101 (Item 37). Pauli added that Epstein
had already made a similar remark, but I have been unable to trace this. Pauli's remark
may have been linked with Born's work at the time, for Born ('Quantentheorie') ran
into considerable difficulties with high-frequency electron-electron coupling in the atom.
78 Pauli to Bohr, 21 February 1924, PB, 147-149 (Item 56).
79 Heisenberg, Physics and Beyond, 36.
80 Heisenberg to Pauli, 9 October 1923, PB, 125-128 (Item 47).
81 A. Sommerfeld and W. Heisenberg, 'Eine Bemerkung iiber relativistische Ront-
gendubletts und Linienscharfe', Zeit. Phys. 10 (1922), 393-398, esp. 398.
82 Heisenberg to Lande, 29 October 1921, SHQP 6,2.
83 See for example A. Sommerfeld and W. Heisenberg, 'Die Intensitat der Mehrfachlinien
und ihre Zeemankomponenten', Zeit. Phys. 11 (1922), 131-154, esp. 132.
84 Pauli to Bohr, 11 February 1924,PB, 143-145 (Item 54).

CHAPTER 5. FROM BOHR'S VIRTUAL OSCILLATORS TO THE

NEW KINEMATICS OF HEISENBERG AND PAULI
1 N. Bohr, H. A. Kramers and J. C. Slater, 'The quantum theory of radiation', Phil.
Mag. 47 (1924),785-802, reprinted in SQM, 159-176.
2 This seems to me to be the true value of the virtual oscillator treatment, but Jammer,
Conceptual Development, 181-195, emphasises the break with classical causality and
conservation laws. See also Klein, 'First phase', 23-29;SQM, 11-14; and K. R. Popper,
Objective Knowledge (London, 1970), 206. For a treatment similar to that given here
but self-contained see Hendry, 'Bohr-Kramers-Slater'.
3 The habitual reference to "the present state of the theory" is omitted, but compare
the statement given be: ~w (Note 27) with Bohr's earlier statement in N. Bohr, 'On
the application of the quantum theory to atomic structure. Part I. The fundamental
postulates', Supplement to hoc. Comb. Phil. Soc. (1924), 20: "According to this method
of treatment, we do not seek a cause for the occurrence of radiative processes, but we
simply assume that they are governed by the laws of probability."
4 See below.
5 L. de Broglie, 'Sur la degradation du quantum dans les transformations successives
des radiations de haute frequence', Comptes Rendus 173 (1921), 1160-1162. See also
A. Einstein, 'Ober einen die Erzeugung und Verwandlung des Lichtesbetreffenden
 NOTES 147

heurlstischen Gesichtspunkt', Ann. der Phys. 17 (190.5), 132-148, and M. de Broglie,

'La relation hv = € dans les phenomenes photetectriques: production de la lumiere dans
Ie choc des atomes par les etectrons et production des rayons de Rontgen', in Institut
International de Physique Solvay,Atomes et electrons (Paris, 1921), 80.-119.
6 L. de Broglie, -'Rayonnement noir et quanta des lumiere', 1. de Phys. 3 (1922), 422-
428, translated in de Broglie and Brillouin, Selected Papers, 1-8.
7 Ibid., 1. See also L. de Broglie, 'Ondes et quanta', Comptes Rendus 177 (1923),
50.7-510.; 'Quanta de lumiere', ibid., 548-550.; 'Les quanta, la theorie cinetique des
gaz et Ie principe de Fermat', ibid., 630.-632; 'A tentative theory of light quanta',
Phil. Mag. 47 (1924), 446-458.
8 The reception of de Broglie's ideas is analysed in Raman and Forman, 'Why Schrti-
dinger' but see also discussion of Born's response below.
9 de Broglie, 'Tentative theory'.
10 Slater to Kramers, 8 December 1923, SHQP 8,10..
11 Interview with J. C. Slater, SHQP. In the classical theory the sharpness of a spectral
line was related to the period over which a wave was emitted, the wave undergoing a
gradual change of frequency as the electron emitting it spiralled inwards.
12 Ibid.
13 Slater to Kramers, SHQP 8, 10.
14 Ibid.
15 See Chapter 4 above, and Stuewer, Compton Effect.
16 H. A. Lorentz, lectures given in California in 1923 and printed in his Problems of
Modem Physics (New York, 1927), especially pp. 150. ff.
17 Slater's conception is closest to that of de Broglie, who was the only writer of any
importance to suggest quanta moving at less than the traditional speed of light. de
Broglie's work appeared in the Comptes Rendus over 1923, and the English version was
communicated to the Philosophical Magazine, for publication in February 1924, by
Slater's host Fowler, so that some influence on Slater would seem to be quite probable.
On the other hand, the idea of light-quanta guided by a system of ghost waves was
very commonplace, and Slater's conversion to the light-quantum concept could also
have been related to the Compton effect results of the previous year. In his interview,
given in 1963, Slater denied the influence of de Broglie, and said that he got nothing
from Fowler but politeness. Recounting his idea in 1925, however, he wrote that "the
theory in this form was developed in England, under the guidance of Mr R. H. Fowler,
to whom my sincerest thanks are due'. (J. C. Slater/The nature of radiation', Nature
116 (1925), 278); and Fowler wrote to Bohr on 14 January 1924, BSC, that "I thought
he [Slater) was on sound lines and I encouraged him as much as I could". Slater's
interview was given only reluctantly, and revealed a marked hostility towards Bohr,
whose vague "handwaving" clearly repelled him. The Bohr-Kramers-Slater episode had
hurt him considerably, and this may have affected his recollections. The question as to
his sources if any must remain an open one.
18 Slater to Kramers, 8 December 1923, SHQP 8, 10.. Bohr's 'Fundamental postulates'
had also been communicated, to the Cambridge Philosophical Society, by Fowler in
1923.
19 Interview with Slater, SHQP.
20 The Iight-quanta were emitted instantaneously, but they were governed by the
waves, which were not.
148 NOTES

21 See Note 17 above.

22 Kemble to Bohr, 4 January 1924, BSC.
23 Kramers and Holst, The Atom, 175. Kramers's views and abilities are difficult to
assess. He was obviously a fust rate physicist, and a philosophical one at that, but on
matters of interpretation he often comes across, perhaps unfairly, as naive. See forth-
coming papers by H. Radder in History of Science and Janus for a more generous view
of Kramers than that taken here.
24 J. C. Slater, 'Radiation and atoms', Nature 113 (1924). 307: "But when the idea
with this [the light-quantum] interpretation was described to Dr. Kramers, he pointed
out that it scarcely suggested the defmite coupling between emission and absorption
processes which light-quanta would provide." The same account is repeated in Slater,
'Nature of radiation'.
25 Slater, 'Radiation and atoms', 307.
26 Interview with Slater, SHQP; Slater to van Vleck, 27 July 1924, SHQP 49, 14, in
which the Bohr-Kramers-Slater theory is clearly rejected.
27 Bohr, Kramers and Slater, 'Quantum theory of radiation', 796.
28 Interview with Slater, SHQP.
29 A. H. Compton, 'The scattering of X-rays', J. Franklin Inst. 198 (1924), 61-72,
esp. 69, where Sommerfeld's views are also quoted.
30 Bohr, Kramers and Slater, 'Quantum theory of radiation', 799.
31 Note 29 above.
32 E. C. Stoner, 'The structure of radiation', Proc. Camb. Phil. Soc. 22 (1925), 577-
594, esp. 592, and see also Note 1 to Chapter 2 above.
33 Einstein to Ehrenfest, 31 May 1924 and see also 12 July 1924 and Einstein to Born,
29 April 1924, all discussed with quotations in Klein, 'First phase', 32-35.
34 Ehrenfest to Einstein, 9 January 1925, ibid., 31.
35 Pauli to Bohr, 2 October 1924, PB, 163-167 (Item 66).
36 Pauli to Bohr, 21 February 1924, PB, 147-149 (Item 56). See also Heisenberg to
Bohr, 8 January 1925, BSC: "[Pauli] does not believe in the virtual oscillators and
denounces the virtualisation of physics. It is not clear to me what he means by this." "Er
glaube ... nicht aber an virtuelle Oszillatoren und schimpft uber die 'Virtualisierung'
der Physik. Mir ist nicht klar, was es damit meint.")
37 Pauli to Bohr, 2 October 1924, PB, 163-167 (Item 66). See also Pauli to Sommer-
feld, November 1924,PB, 173-176 (Item 70).
38 Schrodinger's position is clearest in his letter to Pauli on 8 November 1922, PB
69- 73 (Item 29), for which see Chapter 2 above. For his reaction to Bohr-Kramers-
Slater see SchrOdinger to Bohr, 24 May 1924, BSC, and E. SchrOdinger, 'Bohr's neue
Strahlungshypothese und der Energiesatz', Naturwissenschaften 12 (1924), 720-724.
39 H. A. Kramers, The law of dispersion and Bohr's theory of spectra', Nature 113
(1924), 673-674; 'The quantum theory of dispersion', Nature 114 (1924), 310 (in
response to Breit's criticisms); H. A. Kramers and W. Heisenberg, 'Uher die Streuung
von Strahlen durch Atome', Zeit. Phys. 31 (1925),681-707.
40 Jordan's first published research was a critique of Einstein's 1916 demonstration of
light-quanta (P. Jordan, 'Zur Theorie der Quantenstrahlung', Zeit. Phys. 30 (1924),
297-319, refuted by A. Einstein, 'Bemerkung zu P. Jordans Abhandlung ... " ibid.,
31 (1925), 784-785). During the remainder of the decade his work showed a strong
preference for wave rather than particle formulations, but his explicit discussions of
physical issues remained strictly neutral.
 NOTES 149

41 R. H. Fowler, manuscript notes taken by Dirac on a lecture course 'Recent Develop-

ments' (1925), SHQP 36, 8. R. Becker, 'tiber Absorption und Dispersion in Bohr's
Quantentheorie', Zeit. PhY!1. 27 (1924),173-188.
42 M. Born, 'Ober Quantenmechanik', Zeit. Phy!1. 26 (1924), 379-395, translated in
SQM, 181-198, esp. 189.
43 Heisenberg to Pauli, 8 June 1924,PB, 154-156 (Item 62).
44 Heisenberg to Pauli, 4 March 1924,PB, 149-150 (Item 57).
45 Ladenburg to Kramers, 31 May 1924, SHQP 8,9.
46 In a further letter of 8 June 1924, SHQP 8, 9, Ladenburg wrote to Kramers of
Einstein's reaction to the oscillator theory that "Seine Meinung was entschrieden nicht
ungiinstig". If anything this would support the view that Ladenburg was again talking
of the technique rather than of the interpretation, but it is much more likely to be a
slip of the pen: there is already one slip in "entschrieden" for "entschieden", and the
inclusion of "nicht" may well be another. In neither letter is the interpretation referred
to explicitly.
47 C. Ramsauer, 'Ober der Wirkungsquerschnitt der Gasmolekule gegeniiber langsamen
Elektronen', Ann. der PhY!1. 64 (1921), 513-540, 66 (1922), 546-558, and 72 (1923),
345-352.
48 Geiger to Bohr, 17 April 1925, BSC; W. Bothe and H. Geiger, 'Ober das Wesen des
Comptoneffektes; ein experimenteller Beitrag zur Theorie der Strahlung', Zeit. PhY!1. 32
(1925), 639-663, and 'Experimentelles zur Theorie von Bohr, Kramers und Slater',
Naturwis!1emchaften 13 (1925), 440-441. See also A. H. Compton and A. W. Simon,
'Directed quanta of scattered X-rays', Phys. Rev. 26 (1925), 289-299.
49 Bohr to Heisenberg, 18 April 1925, BSC, written before the letter from Geiger had
been recieved. Bohr wrote that he was preparing himself for Geiger's results, which he
clearly thought would go against his interpretation. Writing to Franck on 21 April 1925,
BSC, he wrote that he had long been worried by Ramsauer's resuits, and he referred
to a much earlier letter, now lost, in which he had discussed these.
50 Born to Bohr, 24 April 1925,BSC.
51 Bohr to Geiger, 21 April 1925, Bohr to Franck, 21 April 1925, and Bohr to Born,
1 May 1925, BSC. See also Stuewer, Compton Effect, 301.
52 Pauli to Kramers, 27 July 1925,PB, 232-235 (Item 97).
53 Interview with J. C. Slater, SHQP.
54 Kramers, 'Law of dispersion'.
55 Interview with W. Heisenberg, SHQP.
56 Kramers, 'Quantum theory of dispersion' and Born, 'Quantenmechanik'. Kramers
wrote that his formula was derived by substituting a quantum expression for the classical
dispersion form, but the substitution given was far from obvious. It could only have been
derived from physical considerations or from Born's general rules, and even in the former
case a knowledge of Born's work would seem to have been necessary to get the exact
form given by Kramers. The timing of the papers allows plenty of scope for this, and
it would indeed be remarkable if Kramers were not aware of Born's work when he
completed his own.
57 See Chapter 4 above.
58 Heisenberg to Pauli, 9 October 1923, PB, 125 -128 (Item 47): see Note 63 to Chapter
4 above.
59 That the theory had such implications was recognised by Heisenberg in his letter to
Bohr of 22 December 1923,BSC.
150 NOTES

60 Heisenberg to Pauli, 8 June 1924,PB, 154-156 (Item 62).

61 Interview with W. Heisenberg, SHQP, and Born,My Life, 216.
62 Born, 'Quantenmechanik'; W. Heisenberg, 'tiber eine Abiinderung der formalen
Regeln der Quantentheorie beim Problem der anomalen Zeemaneffekte', Zeit. Phys. 26
(1924),291-307.
63 Born, 'Quantenmechanik': SQM, 182.
64 It is an historian's job to speculate to some extent, but when a new theory arises
from group discussions, or from ideas already 'in the air', such speculation tends to
distort the reality. This seems to have happened, for example, in respect of the history
of the spin concept, when van der Waerden wrote an excellent detailed study, only to
be censured, also with good reason, by two of the main participants. Goudsrnit and
Uhlenbeck argued that the history had been distorted by overemphasis on "rnicro-
history" and consequent loss of its "irrational" part, that the ideas were in the air
and that it did not really matter who published what or when. See S. Goudsrnit and
G. Uhlenbeck, manuscripts on the advent of spin, SHQP, and Waerden, 'Exclusion
principle'. The present situation was described by Born as a groping towards the new
theory by his whole department jointly.
6S All Born's work on the quantum theory had been conducted through the medium
of perturbation theory, and he took the same approach to the new quantum mechanics
in 1925-1926.
66 Born, 'Quantenmechanik': SQM, 181.
67 Ibid., 182.
68 M. Born, 'Quantentheorie und Storungsrechnung', Naturwissenschaften 11 (1923),
537-542.
69 Born, 'Quantenmechanik': SQM, 193; he also wrote that "the two kinds of resonator
have a different behaviour".
70 Ibid., 190.
71 Kramers and Heisenberg, 'Streuung von Strahlen'.
72 Interview with W. Heisenberg, SHQP. Smekal's treatment was less general than that
of Kramers and Heisenberg, but more so than Kramers's earlier one, and it is somewhat
surprising that Kramers had not apparently read it, or chose not to cite it, when he
published the latter: Kramers, 'Law of dispersion', and A. Smekal, 'Zur Quantentheorie
der Dispersion', Naturwissenschaften 11 (1923),873-875.
73 Interview with W. Heisenberg, SHQP.
74 The work is discussed thoroughly by MacKinnon, 'Heisenberg', but he seems to me
to miss the point rather in treating the virtual oscillator approach as here entailing a
physical model. Given that it did not do so then the otherwise contradictory treatments
of MacKinnon and Serwer (,Unmechanischer Zwang', especially 221) become more or
less consistent.
75 R. Wood and A. Ellett, 'On the influence of magnetic fields on the polarisation of
resonance radiation', Proc. Roy. Soc. A103 (1923), 396-403; P. D. Foot, A. E. Ruark
and F. L. Mohler, 'The D2 Zeeman pattern for resonance radiation', J. Opt. Soc. America
7 (1923), 415-418.
76 N. Bohr, 'ZUI Polarisation des Fluorescenzlichtes', Naturwissenschaften 12 (1924),
1115-1117; W. Heisenberg, 'tiber eine Anwendung des Korrespondenzprinzips auf die
Frage nach der Polarisation des Fluorescenzlichtes',Zeit. Phys. 31 (1925),617-626.
77 Bohr, 'Polarisation', 1115.
78 H. Burger and H. Dorgelo, 'Beziehung zwischen inneren Quantenzahlen und Inten-
 NOTES 151

sitaten von Mehrfachlinien', Zeit. Phys. 23 (1924), 258-266; L. Ornstein and H. Burger,
'Strahlungsgesetz und Intensitat der Komponenten im Zeemaneffekt', ibid., 28 (1924),
135-141, and 29 (1924), 241-242.
79 Interview with W. Heisenberg, SHQP, and see MacKinnon, 'Heisenberg', 154.
80 Heisenberg, 'Polarisation', 621.
81 Ibid., 617.
82 M. Born and W. Heisenberg, 'Ober den Einfluss der Deformierbarkeit der lonen auf
optische und chemische Konstanter, 1', Zeit. Phys. 23 (1924), 38,8-410. A. Lande,
'Zur Struktur des Neonspektrums', ibid., 17 (1923), 292-294, and 'Das Wesen der
relativistisch Rontgendubletts', ibid., 24 (1924), 88-97; A. Lande and W. Heisenberg,
'Termstruktur der Multipletts hoherer Stufe', ibid., 25 (1924), 279-286.
83 W. Pauli, 'tiber den Einfluss der Geschwindigkeitsabhiingigkeit der Elektronenmasse
auf den Zeemaneffekt', ibid., 31 (1925), 373-385.
84 W. Pauli, 'tiber den Zusammenhang des Abschlusses der Elektronengruppen in Atom
mit der Komplexstruktur der Spektren', ibid., 32 (1925), 765-783; the Zweideutigkeit
is introduced in this sense in Pauli, 'Einfluss der Geschwindigkeitsabhiingigkeit', 385.
85 E. C. Stoner, 'On the distribution of electrons among atomic levels', Phil. Mag. 48
(1924},719-736.
86 Pauli, 'Zusammenhang'.
87 Pauli to Bohr, 12 and 31 December 1924,PB, 186-189, 197-199 (Items 74, 79).
88 Pauli to Bohr, 31 December 1924, PB, 197-199 (Item 79). He suggested that
Kramers could make this sound very convincing in his popular lectures.
89 Pauli to Bohr, 12 December 1924,PB, 186-189 (Item 74).
90 Pauli to Bohr, 31 December 1924,PB, 197-199 (Item 79).
91 Ibid.; Serwer, 'Unmechaniseher Zwang', 237ff.; Pauli to Heisenberg, 28 February
1925, and Pauli to Bohr, 30 April 1925,PB, 211-214 (Items 86, 87).
92 A. Sommerfeld, Atombau und Spektrallinien (Braunschweig, 1924).
93 W. Heisenberg, 'Zur Quantentheorie der Multiplettstruktur in der anomalen Zeeman-
effekte', Zeit. Phys. 32 (1925), 841-860. Compare Serwer, 'Unmechanischer Zwang',
239-245, and MacKinnon, 'Heisenberg', 159-163, for different interpretations of this
paper. The account given here follows Serwer.
94 Heisenberg, 'Quantentheorie der Multiplettstruktur', 856.
95 IbkJ., 842.
96 Interview with W. Heisenberg, SHQP; Heisenberg, 'Errinerungen'; Kronig, 'Turning
point'.
97 W. Heisenberg, 'tiber quantentheoretische Urn deutung kinernatischer und meehan-
ischer Beziehungen', Zeit. Phys. 33 (1925), 879-893. I have not gone into the details
of Heisenberg's work here, largely because it has been treated in great depth in several
other places. See especially SQM, Heisenberg, "Eninerungen'; Kronig, 'Turning point';
Cassidy, 'Werner Heisenberg'; and MacKinnon, 'Heisenberg',
98 Heisenberg to Pauli, 21 June 1925,PB, 219-221 (Item 91).
99 See Note 9 to Chapter 6 below.

CHAPTER 6. THE NEW KINEMATICS AND ITS EXPLORATION

1 W. Heisenberg, 'tiber quantentheoretische Umdeutung kinernatischer und mechan-

ischer Beziehungen', Zeit. Phys. 33 (1925), 879-893, translated in SQM, 261-276,
esp.262.
 152 NOTES

2 In the interests of clarity and brevity I have here abandoned Heisenberg's notations.
The conflicting notations he used, their origins and their implications for the genesis
of his paper are examined by van der Waerden, SQM, 30-34. A detailed if not altogether
convincing reconstruction of Heisenberg's thought processes is given in Mackinnon,
'Heisenberg', 164-184.
3 See Chapter 5 above.
4 The differentiation with respect to n could be theoretically justified only in the cor-
respondence principle limit of large n. Practically, however, it was justified by the result
to which it led, and that result was presumably the reason for its introduction. It pro-
vided the only means of translating from the formula obtained by substitution into
(5) to a form that could be quantised as in Born's virtual oscillator mechanics of the
previous year: M. Born, "tlber Quantenmechanik', Zeit. Phys. 26 (1924), 379-395,
translated in SQM, 181-198. The result was akin to that obtained from the virtual
oscillator theory by Kuhn, working in Copenhagen that Spring: W. Kuhn, 'Uber die
Gesamstlirke der von einem Zustande ausgehende Absorptionslinien', Zeit. Phys. 33
(1925),408-412, translated in SQM, 253-257.
5 W. Bothe and H. Geiger, 'tlber das Wesen des Comptoneffektes; ein experimenteller
Beitrag zur Theorie der Strahlung', Zeit. Phys. 32 (1925), 639-663; 'Experimentelles
zur Theorie von Bohr, Kramers urid Slater', Naturwissenschaften 13 (1925),440-441.
6 Heisenberg did establish energy conservation for the case of the anharmonic oscillator;
as his recollections confirm, he could hardly have continued, in the light of the Bothe-
Geiger results, had he been unable to do this: interview with W. Heisenberg, SHQP,
and Heisenberg, Physics and Beyond, 61.
7 Heisenberg had started out by considering hydrogen, but had found it too difficult.
See Heisenberg, 'Errinerungen', and see also MacKinnon, 'Heisenberg'.
8 Heisenberg to Pauli, 24 June 1924,PB, 225-229 (Item 93).
9 Pauli to Kramers, 27 July 1925, reflecting a similar statement in Heisenberg to Pauli,
9 July 1925:PB, 231-235 (Items 97, 96).
10 Born,My Life, 216.
11 Ibid., 217.
12 In particular, Born's assistance is acknowledged in the preface of R. Courant and
D. Hilbert, Methoden der Mathematischen Physik, Bd.. l (Berlin, 1924), the rust chapter
of which is on matrices, and the authors of which were his very close associates at
Gottingen.
13 Had Born's concern been primarily with the multiplication, then his delay in rec-
ognising the matrices would have been most extraordinary. Moreover, as indicated,
below, his recent work had been very much concerned with the physical implications
of posSible alternative theories.
14 Born to Pauli, 23 December 1919,PB, 9-11 (Item 4).
15 Born, 'Quantemnechanik'.
16 For a discussion of these problematic results, see Chapter 9 below. On receiving
the Bothe-Geiger results from Geiger in a letter dated April 17th, Bohr immediately
wrote to Born, who replied that he had already anticipated the result (most physicists
were indeed expecting it) and was working on a new theory to cope with it (Born to
Bohr, 4 April 1925, BSC): "The main thing is to keep the value of the Bohr-Kramers-
Slater theory, namely the emission of radiation during the stationary states. But now
there are besides these periods the jumps, which may be valuable here as momentum pro-
 NOTES 153

cesses. So if one tries generally to arrange the ordering of events in space and time, one
must class the stationary states with the wave emissions, and the jumps with the light-
quantum emissions." (Die Hauptsache ist, dass, man das Wertvolle der Bohr-Kramers-
Slaterschen Theorie beibehiiIt: niimlich die Emission der Wellenstrahlen wiihrend der
stationaren Zustiinde. Nun gibt es aber neben diesen Zeitabschnitten die Spriinge,
die hier als Momentanprozesse gelten mogen. Wenn man also iiberhaupt versucht, die
Ordnung der Vorgiinge in Raum und Zeit vorzuordnen, so muss man den stationaren
Zustiinde die Wellenemission, den Sprungen die Lichtquantenemission zuordnen.")
Born continued to explain that the wave would carry the light-quantum and create
interference effects by controlling the absorption of radiation by matter, This corres-
ponded to the interpretation given by de Broglie in his defmitive account of the matter
wave theory: L. de Broglie, There (Paris, 1924). de Broglie had since replaced it by a
guiding wave concept, L. de Broglie, 'Sur Ia dynamique du quantum de lumiere et
les interrerences', Comptes Rendus 179 (1924), 1309-1311. But Born's aquaintance
with de Broglie's work would most naturally have been through the These, which he had
been studying: Born, My Life, 231.
17 Born to Einstein, 15 July 1925, Born and Einstein, Letters 83-88 (Item 49). This
was despite Bohr's criticisms, for which see Note 20 below.
18 Interviews with J. Franck and W. Elsasser, SHQP.
19 M. Born, and P. Jordan, 'Zur Quantentheorie aperiodischer Vorgiinge', Zeit. Phys.
33 (1925), 479-505, submitted in June. That they should have been pursuing the
virtual oscillator theory at this late stage is further evidence of the heuristic attitude
adopted towards that theory, which was now physically disproven. It is possible that
Born hoped to reinterpret the formal analysis in de Broglie's sense.
20 Note 11 above. Responding to Born's advocacy of the matter wave theory, Bohr
had offered both general criticisms, for which see below, and specific ones: "The content
of your note has naturally interested us very much, but I must confess that I do not
believe that a contradiction-free description of the phenomena can be reached in the way
proposed. It seems to me that according to your picture the binding of the light-quanta
with the waves is not tight enough. On one hand, I do not understand how according
to your treatment it can be achieved that the paths of the light-quanta coincide with
sufficient accuracy with the propagation of the waves. If the interaction between a
quantum and a scattering atom depends only on the classically calculated momentum of
its virtual resonator it could indeed scarcely be avoided that the quantum, for example
in reflection or refraction, should be separated from the wave train to which it was
originally attached. On the other hand it seems to me that your picture can hardly
reproduce the quantitative relations of light absorption, for the supposition that the
probability of the absorption of a light-quantum in an atom should be proportional to
the intensity of the wave indeed leads to quite different laws from those corresponding
to either the wave theory or the corpuscular theory of light. The rough agreement
between the two theories, so far as the rectilinear propagation of light is concerned,
indeed even rests on the fact that the number of corpuscles passing through a plane
surface is proportional to the intensity of the waves, and that therefore the absorption
must be described through the assumption of constant effective cross-sections of the
atoms." ("Der Inahlt Ihrer Note hat uns natiirlich sehr interessiert, aber ich muss
gestehen, dass ich nicht glaube, dass eine widerspruchsfreie Beschreibung der Phiinomene
sich in der vorgeschlagenen Weise erreichen lasst. Es scheint mir, dass nach Ihrem Bilde
154 NOTES

der Verband der Lichtquanten mit den Wellen ein nicht geniigend enger ist. Einerseits
sehe ich nicht ein, wie es sich nach Ihre Vorstellung erreichen lasst, dass die Bahnen
der Lichtquanten mit hinreichender Genauigkeit mit der Fortpflanzung der Wellen
zusammenfallen. Wenn die Wechselwirkung zwischen einem Quant und einem streuenden
Atom nur von dem klassisch berechneten Moment seiner virtuellen Resonatoren abhiingt,
diirfte es ja kaum zu vermeiden sein, dass das Quant z.b. bei Spiegelung oder Brechung
vollkommen von demihm urspriinglich zugeordneten Wellenzug getrennt wird. Anderseits
scheint mir, dass Ihr Bild kaum die quantitativen Verhiiltnisse der Lichtabsorption
wiedergeben kann, denn die Annahme, dass die Wahrscheinlichkeit der Auffangung
eines Lichtquants durch ein Atom mit der Intensitat der Wellen proportional sein sollte,
flihrt ja zu ganz anderen Gesetzmassigkeiten als es so wie der Wellentheorie wie der
korpuskularen Theorie des Lichtes entsprechen wiirde. Die grobe Ubereinstimmung
dieser zwei Theorien, soweit es die gradlinige Ausbreitung des Lichtes betrifft, beruht
ja eben darauf, dass die Anzahl der Korpuskeln, die durch die Flacheneinheit geht
iiberall der Intensitat der Wellen proportional ist, und dass also die Absorption durch die
Annahme eines konstanten wirksamen Querschnitt der Atome beschreiben werden
muss." (Bohr to Born, 1 May 1925, BSC.»
21 Born, My Life, 218.
22 Ibid., 218-219. In the thick of quantum theoretical developments, Born had already
been overworking for some years, and continued to do so, driving himself to a nervous
breakdown in 1928-1929. It should be noted that Born's recollections do not tie in
with the submission dates of the papers he refers to, and I have given the latter the
greater weight. Quotations from the recollections in Jammer, Conceptual Development,
describing Born's holiday as a health cure and describing his meeting Jordan - whom
he already knew well of course - on a train, appear to be spurious.
23 M. Born and P. Jordan, 'Zur Quantenmechanik', Zeit. Phys. 34 (1925),858-888,
abridged translation in SQM, 277 -306.
24 For details of the redaction of the Born-Jordan paper see SQM, 38-40.
25 Born and Jordan, 'Quantenmechanik', 883 (not in translation).
26 Heisenberg to Pauli, 18 September 1925,PB, 236-241 (Item 98); SQM, 44-48.
27 Heisenberg to Jordan, 13 September 1925, in SQM, 43-44.
28 Heisenberg to Pauli, 18 September 1925,PB, 236-241 (Item 98).
29 SQM,49.
30 Born and Jordan, 'Quantenmechanik': SQM, 286.
31 M. Born, W. Heisenberg and P. Jordan, 'Zur Quantenmechanik II', Zeit. Phys. 35
(1925),557 -615, translated in SQM, 321-386.
32 My own phraseology.
33 Born,My Life, 218.
34 M. Born and W. Pauli, 'tiber die Quantelung gesttirter mechanischer Systeme',
Zeit. Phys. 10 (1922),137-158.
35 W. Pauli, review of Born's Vorlesungen iiber Atommechanik in Naturwissenschaften
13 (1925), 487.
36 Pauli to Kronig, 9 October 1925,PB, 242-249 (Item 100); see alsoSQM, 37.
37 Heisenberg to Pauli, 16 November 1925,PB, 255-257 (Item 105); see also SQM, 56.
38 Interview with J. Franck, SHQP.
39 See Chapter 4 above.
40 Heisenberg to Pauli, 16 November 1925,PB, 255-257 (Item 105).
 NOTES 155

41 Ibid.
42 Ibid.
43 When still in his teens his mathematics had been sufficient to argue on equal terms
with Weyl (and a few years later with Eddington) in the context of general relativity
theory: see Chaper 2 above.
44 Born, Heisenberg and Jordan, 'Quantenmechanik II'; for the redaction see SQM,
55-56.
45 W. Pauli, 'tiber das Wasserstoffspektrum vom Standpunkt der neuen Quanten-
mechanik', Zeit. PhY8. 36 (1926), 336-363, translated in SQM, 387-416.
46 P. Jordan, 'tiber kanonische Transformationen in der Quantenmechanik', Zeit. PhYII.
37 (1926),383-386, and 38 (1926), 513-517.
47 Heisenberg to Pauli, 16 November 1926,PB, 255-257 (Item 105).
48 Heisenberg to Born, 5 October 1925, and to Jordan, 7 October 1925, SHQP 18,2;
see SQM, 55, and Born, Heisenberg and Jordan, 'Quantenmechanik II': SQM, 325-326.
49 Ibid., 321-325; see also W. Heisenberg, 'tiber quantentheoretischer Kinematik und
Mechanik', Math. Ann. 95 (1926),694-705, where, despite the mathematical readership
at which the exposition was aimed, there is again no use of the term 'matrix mechanics'.
50 Pauli to Bohr, 17 November 1925,PB, 257-261 (Item 106).
51 Ibid.
52 See Chapter 2 above.
53 Or he may have received a very similar one.
54 Heisenberg to Pauli, 24 November 1925, PB, 262-266 (Item 108). I have translated
Heisenberg's "grobe" by both "rough" and "coarse", depending on the context.
55 Born and Jordan, 'Zur Quantentheorie aperiodischer Vorgange'.
56 P. Jordan, 'tiber das thermische Gleichgewicht zwischen Quantenatomen und
Hohlraumstrahlung', Zeit. PhYII. 33 (1925), 649-655.
57 Born, Heisenberg and Jordan, ,Quantenmechanik II': SQM,375-385.
58 S. N. Bose, 'Planck's Gesetz und Lichtquantenhypothese', Zeit. PhYII. 26 (1924),
178-181.
59 De Broglie, These.
60 Ibid., 79.
61 See also P. A. M. Dirac, manuscript draft of a critical presentation of Einstein-Bose
statistical mechanics, SHQP 36, 9: "It is a disadvantage of [Bose's] theory that the
cells play so important a part in it. One assumes that the whole of phase space is divided
into a number of compartments, and each atom or light-quantum as the case may be
is defmitely in one compartment. One can get over this difficulty by adopting the point
of view, fust proposed by de Broglie, that each particle is associated with a wave, and
letting the waves play the part of the cells in the previous theory. Several particles may
be associated with the same wave. This point of view is possible only because it turns
out that the number of waves associated with a given region in phase space is equal to
the number of cells into which that region of phase space was divided in the previous
theory. The results in the two theories become mathematically equivalent."
62 Born, Heisenberg and Jordan, 'Quantenmechanik II': SQM, 377.
63 Ibid., 378, and see Heisenberg to Pauli, 23 October 1925 and 16 November 1925,
PB, 251-252, 255-257 (Items 102, 105).
64 Heisenberg to Pauli, 16 November 1925, PB, 255-257 (Item 105).
65 Heisenberg to Pauli, 23 October 1925, PB, 251-252 (Item 102).
156 NOTES

66 W. Heisenberg, 'Mehrkorperproblem und Resonanz in der Quantenmechanik', Zeit.

Phys. 38 (1926), 411-427. For a discussion of the earlier history of this problem see
Serwer, 'Unmechanischer Zwang', and Chapter 4 above.
67 Heisenberg, 'Mehrkorperproblem und Resonanz', 422.
68 Ibid., 423.
69 E. Fermi, 'Zur Quantelung des idealen einatomigen Gases', Zeit. Phys. 36 (1926),
902-912. Fermi assumed that the molecules of the gas behaved according to Pauli's
exclusion principle, showed that this led to results in accordance with the experimental
Stern-Tetrode values for entropy at high temperatures, and derived the "Fermi-Dirac"
statistics in the form

for the number of molecules with energy shv, O! and {J being constants.
70 Heisenberg to Pauli, 15 November 1926, PB, 354-356 (Item 146).
71 We may reconstruct the probable source of his confusion. Given two parts of a
system, I and II, Heisenberg considered two possible solutions: either I was in state X,
II in state Y, say (X, y), or vice versa, (Y, X). Bose had treated these two possibilities
as a single state, so reducing the statistical weight from 2 to 1 as Heisenberg said. The
exclusion principle forbad the joint existence of (X, y) and (Y, X), but Heisenberg
took it as forbidding one of the two states altogether.
72 Born, Heisenberg and Jordan, 'Quantenmechanik II': SQM, 348-364; Jammer,
Conceptual Development, 218.
73 Born, Heisenberg and Jordan, 'Quantenmechanik II': SQM, 358, but see Note 78
below.
74 Wiener, I am a Mathematician, 108.
75 N. Wiener, 'The operational calculus', Math. Ann. 95 (1926), 557-584; for the
mathematical background see Jammer, Conceptual Development, 223-228.
76 M. Born and N. Wiener, 'A new formulation of the laws of quantisation of periodic
and aperiodic phenomena',J. Math. and Phys. 5 (1926),84-98; 'Eine neue Formulierung
der Quantengesetze fUr periodische und nicht-periodische Vorgange', Zeit. Phys. 36
(1926),174-187.
77 Ibid., 84. Although references will be given here to the English version of this paper,
it should be noted that as a whole the German is somewhat clearer.
78 The operator theory had been developed mainly by Hilbert and his students in
connection with integral equation theory, to which the method of infinite matrices had
been applied by Fredholm. Infmite matrices were primary to this research, operators
only secondary; but Born had already encountered the fact that while infinite matrix
theory and operator theory were equivalent for finite-dimensional spaces and for bounded
forms on more general spaces this did not seem to extend to unbounded forms, i.e.,
to the context of quantum mechanics. As we have seen, he had glossed over this problem
in the three-man-paper, but its existence provided strong mathematical grounds for
preferring the operator to the matrix formulation. An exact account of the situation,
and with it a mathematically rigorous formulation of quantum mechanics, was provided
only after von Neumann had taken up his general study of Hilbert spaces in 1927-1928:
see especially J. von Neumann, 'Allgemeine Eigenwertetheorie Hermetischer Funktion-
soperatoren', Math. Ann. 102 (1929),49-131.
 NOTES 157

79 The work of Born and Wiener also raised an interesting point concerning an aspect
of quantum mechanics that is not often treated, namely the intrusion of the imaginary
constant i. They noted that their ~eory seemed to attribute a motion to a particular
state of the form qk(t) = 1: qmk e 21Tlv (m, k)t, and that this was complex even when the
matrix (qmn) was Hermitian. From this they deduced that "there are then two real
motions belonging to every state, corresponding respectively to the real and pure imagi-
nary parts of the line of the matrix" (Born and Wiener, 'New formulation', 86). Unable
to make anything of this they did not pursue it, but Dirac had already noticed the
property in October 1925 and had followed it slightly further, though without publish-
ing his considerations: P. A. M. Dirac, manuscript treating virtual oscillators according
to the new quantum mechanics, October 1925, SHQP 36, 9. In the virtual oscillator
theory the emission and absorption oscillators had each been defmed by the real part
of a form C eiwt , and although they did not carry energy they were related probabilist-
ica11y to a transition process. In the new quantum mechanics, however, the restriction
to the real part was dropped, and an atom descnbed by complex oscillators of the form
C eiwt . Reviewing this situation, Dirac deduced that "the imaginary exponential is
essential and fundamental in the new theory", and noted that one need a combination
of eiwt and e-iwt oscillators in order to get any radiation at all. Taking another line of
approach, Jordan later linked the presence of i with the absence of classical causality,
for which see below.

CHAPTER 7. WAVE MECHANICS AND THE PROBLEM OF

INTERPRETATION
1 M. Born and N. Wiener, 'A new formulation of the laws of quantization of periodic
and aperiodic phenomena', J. Math. and Phys. 5 (1926), 84-98; Zeit. Phys. 36 (1926),
174-187.
2 E. Schrodinger, 'Quantisierung als Eigenwertproblem', Ann. der Phys. 79 (1926),
362-376; 489-527; 80 (1926),437-490; 81 (1926), 109-139. The frrst, second and
fourth communications are translated as 'Quantisation as an eigenvalue problem', in
Ludwig, Wave Mechanics, 94-105, 106-126, 151-167. For the background to the
theory see Jammer, Conceptual Development; Klein, 'Einstein and duality'; Gerber,
'Gesichte der Wellenmechanik'; Raman and Forman, 'Why Schrodinger'; Hanle, 'The
coming of age' and 'Schrodinger's reaction'; Wessels, 'Schrodinger's route'; Kragh,
'Schrodinger and the wave equation'.
a E. Schrodinger, 'tiber das Verhiiltnis der Heisenberg-Born-Jordanschen Quanten-
mechanik zu der meinen', Ann. der Phys. 79 (1926), 743-756, translated in Ludwig,
Wave Mechanics, 127-150. For the extent and limitations of Schrodinger's demonstra-
tion see Waerden, From matrix mechanics'.
4 Schrodinger, 'Quantisation', 163.
5 Ibid.
6 L. de Broglie, These (Paris, 1924), 69, and 'Sur la dynamique du quantum', Comptes
Rendus 179 (1924), 1309, emphasises the priority of the wave picture, although his
whole theory had been based upon the primacy of the particle concept for light.
7 Schrodinger, 'Quantisation', 163-164.
8 Schrodinger, 'Verhiiltnis', 128, and see Przibram, Letters on Wave Mechanics, for the
responses of Einstein, Lorentz, Wien and Planck.
158 NOTES

9 Lorentz to Schrodinger, 27 May 1926, and Einstein to Schrodinger, 31 May 1928,

in Przibram, Letters on Wave Mechanics, 43-54,30-31 (Items 19, 14). L de Broglie,
'La mecanique ondulatoire et la structure atomique de la matiere et du rayonnement',
Comptes Rendus 184 (1927), 273-274, was also very critical.
10 Interview with M. Born, SHQP, quoted in Jammer, Conceptual Development, 223.
11 Pauli to Jordan, 12 April 1926, PB, 315-320 (Item 131), translated in Waerden,
'From matrix mechanics'.
12 For a full analysis of Pauli's work and comparison with Schrodinger's, see Waerden,
'From matrix mechanics'. For Pauli's use of the operator formulation see Pauli to
Heisenberg, 31 January 1926,PB, 283-288 (Item 118).
13 Pauli to Jordan, 12 Apri11926,PB, 315-320 (Item 131)'
14 Heisenberg to Jordan, 28 July 1926,SHQP 18, 2.
15 See Chapter 2 above.
16 Pauli to SchrOdinger, 24 May 1926,PB, 324-327 (Item 134).
17 Pauli to Schrodinger, 22 November 1926,PB, 356-357 (Item 147).
18 Pauli to Schrodinger,12 December 1926,PB, 364-366 (Item 150).
19 Heisenberg to Dirac, 26 May 1926, SHQP 59,2.
20 W. Heisenberg, 'Mehrkorperproblem und Resonanz in der Quantenmechanik', Zeit.
Phys. 38 (1926),411-427, esp. 422; Heisenberg to Pauli, 8 June 1926, PB, 328-329
(Item 136).
21 Heisenberg to Pauli, 8 June 1926, PB, 328-329 (Item 136): literally "dung" or
"manure".
22 Ibid.
23 Heisenberg to Pauli, 28 July 1926, PB, 337-340 (Item 142).
24 Sommerfeld to Pauli, 26 July 1926,PB, 337 (Item 141).
25 Bohr to Kronig, 28 October 1926, SHQP 16, 1.
26 Born to Schrodinger, 16 May 1927, SHQP 41,7.
27 Jordan to SchrOdinger, catalogued as May 1926 but date uncertain, SHQP 41, 8. It
should be said that Schrodinger does not appear to have been so naive as his opponents
made out. His emphasis on the wave interpretation was justified in his 'Quantisation',
163-164, partly in terms of the danger of reverting to the untenable concept of three-
dimensional spatial representations if the particle picture were adopted. His own inter-
pretation centred on the proposition that the energy concept was inapplicable on the
microscopic scale: E. Schrodinger, 'Energieaustausch nach der Wellenmechanik', Ann.
der Phys. 83 (1927), 956-968; Schrodinger to Planck, 31 May 1926, and to Lorentz,
6 June 1926, in Przibram, Letters on Wave Mechanics, 8-11 and 55-66 (Items 4, 20).
See also L Wessels, 'Schrodinger's interpretations of wave mechanics', Ph.D. dissertation,
Indiana University, 1975.
28 Born to Schrodinger, 16 May 1927, SHQP 41, 7: "Heisenberg war von vornherein
nicht meiner Meinung, dass Ihre Wellenmechanik physikalisch mehr bedeute, als unsere
Quantenmechanik .... Inzwischen habe ich rnich aber wieder zu Heisenbergs Stand-
punkt zuriickgefunden."
29 Ibid.: ..... die einfache Art, aperiodische Vorgange (Stosse) zu behandeln, bracht
mir zuniichst zu dem Glauben der tiberlegenheit Ihre Anschauungsweise."
30 M. Born, 'Quantenmechanik der Stossvorgange', Zeit. Phys. 38 (1926), 803-827,
translated in Ludwig, Wave Mechanics, 206-225, esp. 224.
31 M. Born, 'Physical aspects of quantum mechanics', Nature 119 (1926), 354.
 NOTES 159

32 M. Born, 'Zur Quantenmechanik der Stossvorgange', Zeit. Phys. 37 (1926), 863-

867.
33 Ibid., 864.
34 Ibid., 865-866.
35 Born, 'Quantenmechanik der Stossvorgange': Ludwig, Wave Mechanics, 225.
36 Born, 'Zur Quantenmechanik der Stossvorgiinge'.
37 Born, 'Quantenmechanik der Stossvorgange': Ludwig, Wave Mechanics, 207.
38 Ibid., 207.
39 Ibid., 207.
40 Einstein to Born, 4 December 1926, Born and Einstein, Letters, 90-91 (Item 52).
41 Ibid., 91.
42 Born, Experiment and Theory, 23, and 'Bedeutung zur statistischen Deutung der
Quantenmechanik', in Bopp, Wemer Heisenberg, 103-118, esp. 103.
43 Chapter 6 above, and Born, My Life, 231.
44 This possibility had long been discounted: see for example Hendry, 'Wave-particle
duality'.
45 H. A. Lorentz, Problems of Modem Physics (New York, 1927 and 1957), 156.
46 Hendry, 'Weimer culture'.
47 Einstein to Born, 27 January 1920, and 3 March 1920, Born and Einstein, Letters,
20-26 (Items 13, 14).
48 His assistance in the solution of Schrodinger's eigenvalue problems was acknowledged
by Schrodinger, 'Quantisation': Ludwig, Wave Mechanics, 97. His assistance on aspects
of the matrix mechanics was acknowledged by Born: Born to Weyl, 3 October 1925,
ETHZ 91, 488.
49 M. Born, 'Das Adiabatenprinzip in der Quantenmechanik', Zeit. Phys. 40 (1926),
167-192, esp. 167.
50 Hendry, 'Weimar culture'.
51 Heisenberg to Pauli, 24 November 1925,PB, 262-266 (Item 108).
52 Born, 'Zur Quantenmechanik der Stossvorgiinge', 863.
53 Note 3 of Chapter 3 above. Born's training had been with Hilbert and Minkowski.
54 For Pauli's reaction to the virtual oscillator theory see Chapter 5 above.
55 W. Pauli, 'Quantentheorie', Handbuch der Physik 23 (1926), 1-278. The corre-
spondence is not complete and Pauli may well have made comments now lost.
56 Interview with Heisenberg, SHQP.
57 Heisenberg to Pauli, 28 July 1926,PB, 337-340 (Item 142).
58 K. Lanczos, 'Dber eine feldmiissige Darstellung der neuen Quantenmechanik', Zeit.
Phys. 35 (1926), 812-830.
59 P. A. M. Dirac, 'The fundamental equations of quantum mechanics', Proc. Roy. Soc.
AI09 (1925),642-653, reprinted in SQM, 307-320.
60 P. A. M. Dirac, 'Quantum mechanics and a preliminary investigation of the hydrogen
atom', Proc. Roy. Soc. AllO (1926),561-579, reprinted in SQM, 417-427-.
61 Ibid., 418.
62 P. A. M. Dirac, 'On the theory of quantum mechanics',Proc. Roy. Soc. Al12 (1926),
661-667, esp. 667.
63 Ibid., 662.
64 See Note 69 of Chapter 6 above.
65 Dirac, 'Theory of quantum mechanics', 662, 666.
160 NOTES

66 Pauli to Heisenberg, 19 October 1926,PB, 340-349 (Item 143).

67 M. Born, W. Heisenberg and P. Jordan, 'Zur Quantenmechanik II', Zeit. Phys. 35
(1925),557-615, translated in SQM, 321-386, esp. 385.
68 Pauli to Heisenberg, 19 October 1926,PB, 340-349 (Item 143).
69 Born, Heisenberg and Jordan, 'Zur Quantenmechanik II': SQM, 380.
70 Heisenberg to Pauli, 15 November 1926, PB, 354-356 (Item 146).
71 Pauli to Heisenberg, 19 October 1926, PB, 340-349 (Item 143).
72 H. A. Kramers, 'Wellenmechanik und halbzahlige Quantisierung', Zeit. Phys. 39
(1926),828-840.
73 W. Heisenberg, 'Schwankungserscheinungen und Quantenmechanik', Zeit. Phys. 40
(1926),501-506.

CHAPTER 8. TRANSFORMATION THEORY AND THE

DEVELOPMENT OF THE PROBABILISTIC INTERPRETATION

1 Heisenberg to Pauli, 28 October 1926,PB, 349-352 (Item 144).

2 Ibid.
3 Reported in Bohr to Kronig, 28 October 1926, SHQP 16, 1.
4 Heisenberg to Pauli, 4 November 1926,PD, 352-353 (Item 145).
5 W. Heisenberg, 'Schwankungserscheinungen und Quantenmechanik', Zeit. Phys. 40
(1926),501-506.
6 Ibid., 501.
7 Heisenberg to Pauli, 4 November 1926,PD, 352-353 (Item 145).
8 Ibid.
9 Heisenberg to Pauli, 28 October 1926,PB, 349-352 (item 144).
10 Heisenberg to Pauli,4 November 1926, PB, 352-353 (Item 145); O. Klein, 'Quan-
tentheorie und fUnfdimensionale Relativitiitstheorie', Zeit. Phys. 37 (1926), 895-906;
T. Kaluza, 'Zum Unitiitsproblem der Physik', S.-B. Preuss. Akad. Wiss. (1921),966-972.
11 Heisenberg to Pauli, 23 November 1926, PB, 357 -360 (Item 148).
12 P. A. M. Dirac, 'The phYSical interpretation of the quantum dynamics', Proc. Roy.
Soc. All3 (1927), 621-641, esp. 621.
13 P. Jordan, 'Uber eine neue Begriindung der Quantenmechanik', Zeit. Phys. 40 (1927),
809-838.
14 We may note also that his ftrst research had been on a critique of Einstein's argu-
ments in support of the necessity of light-quanta: P. Jordan, 'Zur Theorie der Quan-
tenstrahlung',Zeit. Phys. 30 (1924), 297-319.
15 Jordan, 'Neue Begriindung', 81l.
16 Dirac to Jordan, 24 December 1926,SHQP 18, 1.
17 Ibid.
18 Dirac, 'Physical interpretation', 641; Pauli to Bohr, 17 November 1925, PD, 257-
261 (Item 106).
19 See the discussion of Bohr's ideas in Chapter 9 below.
20 See Chapter 2 above.
21 D. Hilbert, J. von Neumann and L. Nordheim, 'Uber die Grundlagen der Quanten-
mechanik',Math. Ann. 98 (1927),1-30.
22 Jammer, Conceptual Development, 314-318.
 NOTES 161

23 Although they left their transformations in general form the authors cannot but have
been aware of the specific form required: Dirac had, after all, noted it explicitly. They
did not, however, discuss the matter.
24 Dirac, 'Physical interpretation', 625.
25 J. von Neumann, 'Mathematische Begriindung der Quantenmechanik',Nach. konigliche
Ges. Wiss. Gottingen, Math. ·Phys. Kl (1927), 1-57; 'Wahrscheinlichkeitstheoretischer
Aufbau der Quantenmechanik', ibid., 245 -272; 'Thermodynarnik Quantenmechanischer
Gesamtheiten', ibid., 273-291; 'Beweis der Ergodensatzes und des H-Theorems in
der neuen Mechanik', Zeit. Phys. 57 (1929), 30-70; 'Allgemeine Eigenwertetheorie
Hermetischer Funktionsoperatoren', Math. Ann. 102 (1929),49-131.
26 von Neumann, 'Mathematische Begriindung'; E. Fischer, 'Sur la convergence en
moyenne', Comptes Rendus 144 (1907), 1022-1024; F. Riesz, 'Sur les systemes ortho-
gonaux de fonctions', ibid., 615-619.
27 von Neumann, 'Warscheinlichkeitstheoretischer Aufbau'.
28 For a relatively comprehensible presentation of the theory underlying von Neumann's
statement of the probability interpretation, see B. L. van der Waerden, Group Theory
and Quantum Mechanics (New York, 1974), 12-16.

CHAPTER 9. THE UNCERTAINTY PRINCIPLE AND

THE COPENHAGEN INTERPRETATION

1 Heisenberg to Pauli, 28 October 1926,PB, 349-352 (Item 144).

2 Ibid.
3 Ibid.
4 See W. Pauli, 'tiber Gasentartung und Paramagnetismus', Zeit. Phys. 41 (1927), 81-
102, submitted on 6 December 1926, but not very helpful in the present context. On
Heisenberg's letter to him of 4 November 1926, PB, 352-353 (Item 145), Pauli wrote
the words "Ferromagnetismus geht nicht!"
5 Heisenberg to Pauli, 15 November 1926,PB, 354-356 (Item 146).
6 The precise status of the time variable in quantum mechanics, as compared with the
other kinematic and mechanical quantities, was and continued to be somewhat pro-
blematic. See for example Jammer, Philosophy, 140-141.
7 W. Heisenberg, 'Schwankungserscheinungen und Quanten-mechartik', Zeit. Phys. 40
(1926),501-506.
8 Heisenberg to Pauli, 23 November 1926, PB, 357-360 (Item 148).
9 See the discussion of Bohr's ideas below, this chapter.
10 Interview with Heisenberg, SHQP; Heisenberg, Physics and Beyond, 77.
11 P. Jordan, 'Kausalitiit und Statistisch', Naturwissenschaften 15 (1927), 105-107,
and 'Philosophical foundations of quantum theory',Nature 119 (1927), 566-568.
12 Heisenberg to Pauli, 5 February 1927, PB, 373-376 (Item 153).
13 Heisenberg to Pauli, 23 February 1927,PB, 376-382 (Item 154).
14 W. Heisenberg, 'tiber den anschaulichen Inhalt der quantentheoretischer Kinematik
und Mechanik', Zeit. Phys. 43 (1927),172-198.
15 Ibid.; Bohr's criticisms are given in the postscript on 197-198. See Jammer, Philo-
sophy, Chapter 3, for an extensive discussion of the measurement thought experiment.
16 Heisenberg, 'Anschaulichen Inhalt'; Heisenberg to Pauli, 23 February 1927, PB,
376-382 (Item 154).
162 NOTES

17 Heisenberg, 'Anschaulichen Inhalt', 197.

18 Heisenberg to Pauli, 23 February 1927,PB, 376-382 (Item 154).
19 Heisenberg, 'Anschaulichen Inhalt', 173.
20 Heisenberg to Pauli, 23 February 1927,PB, 376-382 (Item 154).
21 M. Born, 'Zur Quantenmechanik der Stossvogiinge', Zeit. Phys. 37 (1926), 863-867.
22 Jordan, 'Kausalitiit und Statistisch'.
23 M. Born 'Quantenmechanik und Statistik', Naturwissenschaften 15 (1927), 238-
242.
24 H. A. Senftleben, 'Zur Grundlagen der Quantentheorie', Zeit. Phys. 22 (1923),127-
156, esp. 131.
25 Heisenberg, 'Anschaulichen Inhalt', 197.
26 Heisenberg to Pauli, 23 February 1927, PB, 376-382 (Item 154).
27 Heisenberg to Pauli, 9 March 1927,PB, 383-385 (Item 156).
28 Heisenberg, 'Anschaulichen Inhalt', 172-173.
29 Ibid., 172.
30 Heisenberg, Physics and Beyond, 63.
31 Heisenberg to Pauli, 9 March 1927,PB, 383-385 (Item 156).
32 See for example N. Bohr, 'On the application of the quantum theory to atomic
structure. Part I. The fundamental postulates', Supplement to Proc. Camb. Phi/.- Soc.
(1924),1 and 35.
33 Ibid., 35.
34 N. Bohr, H. A. Kramers and J. C. Slater, 'The quantum theory of radiation', Phil.
Mag. 47 (1924), 795-802. See Chapter 5 above.
35 A. Einstein, 'Quantentheorie des einatomigen idealen Gases', S.-B. Preuss. Akad.
Wiss. (1924), 261-267, and ibid. (1925), 3-14. C. Ramsauer, ''Ober das Wirkungs-
querschnitt der Gasmolekule gegeniiber langsamen Elektronen', Ann. der Phys. 64
(1922), 513-540; ibid. 66 (1922), 546-558; ibid. 72 (1923),345-352. C. Davisson,
'The scattering of electrons by a positive nucleus of limited field',Phys. Rev. 21 (1923),
637-649; C. Davisson and C. H. Kunsman, 'The scattering of low speed electrons by
platinum and magnesium', ibid. 22 (1923), 242-258. Ramsauer's results had long been
recognised as presenting a problem, while the implications of the Davisson-Kunsman
experiments were realised during the Spring of 1925 during discussions on de Broglie's
matter-wave theory between Born, Franck, and the research student Elsasser (Born, My
Life, 231). They were published in W. Elsasser, 'Bemerkung zur Quantenmechanik freier
Elektronen', Naturwissenschaften 13 (1925), 711.
36 See Note 48 to Chapter 5 above and, for physicists' expectations, Born to Bohr, 15
January 1925 and 24 April 1925,BSC.
37 N. Bohr, speech to the Royal Danish Academy, reported in Nature 116 (1925), 262.
38 Bohr to Heisenberg, 18 April 1925, BSC: "Besonders durch Gesprache mit Pauli
angeregt, quiile ich mich in diesen Tagen nach besten Kriiften, mich in die Mystik der
Natur einzuleben und versuche, mich auf aIle Eventualitiiten vorzubereiten, ja sogar auf
die Annahme einer Kopplung der Quantenprozesse in entferten Atomen. Die Kosten
dieser Annahme sind allerdings so gross, dass sie nicht in der gewOhnlichen, raum-
zeitlichen Beschreibung ermessen lassen."
39 Bohr to Geiger, 21 April 1925, BSC, in Stuewer, Compton Effect, 301.
40 See Note 38 above.
41 Bohr to Franck, 21 April 1925, BSC: "Ich hatte schon lange die Absicht, Ihnen
 NOTES 163

wieder zu schreiben, denn der Zweifel an der Richtlgkeit meiner Oberlegnungen iiber die
Stosserscheinungen, dem die Nachschrift in meinem letzten Brief Ausdruck gab, hat sich
seitdem immer verstiirkt. Es sind besonders die Ramsauerschen Ergebnisse der Durch-
dringung langsamer Elektronen durch Atome, die sich anscheinend dem angenommen
Gesichtspunkte nicht eintiigen. In der Tat diirften diese Ergebnisse unserer gewohnlichen
raumzeitlichen Naturbeschreibung Schwierigkeiten iihnlicher Art darbieten wie eine
Koppelung der Zustandsiinderung entfemter Atome durch Strahlung. Dann ist aber kein
Grund mehr, an einer solchen Koppelung und an den Erhaltungssatzen iiberhaupt zu
zweifeln. Dies ist nur eine grosse Befriedigung denn, wie Sie hervorheben, wird ja dann so
vieles bei den Stossen so ungemein viel einfacher. Auch waren die thermodynamischen
Betrachtungen von Einstein ja sehr beunruhigend. Ich habe schon diesen Morgen an
Fowler geschrieben, dass ich eine englische Arbiet iiber die Bremsung der a-strahlen
zurUckziehen und werde Ahnliches an Scheel betreffend die ibm zugesandte Arbeit
schreiben. Ausserdem habe ich eben jetzt von Geiger gehort, dass seine Versuche flir die
Koppelung entschieden haben, und es ist wohl nichts anderes zu tun als unseren Re-
volutionsversuch moglichst schmerzlos in Vergessenheit zu bringen. Unsere Ziele werden
wir aber doch nicht so leicht vergessen konnen und in den letzten Tagen habe ich mit
allerlei wilden Spekulationen geqiialt, urn eine adiiquate Grundlage der Beschreibung der
Strahlungsphiinomene zu finden. Dariiber habe ich viel mit Pauli diskutiert, der jetzt
hier ist, und dem seit langem unser "Kopenhagener Pretsch" unsympatisch war." The
earlier letter referred to appears to have been lost.
42 Bohr to Born, 1 May 1925, BSC: "Ganz abgesehen von der Frage der Richtigkeit
derartiger Einwiinde gegen Ihr Theorie, mochte ich gem betonen, dass ich der Ansicht
bin, dass die Annahme einer Koppelung zwischen den Zustandiinderungen in entfernten
Atomen durch Strahlung einer einfache Beschreibungsmoglichkeit des physikalischen
Geschehens mittels anschaulicher Bilder ausschliesst. Mit meinen Ausserungen in den
Brief an Franck iiber die Koppelung war nur gemeint, dass ich den Verdacht bekommen
hatte, dass schon fUr die Stosserscheinungen solchen Bildern ein noch geringeren
Anwendbarkeit zukommt als gewohnlich angenommen. Dies ist ja zuniichst eine rein
negativ Aussage, aber ich ruhle, besonders wenn die Koppelung wirklich eine Tatsache
sein sollte, dass man dann in noch hoheren Grade wie bisher seine Zuflucht zu sym-
bolischen Analogien nehmen muss. Eben in letzter Zeit habe ich mir den Kopfe
zerbrochen in solche Analogien mich hineinzutriiumen."
43 N. Bohr, 'tiber die Wirkung von Atomen bei Stossen', Zeit. Phys. 34 (1925), 142-
157, esp. 154.
44 Interview with W. Heisenberg, SHQP.
4S See Chapter 4 above.
46 See Chapter 2 above, and Heisenberg to Pauli,S February 1927,PB, 373-376 (Item
153), where Heisenberg himself makes a similar statement.
47 The main argument appears to have been between Ehrenfest, supporting the five-
dimensional theory, and Pauli and Heisenberg, opposed to it. In the Spring of 1927 Bohr
sided with Klein, who continued to work with the wave mechanics, against Heisenberg;
but his position the previous Autumn is unclear. See Ehrenfest to Pauli, 25 January
1927, and Heisenberg to Pauli,S February 1927,4 April 1927, 16 May 1927, and 31
May 1927,PB, 371-376,390-397 (Items 152, 153, 161, 163 and 164).
48 Interview with W. Heisenberg, SHQP; Heisenberg, PhYSics and Beyond, 16, and in
Rosental, Niels Bohr, 104.
164 NOTES

49 Heisenberg, 'Anschaulichen Inhalt'.

50 Pauli to Bohr, 6 August 1927, PB, 402-406 (Item 168). The context was the quantum
electrodynamics of Jordan.
51 Jordan recalled (Jammer,PhOoaophy, 68) that only Pauli's skillful diplomacy avoided
a serious conflict between Bohr and Heisenberg; but this trouble seems to have been on a
personal level, arising out of a misunderstanding, quite unintentional, between Heisenberg
and Klein, corrected by Pauli at Heisenberg's request: Heisenberg to Pauli, 16 May 1927
and 31 May 1927,PB, 394-397 (Items 163, 164).
52 Heisenberg, 'Anschaulichen Inhalt', 197-198.
53 Heisenberg to Pauli, 14 March 1927,PB, 387-388 (Item 158).
54 Heisenberg to Pauli, 4 April 1927,PB, 390-393 (Item 161).
55 Heisenberg to Pauli, 16 May 1927,PB, 394-396 (Item 163).
56 Heisenberg, 'Anschaulichen Inhalt'; the title speaks for itself.
57 For further discussion of Heisenberg, Bohr and the meaning of Anschaulichkeit see
Miller, 'Beyond Anschaulichkeit' and 'Visualisation lost and regained'.
58 Heisenberg to Pauli, 31 May 1927,PB, 396-397 (Item 164).
59 Ehrenfest to Pauli, 24 January 1927, and Heisenberg to Pauli,S February 1927,
PB, 371-376 (Items 152, 153). This debate was also related to that on the interpreta-
tion of spin, which had occupied both Pauli and Heisenberg the previous Spring.
60 See Note 51 above.
61 Heisenberg to Pauli, 31 May 1927,PB, 396-397 (Item 164).
62 N. Bohr, 'The quantum postulate and the recent development of atomic theory',
Atti del Congresso Intemazionale dei Fisici, VoL 2 (Bologna, 1928),568-588; see also
Nature 121 (1928),580-590 and Naturwissenschaften 16 (1928), 245-257.
63 Ibid., (Nature), 580.
64 Ibid., 580.
6S Ibid., 580.
66 Ibid., 581.
67 Ibid., 582-584.
68 Pauli to Bohr, 17 October 1927, PB, 411-413 (Item 173).
69 For example, P. Jordan, 'Die Quantenmechanik und die Grundprobleme der Biologie
und Psychologie', Naturwissenschaften 20 (1932), 815-821.
70 Pauli to Bohr, 6 August 1927,PB, 402-406 (Item 168).
71 Bohr to Pauli, 13 August 1927,PB, 406-407 (Item 169). M. Born and W. fleisenberg,
'La mecanique des quanta', in Institut International de Physique Solvay, Electrons et
Photons (Paris, 1928), 143-181.
72 P. Jordan and W. Pauli, 'Zur Quantenelektrodynamik ladungsfreier Felder', Zeit.
Phys. 47 (1928),151-173.
73 Born and Heisenberg, 'Mecanique des quanta'; N. Bohr, 'Le postulat des quanta et Ie
nouveau developpement de I'atomistique', in Institut International de Physique Solvay,
Electrons et Photons, 215-247; Dirac, in the discussion on Bohr's paper, ibid., 261-263.
74 de Broglie was advocating his theory of double solution, itself as radical a departure
from classical physics as was the Copenhagen interpretation. SchrOdinger was pursuing
his semi-classical interpretation with the abandonment of the classical energy concept.
Einstein was already talking in terms of a statistical or ensemble interpretation as being
all that the theory could unambiguously support (from a completely opposite viewpoint
this actually came very close to Dirac's position). Lorentz had criticisms to make of
 NOTES 165

everybody, but knew not what to do himself. The various p.ositi.ons are discussed and
referenced in Jammer, Philosophy.
75 This m.o.od was apparent from the pr.oceedings and was als.o recalled by Heisenberg:
interview with W. Heisenberg, SHQP.
76 J.ordan's early pr.obings in this directi.on have been discussed above. The subject
then d.ominated his and Dirac's w.ork, and Pauli's c.orrespondence, fr.om February 1927.
See P. A. M. Dirac, 'The quantum the.ory .of the emissi.on and abs.orpti.on .of radiati.on',
Proc. Roy. Soc. Al14 (1927), 243-265, and The quantum the.ory .of dispersi.on', ibid.,
710-728;J.ordan and Pauli, 'Quantenelektr.odynamik';PB, 385ff.
77 W. Heisenberg and W. Pauli, 'Zur Quantendynamik der Wellenfelder', Zeit. Phys. 56
(1929),1-61;59 (1930), 168-190.
78 See Jammer,Philosophy, Chapter 6.
79 Ibid., 76.
80 B.orn and Heisenberg, 'Mecanique des quanta'.

CHAPTER 10. CONCLUDING REMARKS

1 See especially the work .of HanIe, and als.o that of Wessels, cited in the bibliography.
2 See especially Cassidy, 'Werner Heisenberg' and 'Heisenberg's rust model'; see also
MacKinnon, 'Heisenberg'.
3 See Kragh, 'Methodology and philosophy of science'.
4 See Kragh, 'Niels Bohr's second at.omic theory', and the introductions to the volumes
of the Collected Works .of Bohr.
5
See Serwer, 'Unmechanischer Zwang'.
6
See for example, Klein, 'First phase', and Raman and Forman, 'Why Schrodinger'.
7 F.orman, 'Weimar culture'.
8 Brush, 'Chimerical cat'.
9 C.onfusingly, alth.ough the language .of Anschaulichkeit may itself be traced t.o Kant,
the way in which the word was used changed significantly: see Miller, 'Bey.ond Anschau-
lichkeit'.
10 The clearest indication of this is in the overwhelmingly h.ostile reception afforded t.o
the purest Kantian of the quantum mechanics period, Eddingt.on.
 SELECT BIBLIOGRAPHY

A. UNPUBLISHED PRIMAR Y SOURCES

The great majority of the known extant source material for the history of quantum
mechanics is to be found in the micro fUm archive, Sources for the History of Quantum
Physics (SHQP), copies of which are located at the University of California at Berkeley,
the american Philosophical Society at Philadelphia, the American Institute of Physics in
New York, the Niels Bohr Institute in Copenhagen and (shortly) the Science Museum in
London. The archive contains the Bohr Scientific Correspondence (BSC) and the Bohr
Manuscript Collection (Bohr MSS), also the Pauli Correspondence. It is frequently
augmented, but the original catalogue remains invaluable, including not only items in the
archive but also those in other collections, together with biographical details of the
physicists: Kuhn, T. S., et al., (eds.): 1967, Sources for History of Quantum Physics
(Philadelphia). Interviews with many of the quantum physicists are also to be found in
these archives.
Apart from the SHQP set of archives, the only other archive to be used in this study
is the Weyl collection at the ETH Ziirich (ETHZ).

B. PUBLISHED CORRESPONDENCE AND PAPERS

The most important sources for this study have been the published correspondence and
papers of Pauli and Bohr:

Pauli, W.: 1979, Wissenschaftlicher Briefwechsel mit Bohr, Einstein, Heisenberg u.a.,
Band 1 :1919-1929 (Berlin, Heidelberg, New York).
Bohr, N.: 1975, Collected Works, Vol. 3 (Amsterdam).

The works of Bohr (Volumes 1, 2 and 4 are also relevant) contain a selection of his
correspondence as well as previously unpublished papers. Other useful collections of
correspondence include:

Przibram, K. (ed.): 1967, Letters on Wave Mechanics (New York).

Born, M. and A. Einstein: 1971, The Born·Einstein Letters (New York).
Einstein, A. and M. Besso: 1972, Correspondance 1903-1955 (Paris).
Einstein, A. and A. Sommerfeld: 1968, Briefwechsel (Basel).

C. REPRINT COLLECTIONS AND TRANSLATIONS

As well as the reprints and translations of individual papers there are also several thematic
collections. Among the most useful of these are:

166
 SELECT BIBLIOGRAPHY 167

de Broglie, L. and L. Brillouin: 1928, Selected Papers on Wave Mechanics (London).

de Broglie, L.: 1953: La Physique Quantique Restera-t-elle Indeterministique? (Paris).
ter Haar, B.: 1967, The Old Quantum Theory (Oxford). [But note that the translations
here are not very reliable.]
Bohr, N.: 1924, Theory of Spectra and Atomic Constitution (Cambridge).
Bohr, N.: 1934, Atomic Theory and the Description of Nature (Cambridge).
Hermann, A. (ed.): 1962-1964, Dokumente der Naturwissenschaften, Bd.I-4 (Stuttgart).
Lorentz H. A. et al.: 1923, The Principle of Relativity (London, 1923; New York, 1952).
Ludwig, G.: 1967, Wave Mechanics (Oxford).
SchrOdinger, E.: 1928, Collected Papers on Wave Mechanics (London).
van der Waerden, B. L.: 1967, Sources of Quan tu m Mechanics (Amsterdam).

Collected editions of the work of individual physicists include:

Bohr, N.: 1974-, Collected Works (Amsterdam).

Born, M.: 1963, Ausgewahlte Abhandlungen (Gottingen).
Ehrenfest, P.: 1959, Collected Scientific Papers (Amsterdam).
Fermi, E.: 1962, Collected Papers (Chicago).
Kramers, H. A.: 1956, Collected Scientific Papers (Amsterdam).
Pauli, W.: 1964, Collected Scientific Papers (New York).
Von Neumann, J.: 1961, Collected Works (Oxford).
Weyl, H.: 1968, Gesammelte Abhandlungen (Berlin).

At the time of writing a collection of Heisenberg's papers is also in preparation.

D. SECONDARY WORKS

The standard and invaluable general history of quantum theory is Jammer's Conceptual
Development of Quantum Mechanics. This contains a few mistakes, notably in the treat-
ment of the causality issue (see Hendry, 'Weimar Culture', 161-162) and in the trans-
mission of Born's recollections (for which see instead the originals, published since), but
it is for the most part a reliable and comprehensive guide to the published primary
material. More comprehensive but still to be assessed is the massive multi-volume work
by Mehra and Rechenberg. Compared with these works none of the other general
histories, only a few of which are listed below, add anything significant.
There is a shortage of good biographical treatments of the quantum physicists, but
the situation is gradually improving. The Collected Works of Bohr and Briejwechsel of
Pauli (see Section B above) both include excellent and extensive introductory and
editorial material and will when complete constitute good scientific biographies. Other-
wise, the best studies of Bohr are those by Rosenfeld, Honner and StoIzenburg, to be
supplemented by the volume edited by Rozental. A major biography of Pauli is currently
in preparation, but meanwhile see Richter's book, the introduction to Pauli's Papers
(Section C above) and the article by Fierz. For Heisenberg and Born, the most thorough
treatments are their respective autobiographical works, those of Heisenberg being
particularly interesting and that of Born both accurate and informative. There is also a
popular but rather good biography of Heisenberg by Hermann. Of the many biographies
168 SELECT BIBLIOGRAPHY

of Einstein the most valuable are those by Pais, Frank, Seelig and Kuznetsov. Of the full
length biographies available of other physicists, those by Reid (Hilbert), Scott
(SchrOdinger), Douglas (Eddington), Klein (Ehrenfest) and Segre (Fermi) deserve special
mention, though Scott in particular has been surpassed by more recent work. For
Sommerfeld, see especially the article by Born.
In general the thematic treatments of aspects of the history of quantum mechanics,
many of them dealing largely with the work of a single physicist, are rather better than
the specifically biographical works. The intellectual and soCial context of quantum
mechanics has as yet received relatively little attention, but Holton's Thematic Origins of
Scientific Thought, Miller's papers on Anschaulichkeit, and Forman's work on the
causality issue (but see also the corresponding paper by Hendry) are essential reading.
The papers of Brush are also interesting in this respect. Much has been written on the old
quantum theory of Planck and Einstein, and selected references are included below. Of
particular interest as background to the history of quantum mechanics are the works of
Kuhn, Klein and McCormmach, together with Hendry's paper on the wave-particle duality
and Stuewer's book on the Compton effect. Relativity theory has also received a lot of
attention, though most of this has been devoted to the special theory, for which see
especially the papers of Holton and Miller. Interesting contributions to the history of
general relativity theory include the books of Pauli, North and Mehra, as well as the
excenent papers by Earman and Glymour.
The history of Bohr's atomic theory leading up to the creation of matrix mechanics is
particularly wen documented in a set of papers fitting into a natural sequence, by
Heilbron and Kuhn, Heilbron, Forman ('Lande'), Cassidy, Forman ('Doublet riddle'),
Serwer and Hendry. The present work is particularly indebted to this set of papers. The
best treatment of matrix mechanics itself remains that by van der Waerden in his Sources,
though this should be supplemented by the recollections of Heisenberg and Born and by
the papers of Serwer, Hendry and MacKinnon. On wave mechanics the work of de
Broglie has been treated by many writers, but never very well, while that of SChrodinger
and its origins are covered in an excellent group of papers by Klein ('Einstein and the
wave-particle duality'), Raman and Forman, Hanle and Kragh. For the evolution of
quantum mechanics the best works remain the two books by Jammer and the article by
van der Waerden.

Bemkopf, M.: 1967, 'A history of inilDite matrices', Arch.Hist.Exact Sci. 4, 308-358.
Bloch, F.: 1976, 'Heisenberg and the early days of quantum mechanics', Phys. TodJZy
(December), 23-27.
Bohr, N.: 1949, 'Discussions with Einstein', in Schilpp, P. A.: 1949, below.
Bohr, N.: 1961, 'Reminiscences of the founder of nuclear science and of some develop-
ments based on his work (Rutherford memorial lecture)', Proc.Phys.Soc. 78, 1083-
1115.
Bopp, F. (ed.): 1966, Werner Heisenberg und die Physik unserer Zeit (Braunschweig).
Born, M.: 1943,Experimentand Theory in Physics (Oxford).
Born, M.: 1949,Natural Philosophy of Cause and Chance (Oxford).
Born, M.: 1956, Physics in My Generation (Oxford).
Born, M.: 1963, 'Arnold Sommerfeld, 1868-1951', in Born, M.: 1963, Ausgewiihite
Abhandlungen (Section C above).
Born, M.: 1978, My Life. RecoUections of a Nobel Laureate (London).
 SELECT BIBLIOGRAPHY 169

Bromberg, J.: 1976, 'The concept of particle creation before and after quantum
mechanics', Hist.Stud.Phys. Sci. 7,161-183.
S. Brush,: 1976, 'Irreversibility and indeterminism: Fourier to Heisenberg', J.Hist.Ideas
37,603-630.
Brush, S.: 1980, The chimerical cat: philosophy of quantum mechanics in historical
perspective',Soc.Stud.Sci. 10,393-447.
Cassidy, D.: 1976, 'Werner Heisenberg and the crisis in quantum theory, 1920-1925',
Ph.D. dissertation, Purdue University.
Cassidy, D.: 1979, 'Heisenberg's flIst core model of the atom: the formation of a pro-
fessional style', Hist.Stud.Phys. Sci. 10,123-186.
Clark, R. W.: 1971, Einstein: The Life and Times (New York).
Cropper, W. H.: 1970, The Quantum Physicists (Oxford).
Dirac, P. A. M.: 1963, 'The evolution of the physicist's picture of nature', Sci.American
208,45-53.
Dirac, P. A. M.: 1971, The Development of Quantum Theory (New York).
Dirac, P. A. M.: 1977, 'Recollections of an exciting era', in Weiner, C.: 1977, below.
Earman J. and C. Glymour: 1978a, 'Einstein and Hilbert: two months in the history of
general relativity', Arch.Hist.Exact Sci. 17, 291-308.
Earman J. and C. Glymour: 1978b, 'Lost in the tensors: Einstein's struggles with
covariance principles, 1912-1916', Stud.Hist.Phil. Sci. 9,251-278.
Elkana, Y. (ed.): 1974, The Interaction between Science and Philosophy (Atlantic High-
lands, New Jersey).
Elsasser, W. M.: 1978, Memoirs of a Physicist in the Atomic Age (New York).
Enz, C. P.: 1973, 'W. Pauli's scientific work', in Mehra, J.: 1973, below.
Fierz, M.: 1974, 'Wolfgang Pauli', Dict.Sci.Biog. 10,422-425.
Fierz M. and V. S. Weisskopf (eds.): 1960, Theoretical Physics in the Twentieth Century
(New York).
Forman, P.: 1967, 'The environment and practice of atomic physics in Weimar Germany',
Ph.D. dissertation, University of California at Berkeley.
Forman, P.: 1968, 'The doublet riddle and atomic physics circa 1924',/sis 59, 156-174.
Forman, P.: 1970, 'Alfred Lande and the anomalous Zeeman Effect, 1919-1921',
Hist.Stud.Phys. Sci. 2,153-261.
Forman, P.: 1971, 'Weimar culture, causality, and quantum theory, 1918-1927', Hist.
Stud.Phys.Sci. 3, 1-116; republished in a shortened and in some ways improved
version in Chant, C. and J. Fauvel (eds.): 1980, Darwin to Einstein: Historical Studies
on Science and Belief (Harlow, Essex).
Forman, P.: 1978, 'The reception of an acausal quantum mechanics in Germany and
Britain', in Mauskopf, S. H. (ed.): 1980, The Reception of Unconventional Science
(Boulder, Colorado).
Frank, P.: 1947, Einstein, His Life and Times (New York).
Gerber, J.: 1969, 'Geschichte der Wellenmechanik',Arch.Hist.Exact Sci. 5,349-416.
Goldberg, S.: 1969, 'The Lorentz theory of the electron and Einstein's theory of rela-
tivity',Am.J.Phys. 37,982-994.
Goudsmit, S. A.: 1961, 'Pauli and nuclear spin', Phys. Today (June), 18-21.
Goudsmit, S. A.: 1976, 'It might as well as be spin', Phys. Today (June), 40-43.
Hanle, P. A.: 1977a, 'The coming of age of Erwin Schrodinger: his quantum statistics of
ideal gases', Arch.Hist.Exact Sci. 17, 165-192.
170 SELECT BIBLIOGRAPHY

Hanle, P. A.: 1977b, 'Schrooinger's reaction to de Broglie's thesis', Isis 68, 606~609.
Hanle, P. A.: 1979a, 'The Schrodinger-Einstein correspondence and the sources of wave
mechanics',Am.J.Phys. 47,644-648.
Hanle, P. A.: 1979b, 'Indeterminacy before Heisenberg: the case of Franz Exner and
Erwin Schrodinger', Hist.Stud.Phys. Sci. 10,225-270.
Hanson, N. R.: 1961, 'Are wave mechanics and matrix mechanics equivalent theories?',
in Feigl, H. and G. Maxwell (eds.): 1961, Cu"ent Issues in the Philosophy of Science
(New York).
Hanson, N. R.: 1963, The Concept of the Position (Cambridge).
Heelan, P. A.: 1975, 'Heisenberg and radical theoretic change', Zeit. aJlgem. Wissens-
chaftstheorie 6,113-136.
Heilbron, J. L.: 1964, 'A history of atomic structure from the discovery of the electron
to the beginning of quantum mechanics', Ph.D. dissertation, University of California
at Berkeley.
Heilbron, J. L.: 1967, 'The Kossel-Sommerfeld theory and the ring atom', Isis 58,451-482.
Heilbron, J. L. and T. S. Kuhn: 1969, 'The genesis of Bohr's atom', Hist.Stud.Phys.
Sci. 1,211-290.
Heisenberg, W.: 1960, 'Errinerungen an die Zeit der Entwicklung der Quantenmechanik',
in Fierz, M. and V. S. Weisskopf: 1960, above.
Heisenberg, W.: 1967, 'Quantum theory and its interpretation', in Rozental, S.: 1967,
below.
Heisenberg, W.: 1971, Physics and Beyond: Encounters and Conversations (New York).
Heisenberg, W.: 1973, 'Development of concepts in the history of quantum theory', in
Mehra, J.: 1973, below.
Heisenberg, W.: 1974, 'Wolfgang Pauli's philosophical outlook', in his 1974, Across the
Frontiers (New York).
Hendry, J.: 1980a, 'Weimar culture and quantum causality', Hist.Sci. 18,155-180.
Hendry, J.: 1980b, 'The development of the wave-particle duality of light and quantum
theory, 1900-1920', Ann.Sci. 37(1980),59-79.
Hendry, J.: 1981, 'Bohr-Kramers-Slater: a virtual theory of virtual oscillators and its role
in the history of quantum mechanics', Centaurus 25, 189-221.
Hermann, A.: 1962, 'Max Born', in Dokumente der Naturwissenschaften, Bd. 1 (Section
Cabove).
Hermann, A.: 1975, 'Erwin Schrodinger', Dict.Sci.Biog. 12,217-223.
Hermann, A.: 1976, Werner Heisenberg, 1901-1976 (Bonn).
Hesse, M.: 1961, Forces and Fields (London).
Hoffmann, B. and H. Dukas: 1972, Albert Einstein, Creator and Rebel (New York).
Holton, G.: 1973, The Thematic Origins of Scientific Thought (Cambridge, Mass.).
Holton, G.: 1978, The Scientific Imagination: Case Studies (Cambridge, Mass.).
Honner, J.: 1982, 'The transcendental philosophy of Niels Bohr', Stu d.Hist.Phil. Sci.
13,1-30.
Horz, H.: 1966,. Werner Heisenberg und die Philosophie (Berlin).
Hund, F.: 1974, The History of Quantum Theory (New York).
Jammer, M.: 1966, The Conceptual Development of Quantum Mechanics (New York).
Jammer, M.: 1974, The Philosophy of Quantum Mechanics (New York).
Klein, M. J.: 1959, 'Ehrenfest's contribution to the development of quantum statistics',
Kon. Ned. Akad. v. Wetenschappen B62, 41-62.
 SELECT BIBLIOGRAPHY 171

Klein, M. J.: 1964, 'Einstein and the wave-particle duality',Natural Philosopher 3,1-49.
Klein, M. J.: 1970a,Paul Ehrenfest: The Makingofa Theoretical Physicist (Amsterdam).
Klein, M. J.: 1970b, 'The first phase of the Bohr-Einstein dialogue', Hist.Stud.Phys. Sci.
2,1-39.
Klein, M. J.: 1977, 'The beginnings of quantum theory', in Weiner, C.: 1977, below.
Konro, H.: 1978, 'The historical roots of Born's probabilistic interpretation', Jap.Stud.
Hist.Sci. 17,129-145.
Kragh, H.: 1979a, 'On the history of early wave mechanics', IMFUFA tekst 23, Roskilde
University Centre.
Kragh, H.: 1979b, 'Methodology and philosophy of science in Paul Dirac's physics',
IMFUF A tekst 27, Rosklide University Centre.
Kragh, H.: 1979c, 'Niels Bohr's second atomic theory', Hist.Stud.Phys.Sci. 10,123-186.
Kragh, H.: 1982, 'Erwin SchrOdinger and the wave equation: the crucial phase',
Centaurus 26, 154-197.
Kramers, H. A. and H. Holst: 1923, The Atom and the Bohr Theory of its Structure
(Copenhagen).
Kronig, R.: 1960, 'The turning point', in Fierz, M. and V. S. Weisskopf: 1960, above.
Kubli, F.: 1970, 'Louis de Broglie und die Entdeckung der Materiewellen', Arch.Hist.
Exact Sci. 7, 26-78.
Kuznetsov, B.: 1965, Einstein (Moscow).
McCormmach, R.: 1970, 'Einstein, Lorentz and the electron theory',Hist.Stud.Phys.Sci.
2,41-87.
MacKinnon, E.: 1976, 'De Broglie's thesis: a critical retrospective', Am.J.Phys. 44,
1047-1055.
MacKinnon, E.: 1977, 'Heisenberg, models and the rise of matrix mechanics', Hist.Stud.
Phys.Sci. 8,137-188.
MacKinnon, E.: 1982, Scientific Explanation and Atomic Physics (Chicago).
Margenau, H.: 1944, 'The exclusion principle and its philosophical importance', Phil.
Sci. 11,187-208.
Mehra, J.: 1972, 'The golden age of theoretical physics', in Salam, A. and E. Wigner:
1972, below.
Mehra, J.: 1974, Einstein, Hilbert, and the Theory of Gravitation (Dordrecht).
Mehra, J. (ed.): 1973, The Physicist's Conception of Nature (Dordrecht).
Mehra, J. and H. Rechenberg: 1982-, The Historical Development of Quantum Theory
(New York): Volume 1 (in two parts), The Quantum Theory of Planck, Einstein,
Bohr and Sommerfeld; Volume 2, The Discovery of Quantum Mechanics; Volume 3,
The Formulation of Matrix Mechanics and Its Modifications; Volume 4 (in two
parts), The Fundamental Equations of Quantum Mechanics, and The Reception of
the New Quantum Mechanics. (Further volumes in preparation.)
von Meyenn, K.: 1980, 1981, 'Pauli's Weg zum Ausschiessungsprinzip', Phys. Bl. 36
(1980),293-298,37(1981),13-20.
Meyer-Abich, K.: 1965, Ko"espondenz. lndividualitiit und Komplementaritiit (Wies-
baden).
Miller, A. I.: 1978, 'Visualisation lost and regained', in Wechsler, 1. (ed.): 1978, On
Aesthetics in Science (Cambridge, Mass.).
Miller, A. I.: 1982, 'Beyond Anschaulichkeit', in Laszo Tisza Festschrift (Cambridge,
Mass.).
172 SELECT BIBLIOGRAPHY

Moore, R.: 1966, Niels Bohr (Copenhagen).

Mott, N. and R. Peierls: 1977, 'Heisenberg', Biog.Mem.Fell.Roy.Soc. 23,213-252.
Newman, M. H. A.: 1957, 'Weyl',Biog.Mem.Fell.Roy.Soc. 3,305-328.
Nisio, S.: 1973, 'The formation of the Sommerfeld quantum theory of 1916', Jap.Stud.
Hist.Sci. 12,39-78.
Nobel Foundation: 1964, Nobel Laureates in Physics, 1942-1962 (Amsterdam).
Pais, A.: 1979, 'Einstein and the quantum theory', Rev.Mod.Phys. 51,869-914.
Pais, A.: 1982, 'Subtle is the Lord ... ' The Science and Life ofAlbert Einstein (Oxford).
Peierls, R.: 1960, 'Paull', Biog.Mem.Fell.Roy.Soc. 5, 175-192.
Peterson, A.: 1963, 'The philosophy of Niels Bohr', Bull.Atom.Sci. 19, No.7 (Septem-
ber),8-14.
Pyenson, L.: 1974, 'The Gottingen reception of relativity', Ph.D. dissertation, The Johns
Hopkins University.
Radder, H.: 1982, 'Between Bohr's atomic theory and Heisenberg's matrix mechanics. A
study of the role of the Dutch physicist H. A. Kramers', Janus 69, 223-252.
Raman, V. V.: 1972, 'Relativity in the early twenties', Indian J.Hist.Sci. 7,119-145.
Raman, V. V. and P. Forman: 1969, 'Why was it Schrodinger who developed de Broglie's
ideas?',Hist.Stud.Phys.Sci. 1,291-314.
Reid, C.: 1970, Hilbert (London).
Richter, S.: 1979, Wolfgang Pauli (Aurau).
Robertson, P.: 1979, The Early Years. The Niels Bohr Institute, 1921-1930 (Copen-
hagen).
Rosenfeld, L.: 1936, 'La premiere phase de l'evolution de la theorie des quanta', Osiris
2,149-196.
Rosenfeld, L.: 1945, Nieis Bohr - An Essay (Amsterdam).
Rosenfeld, L.: 1970, 'Niels Bohr', Dict.Sci.Biog. 2,239-254.
Rozental, S. (ed.): 1967,Niels Bohr (Amsterdam).
Salam, A. and E. Wigner (eds.): 1972, Aspects of Quantum Theory (Cambridge).
Schilpp, P. A. (ed.): 1949, Albert Einstein: Philosopher-Scientist (Evanston, Illinois).
Scott, W. T.: 1967, Erwin Schrodinger: An Introduction to his Writings (Amherst,
Mass.).
Seelig, C.: 1956,Albert Einstein. A Documentary Biography (London).
Segre, E.: 1970, Enrico Fermi, Physicist (Chicago).
Segre, E.: 1980: From X-rays to Quarks (San Francisco).
Serwer, D.: 1977, 'Unmechanischer Zwang: Pauli, Heisenberg, and the rejection of the
mechanical atom, 1923-1925', Hist.Stud.Phys.Sci. 8,189-256.
Small, H. G.: 1971, 'The helium atom in the old quantum theory', Ph.D. dissertation,
University of Wisconsin at Madison.
Sopka, K. R.: 1980, Quantum Physics in America, 1920-1935 (New York).
Stolzenburg, K.: 1977, 'Die Entwicklung des Bohrschen Komplementaritatsgedankens in
den Jahren 1924 bis 1929', Ph.D. dissertation, University of Stuttgart.
Stuewer, R.: 1975, The Compton Effect - Turning Point in Physics (New York).
Tonnelat, M. A.: 1965, Les Theories Unitaires de I'Electromagnetism et de 10 Gravita-
tion (Paris).
Tonnelat, M. A.: 1973, 'L'influence de la relativite sur J'oevre de Louis de Broglie', in
Louis de Broglie. Sa Conception du Monde Physique (Paris).
Uhlenbeck, G.: 1976, 'Personal reminiscences', Phys. Today (June), 43-48.
 SELECT BIBLIOGRAPHY 173

van der Waerden, B. L.: 1960, 'Exclusion principle and spin', in Fierz, M. and V. S.
Weisskopf: 1960, above.
van der Waerden, B. L.: 1967, Sources of Quantum Mechanics (Amsterdam).
van der Waerden, B. L.: 1973, 'From matrix mechanics and wave mechanics to unified
quantum mechanics', in Mehra, J.: 1973, above.
Weiner, C. (ed.): 1977, History of Twentieth Century Physics (New York).
Wessels, L.: 1979, 'SchrOdinger's route to wave mechanics', Stud.Hist.Phil.Sci. 10,311-
340.
Weyl, H.: 1944, 'David Hilbert and his work', Bull.Am.Math.Soc. 50,612-654.
Wiener, N.: 1956,1 am an Mathenuztician (New York).
 INDEX

Abraham, Max 135 building up principle 40,41

absorption phenomena 6, 17 Burger, Hermann Carel 62
action principle 10-14
anomalous Zeeman effect 38, 40, 42- canonical equations of motion 71, 72
44,63,65 causality 3, 4, 14-18, 20, 30-35, 49,
Anschaulichkeit 5, 64, 87, 114, 118, 51, 55, 56, 89-93, 117, 121, 128,
124,165 132,138
aperiodic phenomena 81, 82 (see a/so classical concepts, role of 25, 26, 29,
collision theory) 30, 33, 34, 37, 38, 45, 52, 56, 64,
93,107,111-119,122-131
barrier penetration 57,70,92,119,121 classical mechanics, rejection of 22, 27,
Becker, Richard 56 33,36,40-43,59,64,65
black-body radiation 6, 7, 78, 79 Coleridge, Samuel Taylor 134
(see a/so Planck's law) collision theory 89, 90, 95-100, 102,
Blackett, Patrick M. S. 134 115,120,121
Bohr, Niels I, 5, 7, 8, 21, 24-43, commutation relations 71, 72, 74, 86,
45-58, 61,62, 64-66, 70, 75-77, 88,103,111,115
87, 95, 102, 103, 108, 113-115, complementarity 5,32,124-129,131
119-131, 140, 141, 146, 152, 153, completeness 92, 128
154 Compton, Arthur Holly 6, 37, 55, 134,
Bohr theory of the atom 4,7,19,24- 143
28,32,36,38-53,80,130-132 Compton effect 6,7,37,38,54,55,57,
Born, Max I, 8, 13, 14, 24, 36-40, 77,115,123,124,134,143
42-46, 49, 56-60, 63, 69-75, 79, conservation of energy and momentum
81-83,85,86,88-93,96-98,101, 4, 10, 16-18, 20, 30-38, 49, 51,
102, 105, 107, 114, 117, 119,121, 55-57,68,69,119-121
122, 126, 128, 132, 140, 146, 152, Copenhagen interpretation 1, 2, 5, 127,
153,154,155,156,157 129,132
Bose, Satyandra Nath 76,155,156 core theory 39-41, 43-45, 48, 50,63,
Bose-Einstein statistics 70, 79, 80, 131
94-96 Courant, Richard 145
Bothe, Walter 37,57,69,70,119,120 Cunningham, Ebenezar 54
Breit, Gregory 56,60
Bridgman, Percy W. 19,21 Darwin, Charles Galton 28, 31,33,47,
Brillouin, L. Marcel 101 124
Broglie, Louis de I, 7, 18, 52-54, Davisson, Clinton 70,119
57, 70, 78, 79, 83-87, 90, 103, Debye, Peter 2,37
127,155,157,164 difference equations 44, 45, 49, 58-
Broglie, Maurice de 52 60,73
Brush, Stephen G. 132 Dirac, Paul A. M. 1,5,7,8,87,93-96,

174
 INDEX 175

101-109, 113, 116, 117, 124, Grossmann 134,135

126-128, 130, 132, 155, 156,
157 half-integral quantum numbers 39-41,
discretisation programme 43-46, 49, 43,45,63,65
56,58,119 Hamilton, William Rowan 83
dispersion theory 46-49, 51, 56- 61, Hanle, Paul 19
69,150,151 Hartree, D. R. 134
Dorgelo, Hendrik B. 62 Heisenberg, Werner 1, 4, 5, 7, 8, 25,
double solution, theory of 164,165 35-37, 39,40,42-46,49,50,56-
Duane, William 37 81,86,87-96,101-108,111-131
Dymond, E. G. 134 Hellinger, Ernst 81
Helmholtz, Hermann von 132
Eddington, Arthur Stanley 10, 12, 15, hidden variables 89, 90
19,22,23,43,136-139 Hilbert, David 8, 12, 44, 73, 74, 108,
Ehrenfest, Paul 6, 36, 37,55,96, 124, 109,135
138 HMfding, Harald 29,32
Einstein, Albert 1, 7-18, 20, 26-28, Hund, Friedrich 102
30-32, 35-37, 39, 52, 54-56, 70,
78-80, 85, 90, 91, 96, 117, 119, imaginary constant, significance of 157
120, 127, 128, 131, 134-138, instrumentalism 132
148 interpretative 1, 2
electron diffraction 70, 119 Ishiwara, Jun 135
electron orbits, rejection of 20,25,26,
43,49,50,57,64,66,67,114-117 Jammer, Max 167
Ellett, Alexander 61,62 Jeans, James 6
Elsasser, Walter 162 Jones, J. E. 134
energy fluctuations 78, 79,96, 102 Jordan, Pascual 1, 7,56,70-75, 78-
Epstein, Paul S. 37, 38, 146 82, 87,94-96, 101, 105-108, 116,
exclusion principle 64, 80,94,95 117,124,126,127,148

Fermi, Enrico 80,93-95, 156 Kaluza, Theodor 103,122

Fermi-Dirac statistics 80,95,96, 156 Kant, Immanuel 132,133
field-particle problem 8, 13-17, 20 Kapitza, Peter 134
fluorescence polarisation 61, 62, 65, Kemble, Edwin 54
130 Kierkegaard, S¢ren 32
Forman, Paul 3,16,19,35,132 Klein, Oscar 103,122,124,127
Forsterling, Karl 18,19 Kramers, Hendrik Anthony 38, 43,46,
Fowler, Ralph Howard 53, 54, 56, 48, 50, 51, 53-61, 64, 69, 101,
147 147,148,149,150
Franck, James 24,70,73,74,90,120 Kronig, Ralph de Laer 73, 87
Fredholm, Erik Ivar 156 Kunsmann, Charles H. 70, 119
Frenkel, Jankov 89
Ladenburg, Rudolf 46-48, 56, 57,
Geiger, Hans 37, 57,69,70,119,120, 149
138 Lanczos, Cornel 93, 105
Gerlach, Walther 36 Lande, Alfred 26,36,39-41,45,63
Goudsmit, Samuel A. 150 Langmuir, Irving 38
176 INDEX

Larmor theorem 39,42,45 Planck's law 6,7, 17, 27, 31, 32,52,
light-quanta 34, 37, 38, 53-55, 61, 78, 79
78-80, 102 (see also wave-particle Poincare, Henri 6
duality of light) probability interpretation 103-110,
Lorentz, Hendrik Antoon 7, 54, 85, 114,116
127
quantum condition 18, 38, 44, 45, 59,
mathematical versus physical approach 60, 68, 69 j 70, 71 (see also quantum
73-75,107,108 postulate)
matrix mechanics 3, 7, 18, 19, 53, quantum field theory 1,80,127
69-80, 84-88, 90, 93-95, 98- quantu)ll postulate 8,25,125,126
105, 108, 109, 124 (see also new quantum statistics 6, 7, 75, 78, 80,
kinematics) 94, 95, 119, 129, 132 (see also
Maxwell, James Clerk 132 Bose-Einstein statistics, Fermi-Dirac
measurement in quantum theory 104, statistics)
113,114,122-128 Q-numbers 94
Mie, Gustav 10, 13-16, 135
Mises, Richard von 17, 33 Raman, V. V. 19
Ramsauer, Carl 57,70,119,120
Nernst, Walther 11 Reiche, Fritz 47,56
Neumann, John von 1, 108-110, 156, Reichenbach, Hans 91
157 relativity, general theory of 4, 7-16,
Newman, Maxwell H. A. 134 18,24,43,103,132
Nordheim, Lothar 108, 109 relativity, special theory of 18, 117,
Nordstrom, Gunnar 135 118

observability criterion 19, 50, 65-67, Schottky, Walter 17,18,91

69, 70, 74-78, 90, 93, 94, 99, 102, Schrodinger, Erwin 1, 7, 8, 17-19,
104, 107, 108, 114, 116, 117, 122, 30-32, 56, 82-97, 101-103, 121,
128 127, 129, 131, 132, 139, 157,158,
old quantum theory (see Bohr atomic 159
theory) Schrodinger interpretation 84,85
operationalism 5, 14, 15, 19-23, 29, Senftleben, Hans Albrecht 33, 117
34,64,66,67,77,83,86,107,108, Serwer, Daniel 44
114-118, 122,126, 129, 130, 137 Skinner, Herbert 134
operator calculus 81, 82, 85, 86, 88, Slater, John Clark 43, 51-55, 57,
106,156 146,147,148
orbital model 59, 61, 62, 65, 84 (see Smekal, Adolf 61
also Bohr atomic theory) Solvay congresses 28,30,33, 127
Ornstein, Leonard S. 62 Sommerfeld, Arnold 8, 9, 24, 26, 28,
oscillator technique 58-62 35-40, 42, 49, 50, 55, 65, 87,
134, 135, 143
phenomenological approach 22, 23,42, space quantisation 36
62-66 space-time description, rejection of 5,
Planck, Max 6, 11,31,32 29, 30, 33, 49, 70, 88-90, 120,
Planck-Bose statistics (see Bose-Einstein 121
statistics) space-time, discrete 113, 117
 INDEX 177

space-time, statistical 112 visualisation 5, 7, 29, 51, 121, 125,

Spengler, Oswald 91 129,134 (see also Anschaulichkeit)
spin 150 (see also Zweideutigkeit) Vleck, John H. van 38,40,54
statistical interpretation 87, 89-93,
97-101,103,107,116 Waerden, Barthel L. van der 150
statistical weights problem 41, 44, 49, wave mechanics 3, 5, 7, 18, 19, 53,
62,80 83-92, 95, 98, 101, 105, 108, 109,
Stern, Otto 36, 96 124,127,129
Stokes's law 52 wave-particle duality of light 4, 6-8,
Stoner, Edmund C. 55,63,134 13-22, 27-33,36,52,90,91, 119,
123-126,141
Thomson, John Joseph 25 wave-particle duality of matter 70, 80,
transformation theory 5, 72-75, 80, 119,123-126
81,111,114,115,117,131 wave theory of matter 53,57, 70,78,
transition probabilities 26-28, 32, 55, 86-88,90,121,153,154,155
71,77,81,92 Wentzel, Gregor 101
Weyl, Hermann 7-19, 23, 50, 74,91,
Uhlenbeck, George E. 150 108,122,130,138
uncertainty principle 5, 77, 80, 114- Whittaker, Edmund T. 10
118,122-126,131 Wien, Wilhelm 52,85
unified field theories 4, 7 -16, 19, 24, Wiener, Norbert 81, 82, 85, 86, 88,
31,50,103,122 92,157
unmechanischer Zwang 41, 42,48,49, Wood, Robert w. 61,62
63
X-ray phenomena 6, 7, 48, 52 (see also
virtual oscillators 50-54, 56, 57, 61, Compton effect)
63, 68, 70, 78, 93, 119, 121, 130,
148, 153 Zweideutigkeit 63-66
 STUDIES IN THE HISTOR Y
OF MODERN SCIENCE

Editors:

ROBERT S. COHEN (Boston University)

ERWIN N. HIEBERT (Harvard University)
EVERETT I. MENDELSOHN (Harvard University)

1. F. GregorY,Scientific Materialism in Nineteenth Century Germany. 1977, xxiii+

279 pp.
2. E. Manier, The Young Darwin and His Cultural Circle. 1978, xii+242 pp.
3. G. Canguilhem, On the Normal and the Pathological. 1978, xxiv+ 208 pp.
4. H. C. Kennedy, Peano. 1980, xii+230 pp.
5. L. Badash, J. O. Hirschfelder, and H. P. Broida (eds.), Reminiscences of Los Alamos,
1943-1945. 1980, xxi+188 pp.
6. L. L. Bucciarelli and N. Dworsky, Sophie Germain: An Essay in the History of the
Theory of Elasticity. 1980, xi+147 pp.
7. D. J. Struik, The Land of Stevin and Huygens. 1981, xiv+162 pp.
8. I. Langham, The Building of British Social Anthropology. 1981, xxxii+392 pp.
9. Z. Bechler (ed.), Contemporary Newtonian Research. 1982, vi+241 pp.
10. W. T. Sullivan, III, Classics in Radio Astronomy. 1982, xxiv+348 pp.
11. L. Anderson, Charles Bonnet and the Order of the Known. 1982, xvi+160 pp.
12. P. L. Farber, The Emergence of Ornithology as a Scientific Discipline: 1760-1850.
1982, xxi+191 pp.
13. T. Lenoir, The Strategy of Life. 1983, xii+314 pp.