Instant Download A Career In Theoretical Physics 2nd Edition P. W. Anderson PDF All Chapters

Download the full version of the ebook at ebookfinal.
com
A Career In Theoretical Physics 2nd Edition P. W.

Anderson
https://ebookfinal.com/download/a-career-in-theoretical-
physics-2nd-edition-p-w-anderson/
OR CLICK BUTTON
DOWNLOAD EBOOK
Download more ebook instantly today at https://ebookfinal.com

Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...
Theoretical Nuclear And Subnuclear Physics 2nd Edition

John Dirk Walecka
https://ebookfinal.com/download/theoretical-nuclear-and-subnuclear-
physics-2nd-edition-john-dirk-walecka/
ebookfinal.com
Physics and Theoretical Computer Science From Numbers and

Languages to Quantum Cryptography J.-P. Gazeau
https://ebookfinal.com/download/physics-and-theoretical-computer-
science-from-numbers-and-languages-to-quantum-cryptography-j-p-gazeau/
ebookfinal.com
The Ghost in the Atom A Discussion of the Mysteries of

Quantum Physics 1st Edition P. C. W. Davies
https://ebookfinal.com/download/the-ghost-in-the-atom-a-discussion-of-
the-mysteries-of-quantum-physics-1st-edition-p-c-w-davies/
ebookfinal.com
Condensed Matter Physics 2nd Edition Michael P. Marder
https://ebookfinal.com/download/condensed-matter-physics-2nd-edition-
michael-p-marder/
ebookfinal.com
Handbook of Career Counseling for Women 2nd Edition W.
Bruce Walsh
https://ebookfinal.com/download/handbook-of-career-counseling-for-
women-2nd-edition-w-bruce-walsh/
ebookfinal.com
Musculoskeletal Imaging Cases Cases in Radiology Mark W.

Anderson
https://ebookfinal.com/download/musculoskeletal-imaging-cases-cases-
in-radiology-mark-w-anderson/
ebookfinal.com
Equations in Mathematical Physics A practical course 1st

Edition Victor P. Pikulin
https://ebookfinal.com/download/equations-in-mathematical-physics-a-
practical-course-1st-edition-victor-p-pikulin/
ebookfinal.com
Differentiable Manifolds A Theoretical Physics Approach

1st Edition Gerardo F. Torres Del Castillo (Auth.)
https://ebookfinal.com/download/differentiable-manifolds-a-
theoretical-physics-approach-1st-edition-gerardo-f-torres-del-
castillo-auth/
ebookfinal.com
Strings Branes and Gravity 1st Edition Theoretical

Advanced Study Institute In Elementary Particle Physics
https://ebookfinal.com/download/strings-branes-and-gravity-1st-
edition-theoretical-advanced-study-institute-in-elementary-particle-
physics/
ebookfinal.com
r
World Scientific Series in 20th Century Physics - Vol. 35
A CAREER IN
THEORETICAL
PHYSICS
Second Edition
P H I L I P W. A N D E R S O N
A CAREER IN
THEORETICAL
PHYSICS
World Scientific Series in 20th Century Physics
Published
Vol. 21 Spectroscopy with Coherent Radiation — Selected Papers of Norman F. Ramsey

(with Commentary)
edited by N. F. Ramsey
Vol. 22 A Quest for Symmetry — Selected Works of Bunji Sakita
edited by K, Kikkawa, M. Virasoro and S. R. Wadia
Vol. 23 Selected Papers of Kun Huang (with Commentary)
edited by B.-F. Zhu
Vol. 24 Subnuclear Physics — The First 50 Years: Highlights from Erice to ELN
by A. Zichichi
edited by O. Barnabei, P. Pupillo and F. Roversi Monaco
Vol. 25 The Creation of Quantum Chromodynamics and the Effective Energy
by V. N. Gribov, G. 't Hooft, G. Veneziano and V. F. Weisskopf
edited by L N. Lipatov
Vol. 26 A Quantum Legacy — Seminal Papers of Julian Schwinger
edited by K. A. Milton
Vol. 27 Selected Papers of Richard Feynman (with Commentary)
edited by L M. Brown
Vol. 28 The Legacy of Leon Van Hove
edited by A. Giovannini
Vol. 29 Selected Works of Emil Wolf (with Commentary)
edited by E. Wolf
Vol. 30 Selected Papers of J. Robert Schrieffer — In Celebration of His 70th Birthday
edited by N. E. Bonesteel and L. P. Gor'kov
Vol. 31 From the Preshower to the New Technologies for Supercolliders — In Honour
of Antonino Zichichi
edited by B. H. Wiik, A. Wagner and H. Wenninger
Vol. 32 In Conclusion — A Collection of Summary Talks in High Energy Physics
edited by J. D. Bjorken
Vol. 33 Formation and Evolution of Black Holes in the Galaxy
— Selected Papers with Commentary
edited by H. A. Bethe, G. E. Brown and C.-H. Lee
Vol. 35 A Career in Theoretical Physics, 2nd Edition
by P. W. Anderson
For information on Vols. 1-20, please visit http://www.worldscibooks.com/series/wsscp_series.shtml

A CAREER IN
THEORETICAL
PHYSICS
P H I L I P W. A N D E R S O N
Princeton University, USA
YJ? World Scientific

NEWJERSEY • LONDON • SINGAPORE • BEIJING • SHANGHAI • HONGKONG • TAIPEI • CHENNAI
Published by
World Scientific Publishing Co. Pte. Ltd.
5 Toh Tuck Link, Singapore 596224
USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE
The author and publisher would like to thank the following publishers for their permission to
reproduce the reprinted papers found in this volume:
American Association for the Advancement of Science (Science)

American Institute of Physics (AIP Conf. Proc, Physics Today)
American Physical Society (Phys. Rev., Phys. Rev. Lett., Rev. Mod. Phys.)
Birkhauser Verlag (Ann. Henri Poincare)
EDP Sciences (J. de Physique Collogue)
Elsevier Science Publishers (J. Phys. Chem. Solids, Solid State Commun., Physica, Phys. Reports,
Stud. Hist. Phil. Mod. Phys.)
Institute of Physics (J. Phys. F, J. Phys. Condens. Matter)
Perseus Books (Emerging Syntheses in Science)
Springer-SBM (Valence Instabilities and Related Narrow Band Phenomena,
Proc. of 1983 NATO/CAP Inst. "Moment Formation in Solids", The Universality of Physics:
A Festschrift in honor of Deng Feng Wang)
Taylor and Francis [http://www.tandf.co.uk/journals] (Philosophical Magazine, Solid State Physics,
Comments on Solid State Physics, Transition Metal Compounds)
To those publishers who have not granted us permission before publication, we have taken the liberty
to reproduce their articles without consent. However we shall acknowledge them in future editions of
this work.
Library of Congress Cataloguing-in-Publication Data

Anderson, P. W. (Philip W.), 1923-
A career in theoretical physics (2nd ed.) / Philip W. Anderson.
p. cm. — (World Scientific series in 20th century physics; v. 35)
ISBN 981-238-865-6 - ISBN 981-238-866-4 (pbk)
1. Mathematical physics. I. Title. II. Series.
QC20.A55 2004 2004-063652
530.1-dc22 CIP
Copyright © 2004 this edition by World Scientific Publishing Co. Pte. Ltd.
All rights reserved.
Printed in Singapore.
PREFACE TO THE SECOND EDITION
The plan which informed the selection of papers in the original volume was to stick as
far as possible to summaries and reviews, since there was and is no possibility of
covering all possible subjects. I think this may not have been a very good idea, since it
slights the original research papers with their greater historical value. This volume is
the only summary of my career likely to be published — after all, consider its title —
and therefore I have omitted some of these reviews, and substituted some of the
research papers — for instance, the paper on line broadening which was a summary of
my thesis, and the original paper on localization theory.
There is a second somewhat more embarrassing reason for this revision. At the
time of the earlier edition, I was gung-ho for a theory of the mechanism of high Tc
superconductivity which was shortly proven not to be the primary one. Among the
"Central Dogmas" which I proudly reprinted in the book, number V was just plain
wrong. I now belatedly understand what led me astray, but only because I was
conclusively shown to be wrong, in the gentlest possible manner, by the careful
experimental work of several colleagues, notably Kam Moler and collaborators.
Fortunately, the correct theory is also partly based on the ideas I expressed in 1987, so
not all is lost; but the papers in this section — involving nearly ten years of my
professional life — have had to be mostly replaced, and some of the commentary
rewritten.
v
INTRODUCTORY ESSAY TO THE FIRST EDITION
It was in the winter of 1966-7 that I first tried to express my philosophy of what was
important in science — or at least, of what was important that was possible for me. I
has been invited to spend a month in the pleasant climate of La Jolla as the Regents'
lecturer at the UCSD, visiting a department staffed by many of my old friends from
Bell Labs, such as Bernd Matthias, Harry Suhl, and George Feher, who had been
recruited by Walter Kohn, who was also by courtesy a Bell alumnus. One of the
lectures I gave was called "More is Different". The original version, now lost, went
over with the audience like a lead balloon, at least according to that portion of the
audience most closely related to me. But a cleaned up version, which was published in
Science in 1971, has been surprisingly influential. I have chosen to reprint it, out of
chronological order, as an additional introduction to this collection.
I already had this overall vision, if only very vaguely, as I completed my thesis
and other early work on the breadths of spectral lines. This work was among the first
to approach what was normally thought of as a dissipative process and dealt with by
Boltzmann equations for the microscopic variables, in terms of the macroscopic
moments of the system as a whole. In my own mind called it the "one big molecule"
method. I was beginning to see that somehow, starting from the fundamental atomic-
level laws of quantum theory, and without any more or less mystical appeals to the
difference between the atomic domain and the macroscopic domain where
deterministic Newtonian physics held way, one could really understand many of the
properties of matter in that macroscopic domain. I also experienced two other insights:
that these properties were often extraordinary; and that this understanding was often
intellectually challenging and deep, at a level which was for me at least more exciting
and worthwhile than any other purely intellectual exercise.
Thus I was launched into the search for the "Aufbauprinzips" of the everyday
world. I admired Frohlich's book on dielectric theory because it took the same point
of view; I encountered it in my early work on ferroelectrics. Much more important, I
came to see the crucial importance of the concept of broken symmetry, which has been
a lifelong interest. Broken symmetry is the clearest instance of the process of
emergence which lies behind "More is different". My interest began with the
antiferromagnetic ground state work in 1952-3, from which I developed the concept
of the Goldstone boson. In applying similar thinking to superconductivity, the Higgs
VII
phenomenon and the concept of Generalized Rigidity arose, the latter referring to the
similarities between superfluidity, superconductivity, and elastic rigidity of all sorts.
The topological theory of defects and dissipation is another facet of the conceptual
structure of broken symmetry.
The first natural direction in which to seek for understanding of the real world
is the explicit quantitative calculation of properties of real materials. This enterprise is
particularly rewarding when the basic principles which govern these properties are
new, or when the effects are subtle. This has been the mainspring of much of my work,
from the early work on magnetic resonance and other spectra, to superexchange, the
microscopic theory of superconductivity, and even work on the electronic structure of
solids, foreshadowing the corrections to local density theory, and a paper where the
idea is to develop a method which explains and exploits the strong locality of the
chemical bond, perhaps the most fundamental observation in the field of chemistry,
which nonetheless seems to interest quantum chemists not at all.
In 1956, stimulated by a pioneering paper on percolation by Hammersley
and by some puzzling experiments, I started to look at the next level of complexity,
those phenomena which are intrinsic to disorder. The first result was the theory of
localisation of waves. This was slow of acceptance, and, perhaps discouraged by
this fact, it was not until 1969 that I returned to it and also began to look at other
disorder phenomena such as glass and spin glass. Spin glass stimulated a very fertile
period of worldwide activity, first in the development of totally novel methodology,
such as a new statistical mechanics of nonergodic systems, and then in applications
of the new methodology to a "cornucopia" of new ideas and concepts and problems:
neural networks, computer algorithms, rugged evolutionary landscapes, even
immunology. This has left me involved in such fields as economics and the origin
of life. The interest in complex, emergent phenomena made it natural for me to
participate at an early stage in the Santa Fe Institute, founded on the twin principles of
interdisciplinarity and of the importance and tractability of genuinely complex
systems.
Left behind, in my old field of quantum condensed matter physics, was a worry
about the validity and completeness of the theories of many-body quantum systems
which had occupied that world in the 1950's and 60's. My work on the Anderson
model of magnetic impurities had led on to the "infrared catastrophe" in Fermi
systems which was the key to my work on the Kondo phenomenon which solved that
isolated many-body system but only at the cost of going beyond the conventional
perturbation theory. I was also disturbed by a class of strongly interacting
superconducting materials, as well as our notable lack of success in dealing with the
field of quantum valence fluctuations, i.e. "heavy electrons". With the appearance of
VIII
the high Tc cuprates in 1987, my apprehensions became manifestly justified, and I
began to formulate theories of metals based on new principles. This is the area which
now most absorbs me, and where my efforts have been rewarded with the discovery of
several new and fascinating phenomena. I reach my 70th year still involved in two
major, very different types of problem. Although there is no easy way to round this one
off, I have tried to give a flavor of the ideas involved.
My original vision seems to me to have been sound. The problem of under-
standing the "here and now" world seems to show no sign of being exhausted. The
contribution of physics is the method of dealing correctly both with the substrate from
which emergence takes place, and with the emergent phenomenon itself. Examples are
legion: for one, superconductivity cannot be properly understood simply as a
phenomenology without understanding electrons and their interactions, nor on the
other hand as a property of a small number of electrons without taking into account the
macroscopic system. Ever newer insights into the nature of the world around us will
continually arise from this style of doing science. There will be physics even in a world
without the SSC.
Regarding the choice of articles in this book: the sheer volume of either a
complete selection of research articles or, even more, a complete set of reviews would
have been much too great. I have given preference to materials which are for one
reason or another not easy of access: lecture notes and out of print sources, for
example. I have tried but failed to provide a selection of all major topics; the book is
long enough. I have included a few items on history as I saw it, and on general aspects
of science.
Oxford, December 2, 1993
IX
CONTENTS
Preface to the Second Edition v
Introductory Essay to the First Edition vii
More Is Different 1
Science 111, 393-396 (1972)
1. Pressure Broadening in the Microwave and Infrared Regions 5

Phys. Rev. 76, 647-661 (1949)
2. Theory of Ferroelectric Behavior of Barium Titanate 21

Ceramic Age 57, 29-34 (1951)
3. Use of Stochastic Methods in Line-Broadening Problems 29

Kyoto Lectures, March 1954
4. An Approximate Quantum Theory of the Antiferromagnetic Ground State 51

Phys. Rev. 86, 694-701 (1952)
5. Qualitative Considerations on the Statistics of the Phase Transition in

BaTi03-type Ferroelectrics 61
Conference Proceedings, Lebedev Physics Institute, USSR Academy of
Sciences, Nov. 1958, Fizika Dielektrikov (Moscow), 1960, pp. 290-297
6. Ordering and Antiferromagnetism in Ferrites 71

Phys. Rev. 102, 1008-1013 (1956)
7. Absence of Diffusion in Certain Random Lattices 79

Phys. Rev. 109, 1492-1505 (1958)
8. New Method in the Theory of Superconductivity 95

Phys. Rev. 110, 985-986 (1958)
XI
9. New Approach to the Theory of Superexchange Interactions 99
Phys.Rev. 115,2-13(1959)
10. Theory of Dirty Superconductors 113

J. Phys. Chem. Solids 11, 26-30 (1959)
11. Calculation of the Superconducting State Parameters with Retarded

Electron-Phonon Interaction 119
(with P. Morel)
Phys. Rev. 125, 1263-1271 (1962)
12. Generalized Bardeen-Cooper-Schrieffer States and the Proposed

Low-Temperature Phase of Liquid He 3 129
(with P. Morel)
Phys. Rev. 123, 1911-1934 (1961)
13. Localized Magnetic States in Metals 155

Phys. Rev. 124, 41-53 (1961)
14. Plasmons, Gauge Invariance, and Mass 169

Phys. Rev. 130, 439-442 (1963)
15. Hard Superconductivity: Theory of the Motion of Abrikosov Flux Lines 175
(with Y.B. Kim)
Rev. Mod. Phys. 36, 39-43 (1964)
16. Probable Observation of the Josephson Superconducting

Tunneling Effect 181
(with J. M. Rowell)
Phys. Rev. Lett. 10, 230-232 (1963)
17. Image of the Phonon Spectrum in the Tunneling Characteristic between

Superconductors 185
(with J. M. Rowell and D.E. Thomas)
Phys. Rev. Lett. 10, 334-336 (1963)
18. Exchange in Magnetic Insulators 189

in Transition Metal Compounds, ed. E.R. Schatz, Buhl Int. Conference on
Materials, Pittsburgh (Gordon and Breach, New York), 1964, pp. 17-28
XII
19. Superconductivity (Two Opinions) 207
(with B.T. Matthias)
Science, 144, 373-381 (1964)
20. Coherent Matter Field Phenomena in Superfluids 219

in Some Recent Definitions in the Basic Sciences, Vol. 2,
(Belfer Graduate School of Sciences, Yeshiva University,
New York, 1969), ed. A. Gelbart, pp. 21-^10
21. Considerations on the Flow of Superfluid Helium 241

Rev. Mod. Phys. 38, 298-310 (1966)
22. Multiple-Scattering Theory and Resonances in Transition Metals 255

(with W.L. McMillan)
in Proc. of the Int. School of Physics 'Enrico Fermi' XXXVII
(Academic Press, New York), 1967, pp. 50-86
23. Infrared Catastrophe in Fermi Gases with Local Scattering Potentials 293
Phys. Rev. Lett. 18, 1049-1051 (1967)
24. The Kondo Effect I 297

Comments on Solid State Physics 1, no. 2, 31-36 (1968-69)
The Kondo Effect II

Comments on Solid State Physics 1, no. 6, 190-197 (1968-69) 304
Kondo Effect III: The Wilderness — Mainly Theoretical 312

Comments on Solid State Physics 3, no. 6, 153-158 (1971)
Kondo Effect IV: Out of the Wilderness 318

Comments on Solid State Physics 5, no. 3, 73-79 (1973)
25. Superconductivity in the Past and the Future 325

in Superconductivity, Vol. 2, ed. R. Parks (Marcel Dekker, New York),
1968, pp. 1343-1358
26. Macroscopic Coherence and Superfluidity 343

in Contemporary Physics, Vol. 1 (IAEA, Vienna), 1969, pp. 47-54
XIII
27. The Fermi Glass: Theory and Experiment 353
Comments on Solid State Physics 2, 193-198 (1970)
28. Space-Time and Scaling Techniques in the Kondo Problem 361

in Proc. of 12th Int. Conf. on Low Temperature Physics, ed. E. Kanda
(Academic Press, 1971), pp. 657-660
29. Anomalous Low-temperature Thermal Properties of Glasses and

Spin Glasses 367
(with B.I. Halperin and C M . Varma)
Philos. Mag. 25, 1-9 (1972)
30. Comments on the Maximum Superconducting Transition Temperature 377

(with M.L. Cohen)
AIP. Conf. Proc, 1971, pp. 17-27
Comment on "Model for an Exciton Mechanism of Superconductivity" 389

(with J.C. Inkson)
Phys. Rev. B. 8, 4429-4432 (1973)
31. Resonating Valence Bonds: A New Kind of Insulator? 393

Materials Research Bulletin 8, 153-160 (1973)
32. Anisotropic Superfluidity in 3 He: A Possible Interpretation of

Its Stability as a Spin-Fluctuation Effect 403
(with W.F. Brinkman)
Phys. Rev. Lett. 30, 1108-1111 (1973)
33. Conference Summary 409

in Collective Properties of Physical Systems, eds. B. Lundqvist and
S. Lundqvist, Proc. of Nobel Symposium, Goteborg, Sweden,
13 June 1973 (Academic Press, 1974), pp. 266-271
34. Asymptotically Exact Methods in the Kondo Problem 417

(with G. Yuval)
in Magnetism, vol. V, ed. H. Suhl (Academic Press, 1973), pp. 217-236
XIV
35. Many-Body Effects at Surfaces 439
in Elementary Excitations in Solids, Molecules and Atoms
(Plenum, 1974), pp. 1-29
36. Conductivity from Charge or Spin Density Waves 469

(with P.A. Lee and T.M. Rice)
Solid State Communications 14, 703-709 (1974)
37. Uses of Solid State Analogies in Elementary Particle Theory 477

in Proc. ofConf. on Gauge Theories and Modern Field Theory,
eds. R. Arnowitt and P. Nath (MIT Press, 1976), pp. 311-335
38. Possible Consequences of Negative U Centers in Amorphous Materials 503

J. de Physique Colloque, No. 4, 339-342 (1976)
39. Theory of Spin Glasses 509

(with S.F. Edwards)
J. Phys. F. 5, 965-974 (1975)
40. Solution of "Solvable Model of a Spin Glass" 521

(with D.J. Thouless and R.G. Palmer)
Philos. Mag. 35, 593-601 (1977)
41. Phase Slippage without Vortex Cores: Vortex Textures in Superfluid 3 He 531
(with G. Toulouse)
Phys. Rev. Lett. 18, 508-511 (1977)
42. Scaling Theory of Localization: Absence of Quantum Diffusion in

Two Dimensions 537
(with E. Abrahams, D.C. Licciardello and T.V. Ramakrishnan)
Phys. Rev. Lett. 42, 673-676 (1979)
43. Some General Thoughts about Broken Symmetry 543

in Symmetries and Broken Symmetries in Condensed Matter Physics,
ed. N. Boccara (IDSET, Paris, 1981), pp. 11-20
xv
44. The Rheology of Neutron Stars: Vortex Line Pinning in the Crust
Superfluid 555
(with M.A. Alpar, D. Pines and J. Shaham)
Philos. Mag. A45, 227-238 (1982)
45. Localization Redux 569

Physica 117B & 118B, 30-35 (1983)
46. New Method for Scaling Theory of Localization. II: Multi-Channel

Theory of a "Wire" and Possible Extension to Higher Dimensionality 577
Phys. Rev. B. 23, 4828-4836 (1981)
47. Definition and Measurement of the Electrical and Thermal Resistances 587
(with H.-L. Engquist)
Phys. Rev. B. 24, 1151-1154 (1981)
48. Suggested Model for Prebiotic Evolution: The Use of Chaos 593
Proc. Natl. Acad. Sci. (USA) 80, 3386-3390 (1983)
49. Chemical Pseudopotentials 599

Phys. Reports 110, Nos. 5&6, 311-319 (1984)
50. Spin Glass Hamiltonians: A Bridge between Biology, Statistical

Mechanics and Computer Science 609
in Emerging Syntheses in Science, ed. D. Pines
(Addison-Wesley, 1987), pp. 17-20
51. Measurement in Quantum Theory and the Problem of Complex

Systems 615
in The Lessons of Quantum Theory, ed. J. de Boer, E. Dal and
O. Ulfbeck, talk given at the Niels Bohr Centenary Symposium,
Copenhagen, October 1985 (Elsevier Science, 1986), pp. 23-34
52. It's Not Over Till the Fat Lady Sings 629
Talk given at History of Superconductivity, APS Meeting, March 1987
53. Spin Glass I: A Scaling Law Rescued 639

Physics Today 41#1, 9 (1988)
XVI
Spin Glass II: Is There a Phase Transition? 642
Spin Glass III: Theory Raises Its Head 643

Spin Glass IV: Glimmerings of Trouble 645

Spin Glass V: Real Power Brought to Bear 647

Spin Glass VI: Spin Glass as Cornucopia 649

Spin Glass VII: Spin Glass as Paradigm 651

54. Epilogue 653

in Valence Instabilities and Related Narrow-band Phenomena,
ed. R.D. Parks (Plenum, 1977), pp. 389-396
55. Present Status of Theory: l/N Approach 662

in Proc. of 1983 NATO/CAP Inst. "Moment Formation in Solids",
ed. W.J.L. Buyers (Plenum), pp. 313-326
56. The Problem of Fluctuating Valence in f-Electron Metals 676

in Windsurfing in the Fermi Sea, Vol. 1, eds. T.T.S. Kuo and J. Speth
(Elsevier Science, 1987), pp. 61-67
57. Gutzwiller-Hubbard Lattice-Gas Model with Variable Density:

Application to Normal Liquid 3 He 683
(with D. Vollhardt and P. Woffle)
Phys. Rev. B. 35, 6703-6715 (1987)
58. Some Ideas on the Aesthetics of Science 697

Lecture given at the 50 th Anniversary Seminar of the Faculty of
Science and Technology, Keio University, Japan, May 1989
XVII
59. Theoretical Paradigms for the Sciences of Complexity 712
Nishima Memorial Lecture, Department of Physics, Keio University,
Japan, May 1989
60. 50 Years of the Mott Phenomenon: Insulators, Magnets, Solids, and

Superconductors as Aspects of Strong-Repulsion Theory 723
in Frontiers and Borderlines in Many-Particle Physics,
Proceedings of the Enrico International School of Physics,
Varenna, July 1987 (North-Holland, 1987), pp. 1-40
61. Theories of Fullerene T c 's Which Will Not Work 765
62. The Reverend Thomas Bayes, Needles in Haystacks, and the Fifth Force 775
63. The Eightfold Way to the Theory of Complexity: A Prologue 779

in Complexity, eds. G. Cowan, D. Pines and D. Meltzer
(Addison-Wesley, 1994), pp. 7-16
64. Magnetic Field Induced Confinement in Strongly Correlated

Anisotropic Materials 789
(with S.P. Strong and D.G. Clarke)
Phys. Rev. Lett. 73, 1007-1010 (1994)
65. Physics: The Opening to Complexity 795

Proc. Natl. Acad. Sci. (USA) 92, 6653-6654 (1995)
66. Beyond Chaos: Singular Distributions and Power Laws 799
67. Essay Review — Science: A 'Dappled World' or a 'Seamless Web'? 809

Stud. Hist. Phil. Mod. Phys. 32, 487-494 (2001)
68. RVB Redux: A Synergistic Theory of High T c Cuprates 819

in The Universality of Physics, A Festschrift in Honor of
Deng Feng Wang, ed. Khuri et al. (Kluwer/Plenum, 2001), pp. 3-8
XVIII
69. Physics of the Pseudogap Phase of High T c Cuprates, or,
RVB meets Umklapp 827
J. Phys. Chem. Solids 63, 2145-2148 (2002)
70. In Praise of Unstable Fixed Points: The Way Things Actually Work 833
Physica 5 318, 28-32 (2002)
71. Rise of Complexity, 1953-2002 839

Ann. Henri Poincare 4, S1-S6 (2003)
72. The Physics behind High-temperature Superconducting Cuprates:

The 'Plain Vanilla' Version of RVB 847
(with P.A. Lee, M. Randeria, T. M. Rice, N. Trivedi and F.C. Zhang)
J. Phys. Condens. Matter 16, R755-R769 (2004)
XIX
This page is intentionally left blank
Reprinted from
4 August 1972, Volume 177, pp. 393-396
less relevance they seem to have to the

very real problems of the rest of sci-
ence, much less to those of society.
The constructionist hypothesis breaks
down when confronted with the twin
More Is Different difficulties of scale and complexity. The
behavior of large and complex aggre-
gates of elementary particles, it turns
Broken symmetry and the nature of out, is not to be understood in terms
of a simple extrapolation of the prop-
the hierarchical structure of science. erties of a few particles. Instead, at
each level of complexity entirely new
properties appear, and the understand-
P. W. Anderson ing of the new behaviors requires re-
search which I think is as fundamental
in its nature as any other. That is, it
seems to me that one may array the
sciences roughly linearly in a hierarchy,
The reductionist hypothesis may still planation of phenomena in terms of
known fundamental laws. As always, dis- according to the idea: The elementary
be a topic for controversy among phi- entities of science X obey the laws of
tinctions of this kind are not unambiguous,
losophers, but among the great majority but they are clear in most cases. Solid science Y.
of active scientists I think it is accepted state physics, plasma physics, and perhaps
without question. The workings of our also biology are extensive. High energy
physics and a good part of nuclear physics X Y
minds and bodies, and of all the ani-
mate or inanimate matter of which we are intensive. There is always much less solid state or elementary particle
intensive research going on than extensive. many-body physics physics
have any detailed knowledge, are as- chemistry many-body physics
Once new fundamental laws are discov-
sumed to be controlled by the same set ered, a large and ever increasing activity molecular biology chemistry
of fundamental laws, which except begins in order to apply the discoveries to cell biology molecular biology
under certain extreme conditions we hitherto unexplained phenomena. Thus,
feel we know pretty well. there are two dimensions to basic re-
search. The frontier of science extends all
It seems inevitable to go on uncrit- along a long line from the newest and most psychology physiology
ically to what appears at first sight to modern intensive research, over the ex- social sciences psychology
be an obvious corollary of reduction- tensive research recently spawned by the
ism: that if everything obeys the same intensive research of yesterday, to the But this hierarchy does not imply
fundamental laws, then the only sci- broad and well developed web of exten- that science X is "just applied Y." At
sive research activities based on intensive each stage entirely new laws, concepts,
entists who are studying anything really research of past decades.
fundamental are those who are working and generalizations are necessary, re-
on those laws. In practice, that amounts The effectiveness of this message may quiring inspiration and creativity to just
to some astrophysicists, some elemen- be indicated by the fact that I heard it as great a degree as in the previous one.
tary particle physicists, some logicians quoted recently by a leader in the field Psychology is not applied biology, nor
and other mathematicians, and few of materials science, who urged the is biology applied chemistry.
others. This point of view, which it is participants at a meeting dedicated to In my own field of many-body phys-
the main purpose of this article to "fundamental problems in condensed ics, we are, perhaps, closer to our fun-
oppose, is expressed in a rather well- matter physics" to accept that there damental, intensive underpinnings than
known passage by Weisskopf (1): were few or no such problems and that in any other science in which non-
nothing was left but extensive science, trivial complexities occur, and as a re-
Looking at the development of science which he seemed to equate with device sult we have begun to formulate a
in the Twentieth Century one can dis- engineering. general theory of just how this shift
tinguish two trends, which I will call from quantitative to qualitative differ-
"intensive" and "extensive" research, lack- The main fallacy in this kind of
ing a better terminology. In short: in- thinking is that the reductionist hypoth- entiation takes place. This formulation,
tensive research goes for the fundamental esis does not by any means imply a called the theory of "broken sym-
laws, extensive research goes for the ex- "constructionist" one: The ability to metry," may be of help in making more
reduce everything to simple fundamen- generally clear the breakdown of the
The author is a member of the technical staff tal laws does not imply the ability to constructionist converse of reduction-
of the Bell Telephone Laboratories, Murray Hill, start from those laws and reconstruct ism. I will give an elementary and in-
New Jersey 07974, and visiting professor of
theoretical physics at Cavendish Laboratory, the universe. In fact, the more the ele- complete explanation of these ideas, and
Cambridge, England. This article is an expanded mentary particle physicists tell us about then go on to some more general spec-
version of a Regents' Lecture given in 1967 at
the University of California, La Jolla. the nature of the fundamental laws, the ulative comments about analogies at
Copyright© 1972 by the American Association for the Advancement of Science
1
other levels and about similar phe- leak through the triangle of hydrogens the symmetry laws have been, not re-
nomena. to the other side, turning the pyramid pealed, but broken.
Before beginning this I wish to sort inside out, and, in fact, it can do so If, on the other hand, we synthesize
out two possible sources of misunder- very rapidly. This is the so-called "in- our sugar molecules by a chemical re-
standing. First, when I speak of scale version," which occurs at a frequency action more or less in thermal equi-
change causing fundamental change I of about 3 x 1010 per second. A librium, we will find that there are not,
do not mean the rather well-understood truly stationary state can only be an on the average, more left- than right-
idea that phenomena at a new scale equal superposition of the unsymmetri- handed ones or vice versa. In the ab-
may obey actually different fundamen- cal pyramid and its inverse. That mix- sence of anything more complicated
tal laws—as, for example, general rela- ture does not have a dipole moment. than a collection of free molecules, the
tivity is required on the cosmological (I warn the reader again that I am symmetry laws are never broken, on the
scale and quantum mechanics on the greatly oversimplifying and refer him average. We needed living matter to
atomic. I think it will be accepted that to the textbooks for details.) produce an actual unsymmetry in the
all ordinary matter obeys simple elec- I will not go through the proof, but populations.
trodynamics and quantum theory, and the result is that the state of the system, In really large, but still inanimate,
that really covers most of what I shall if it is to be stationary, must always aggregates of atoms, quite a different
discuss. (As I said, we must all start have the same symmetry as the laws of kind of broken symmetry can occur,
with reductionism, which I fully ac- motion which govern it. A reason may again leading to a net dipole moment
cept.) A second source of confusion be put very simply: In quantum me- or to a net optical rotating power, or
may be the fact that the concept of chanics there is always a way, unless both. Many crystals have a net dipole
broken symmetry has been borrowed by symmetry forbids, to get from one state moment in each elementary unit cell
the elementary particle physicists, but to another. Thus, if we start from any (pyroelectricity), and in some this mo-
their use of the term is strictly an one unsymmetrical state, the system will ment can be reversed by an electric
analogy, whether a deep or a specious make transitions to others, so only by field (ferroelectricity). This asymmetry
one remaining to be understood. adding up all the possible unsymmet- is a spontaneous effect of the crystal's
Let me then start my discussion with rical states in a symmetrical way' can seeking its lowest energy state. Of
an example on the simplest possible we get a stationary state. The symmetry course, the state with the opposite mo-
level, a natural one for me because I involved in the case of ammonia is ment also exists and has, by symmetry,
worked with it when I was a graduate parity, the equivalence of left- and just the same energy, but the system is
student: the ammonia molecule. At that right-handed ways of looking at things. so large that no thermal or quantum
time everyone knew about ammonia (The elementary particle experimental- mechanical force can cause a conversion
and used it to calibrate his theory or ists' discovery of certain violations of of one to the other in a finite time com-
his apparatus, and I was no exception. parity is not relevant to this question; pared to, say, the age of the universe.
The chemists will tell you that ammonia those effects are too weak to affect There are at least three inferences to
"is" a triangular pyramid ordinary matter.) be drawn from this. One is that sym-
Having seen how the ammonia mol- metry is of great importance in physics.
NH By symmetry we mean the existence of
ecule satisfies our theorem that there is
no dipole moment, we may look into different viewpoints from which the sys-
other cases and, in particular, study tem appears the same. It is only slightly
progressively bigger systems to see overstating the case to say that physics
with the nitrogen negatively charged whether the state and the symmetry are is the study of symmetry. The first
and the hydrogens positively charged, always related. There are other similar demonstration of the power of this idea
so that it has an electric dipole mo- pyramidal molecules, made of heavier may have been by Newton, who may
ment (/i), negative toward the apex of atoms. Hydrogen phosphide, PH 3 , which have asked himself the question: What
the pyramid. Now this seemed very is twice as heavy as ammonia, inverts, if the matter here in my hand obeys
strange to me, because I was just being but at one-tenth the ammonia frequency. the same laws as that up in the sky—
taught that nothing has an electric di- Phosphorus trifluoride, PF 3 , in which that is, what if space and matter are
pole moment. The professor was really the much heavier fluorine is substituted homogeneous and isotropic?
proving that no nucleus has a dipole for hydrogen, is not observed to invert The second inference is that the in-
moment, because he was teaching nu- at a measurable rate, although theo- ternal structure of a piece of matter
clear physics, but as his arguments were retically one can be sure that a state need not be symmetrical even if the
based on the symmetry of space and prepared in one orientation would in- total state of it is. I would challenge you
time they should have been correct in vert in a reasonable time. to start from the fundamental laws of
general. We may then go on to more compli- quantum mechanics and predict the am-
I soon learned that, in fact, they were cated molecules, such as sugar, with monia inversion and its easily observ-
correct (or perhaps it would be more about 40 atoms. For these it no longer able properties without going through
accurate to say not incorrect) because makes any sense to expect the molecule the stage of using the unsymmetrical
he had been careful to say that no to invert itself. Every sugar molecule pyramidal structure, even though no
stationary state of a system (that is, made by a living organism is spiral in "state" ever has that structure. It is
one which does not change in time) the same sense, and they never invert, fascinating that it was not until a cou-
has an electric dipole moment. If am- either by quantum mechanical tunnel- ple of decades ago (2) that nuclear phys-
ing or even under thermal agitation at icists stopped thinking of the nucleus as
monia starts out from the above un-
normal temperatures. At this point we a featureless, symmetrical little ball and
symmetrical state, it will not stay in it
must forget about the possibility of in- realized that while it really never has a
very long. By means of quantum me-
version and ignore the parity symmetry: dipole moment, it can become football-
chanical tunneling, the nitrogen can
2
shaped or plate-shaped. This has ob- could be deduced semiempirically in I do not mean to give the impression
servable consequences in the reactions the mid-19th century without any that all is settled. For instance, I think
and excitation spectra that are studied complicated reasoning at all. But some- there are still fascinating questions of
in nuclear physics, even though it is times, as in the case of superconduc- principle about glasses and other amor-
much more difficult to demonstrate di- tivity, the new symmetry—now called phous phases, which may reveal even
rectly than the ammonia inversion. In broken symmetry because the original more complex types of behavior. Never-
my opinion, whether or not one calls symmetry is no longer evident—may be theless, the role of this type of broken
this intensive research, it is as funda- of an entirely unexpected kind and ex- symmetry in the properties of inert but
mental in nature as many things one tremely difficult to visualize. In the case macroscopic material bodies is now un-
might so label. But it needed no new of superconductivity, 30 years elapsed derstood, at least in principle. In this
knowledge of fundamental laws and between the time when physicists were case we can see how the whole becomes
would have been extremely difficult to in possession of every fundamental law not only more than but very different
derive synthetically from those laws; it necessary for explaining it and the time from the sum of its parts.
was simply an inspiration, based, to be when it was actually done. The next order of business logically
sure, on everyday intuition, which sud- The phenomenon of superconductiv- is to ask whether an even more com-
denly fitted everything together. ity is the most spectacular example of plete destruction of the fundamental
The basic reason why this result the broken symmetries which ordinary symmetries of space and time is possi-
would have been difficult to derive is macroscopic bodies undergo, but it is ble and whether new phenomena then
an important one for our further think- of course not the only one. Antiferro- arise, intrinsically different from the
ing. If the nucleus is sufficiently small magnets, ferroelectrics, liquid crystals, "simple" phase transition representing
there is no real way to define its shape and matter in many other states obey a condensation into a less symmetric
rigorously: Three or four or ten par- a certain rather general scheme of rules state.
ticles whirling about each other do not and ideas, which some many-body the- We have already excluded the appar-
define a rotating "plate" or "football." orists refer to under the general heading ently unsymmetric cases of liquids,
It is only as the nucleus is considered of broken symmetry. I shall not further gases, and glasses. (In any real sense
to be a many-body system—in what is discuss the history, but give a bibliog- they are more symmetric.) It seems to
often called the N -» » limit—that such raphy at the end of this article (3). me that the next stage is to consider the
behavior is rigorously definable. We say The essential idea is that in the so- system which is regular but contains
to ourselves: A macroscopic body of called N -> °o limit of large systems (on information. That is, it is regular in
that shape would have such-and-such a our own, macroscopic scale) it is not space in some sense so that it can be
spectrum of rotational and vibrational only convenient but essential to realize "read out," but it contains elements
excitations, completely different in na- that matter will undergo mathematically which can be varied from one "cell"
ture from those which would character- sharp, singular "phase transitions" to to the next. An obvious example is
ize a featureless system. When we see states in which the microscopic sym- DNA; in everyday life, a line of type
such a spectrum, even not so separated, metries, and even the microscopic equa- or a movie film have the same struc-
and somewhat imperfect, we recognize tions of motion, are in a sense violated. ture. This type of "information-bearing
that the nucleus is, after all, not macro- The symmetry leaves behind as its ex- crystallinity" seems to be essential to
scopic; it is merely approaching macro- pression only certain characteristic be- life. Whether the development of life
scopic behavior. Starting with the fun- haviors, for instance, long-wavelength requires any further breaking of sym-
damental laws and a computer, we vibrations, of which the familiar exam- metry is by no means clear.
would have to do two impossible things ple is sound waves; or the unusual mac-
—solve a problem with infinitely many Keeping on with the attempt to char-
roscopic conduction phenomena of the acterize types of broken symmetry
bodies, and then apply the result to a superconductor; or, in a very deep
finite system—before we synthesized which occur in living things, I find that
analogy, the very rigidity of crystal lat-
this behavior. at least one further phenomenon seems
tices, and thus of most solid matter.
to be identifiable and either universal or
There is, of course, no question of the
A third insight is that the state of a remarkably common, namely, ordering
system's really violating, as opposed to
really big system does not at all have (regularity or periodicity) in the time
breaking, the symmetry of space and
to have the symmetry of the laws which dimension. A number of theories of life
time, but because its parts find it ener-
govern it; in fact, it usually has less processes have appeared in which reg-
getically more favorable to maintain cer-
symmetry. The outstanding example of ular pulsing in time plays an important
tain fixed relationships with each other,
this is the crystal: Built from a substrate role: theories of development, of growth
the symmetry allows only the body as
of atoms and space according to laws and growth limitation, and of the mem-
a whole to respond to external forces.
which express the perfect homogeneity ory. Temporal regularity is very com-
of space, the crystal suddenly and un- This leads to a "rigidity," which is monly observed in living objects. It
predictably displays an entirely new and also an apt description of superconduc- plays at least two kinds of roles. First,
very beautiful symmetry. The general tivity and superfluidity in spite of their most methods of extracting energy from
rule, however, even in the case of the apparent "fluid" behavior. [In the for- the environment in order to set up a
crystal, is that the large system is less mer case, London noted this aspect continuing, quasi-stable process involve
symmetrical than the underlying struc- very early (4).] Actually, for a hypo- time-periodic machines, such as oscil-
ture would suggest: Symmetrical as it thetical gaseous but intelligent citizen of lators and generators, and the processes
is, a crystal is less symmetrical than Jupiter or of a hydrogen cloud some- of life work in the same way. Second,
perfect homogeneity. where in the galactic center, the proper- temporal regularity is a means of han-
Perhaps in the case of crystals this ties of ordinary crystals might well be dling information, similar to informa-
appears to be merely an exercise in a more baffling and intriguing puzzle tion-bearing spatial regularity. Human
confusion. The regularity of crystals than those of superfluid helium. spoken language is an example, and it
3
is noteworthy that all computing ma- possible; analysis, on the other hand, mined to try to reduce everything about
chines use temporal pulsing. A possible may be not only possible but fruitful in the human organism to "only" chem-
third role is suggested in some of the all kinds of ways: Without an under- istry, from the common cold and all
theories mentioned above: the use of standing of the broken symmetry in mental disease to the religious instinct.
phase relationships of temporal pulses superconductivity, for instance, Joseph- Surely there are more levels of orga-
to handle information and control the son would probably not have discovered nization between human ethology and
growth and development of cells and his effect. [Another name for the Joseph- DNA than there are between DNA and
organisms (5). son effect is "macroscopic quantum-in- quantum electrodynamics, and each
In some sense, structure—functional terference phenomena": interference ef- level can require a whole new concep-
structure in a teleological sense, as op- fects observed between macroscopic tual structure.
posed to mere crystalline shape—must wave functions of electrons in super- In closing, I offer two examples from
also be considered a stage, possibly in- conductors, or of helium atoms in su- economics of what I hope to have said.
termediate between crystallinity and in- perfluid liquid helium. These phenom- Marx said that quantitative differences
formation strings, in the hierarchy of ena have already enormously extended become qualitative ones, but a dialogue
broken symmetries. the accuracy of electromagnetic mea- in Paris in the 1920's sums it up even
To pile speculation on speculation, I surements, and can be expected to play more clearly:
would say that the next stage could be a great role in future computers, among FITZGERALD: The rich are different
hierarchy or specialization of function, other possibilities, so that in the long from us.
or both. At some point we have to stop run they may lead to some of the major HEMINGWAY: Yes, they have more
talking about decreasing symmetry and technological achievements of this dec- money.
start calling it increasing complication. ade (6).] For another example, biology
Thus, with increasing complication at has certainly taken on a whole new as- References
each stage, we go on up the hierarchy pect from the reduction of genetics to 1. V. F. Weisskopf, in Brookhaven Nat. Lab.
of the sciences. We expect to encounter biochemistry and biophysics, which will Pub}. 888T360 (1965). Also see Nuoro Ci-
mento Suppl. Set 1 4, 465 (1966); Phys. Today
fascinating and, I believe, very funda- have untold consequences. So it is not 20 (No. 5), 23 (1967).
mental questions at each stage in fitting true, as a recent article would have it 2. A. Bohr and B. R. Mottelson, Kgl. Dan.
Vidensk. Selsk. Mat. Fys. Medd. 27, 16 (1953).
together less complicated pieces into the (7), that we each should "cultivate our 3. Broken symmetry and phase transitions: L.
more complicated system and under- own valley, and not attempt to build D. Landau, Phys. Z. Sowjetunion 11, 26, 542
roads over the mountain ranges . . . (1937). Broken symmetry and collective motion,
standing the basically new types of be- aeneral: J. Goldstone, A. Salam, S. Weinberg,
havior which can result. between the sciences." Rather, we Phys. Rev. Ill, 965 (1962); P. W. Anderson,
Concepts in Solids (Benjamin, New York,
should recognize that such roads, while 1963), pp. 175-182; B. D. Josephson, thesis,
There may well be no useful parallel often the quickest shortcut to another Trinity College, Cambridge University (1962).
to be drawn between the way in which part of our own science, are not visible Special cases: antiferromagnetism, P. W.
Anderson, Phys. Rev. 86, 694 (1952); super-
complexity appears in the simplest cases from the viewpoint of one science alone. conductivity, , ibid. 110, 827 (1958);
of many-body theory and chemistry and The arrogance of the particle physi- ibid. 112, 1900 (1958); Y. Nambu, ibid. 117,
648 (1960).
the way it appears in the truly complex cist and his intensive research may be 4. F. London, Superfiuids (Wiley, New York,
cultural and biological ones, except per- behind us (the discoverer of the positron 1950), vol. 1.
haps to say that, in general, the rela- 5. M. H. Cohen, /. Theor. Biol. 31, 101 (1971).
said "the rest is chemistry"), but we 6. J. Clarke, Amer. J. Phys. 38, 1075 (1969); P.
tionship between the system and its have yet to recover from that of some W. Anderson, Phys. Today 23 (No. 11), 23
parts is intellectually a one-way street. (1970).
molecular biologists, who seem deter- 7. A. B. Pippard, Reconciling Physics with Reali-
Synthesis is expected to be all but im- ty (Cambridge Univ. Press, London, 1972).
4
Pressure Broadening in the Microwave and Infrared Regions
Phys. Rev. 76, 647-661 (1949)
This is a condensation of most of my thesis. For two years of postwar grad school,
1945^17,1 took the (mostly wonderful) courses available at Harvard from Schwinger,
van Vleck, Furry, and visitors such as Goudsmit and Gorter, and, aside from that,
enjoyed myself, eventually winding up as one of a group which congealed around Tom
Lehrer, Dave Robinson, and others like Chan Davis and Bob Welker, which was more
about bridge, singing, puzzles and scifi than physics. I had the GI Bill and a tiny
scholarship, and would have been happy to go on indefinitely as Tom Lehrer actually
did. But in the summer of 1947 I met the right girl and things suddenly became more
serious. Van had given me, as one of his last three graduate students (another was Tom
Kuhn), a marvelous problem, in spite of my marginal performance on my qualifying
exam. It arose from one of the postwar revolutions in spectroscopy which radar-
derived coherent sources allowed, unprecedentedly detailed molecular spectroscopy. I
solved it using a series of novel techniques borrowed from Schwinger's courses
among other sources. I felt that after this there was little more to be done in this field
and wanted to move on to other things, an attitude which one job interviewer
interpreted as reprehensible, and chose not to hire me.
5
Reprinted from T H E PHYSICAL REVIEW, Vol. 76, No. 5, 647-661, September 1, 1949
Printed in U. S. A.
Pressure Broadening in the Microwave and Infra-Red Regions*

P . W . ANDERSONf
Harvard University, Cambridge, Massackusettst
(Received May 12, 1949)
In this paper a generalized theory of collision broadening is developed, adequate for predicting line
breadths in the microwave and infra-red regions. This theory differs from previous ones in taking into
account transitions among quantum states caused by collisions, although it is limited to a classical picture
of the relative motion of the colliding molecules as a whole. Formulas are derived for computing approxi-
mate line broadening collision cross-sections from known intermolecular interactions, and the results of the
theory are successfully compared with experiments on self-broadening in the ammonia inversion spectrum
and the vibrational band spectra of HCl and HCN. Some cases of foreign gas broadening in the microwave
region are examined, but it is concluded that in general the Van der Waals interaction, commonly assumed to
be the force important in causing foreign gas broadening, is not adequate to cause the observed broadenings.
The more complicated types of forces which become important at short range would have to be taken into
account to give good agreement with experiment in these cases.
PART I. THEORY computed by means of the adiabatic approximation and

A. Introduction is equivalent to a relative phase-shift of the radiation
before and after the collision, with no amplitude change.
I HE theories of pressure broadening presented up
-*- to this time can easily be shown to be inapplicable B. Generalization of the Fourier Integral
in the microwave and infra-red regions of the spectrum.
The primary purpose of this paper is, therefore, to It is clear that it is necessary, in order to compute pres-
develop a theory which can give accurate results in sure broadening in any frequency range, to develop an
these regions, and to compare these results with the expression for the spectral shape of a pressure-broadened
available experimental data. line which encompasses the two limiting cases of high
and low frequencies. In other words, a formula must be
The older theories, in particular the Fourier integral
obtained which includes the effects both of phase-shifts
treatment developed primarily by Lorentz, Lenz, and
and of transitions. Before presenting this formula, we
Weisskopf,1-4 have given good results in the optical
should emphasize the physical picture upon which it is
region.4'5 However, this theory is essentially in contra-
based. This picture is equivalent to that employed by
diction to the theory of dielectric relaxation developed
Foley9 in his re-derivation of the standard Fourier
by Debye and others, 6-8 which treats broadening in the
integral formula by means of quantum radiation
limiting case of a spectral line of zero frequency.8 Both
theory. The basic assumption we call the "assumption
theories involve computing the spectrum due to a
of a classical path": we assume that for all collisions of
radiating molecule randomly interrupted by collisions
interest the colliding molecules can be taken to be
with other molecules. However, the mechanism by
point dipoles following definite paths; thus their inter-
which the interruption is effected is essentially different
action is a definite function of the time Hi{t). The jus-
in the two theories. Debye's relaxation theory considers
tification for this lies in the fact that one can think of the
interruptions due only to reorientations of the radiating
motion of the molecules in terms of packets of transla-
molecule; in a quantum theory, this is equivalent to
tional wave functions; by the uncertainty principle the
considering transitions among the various spacially
allowable spreads Ap and &q in position and momentum
degenerate quantum states. Contrariwise, the Fourier
space are ApAq=h. We seldom or never consider in
integral theory explicitly assumes that no transitions
detail collisions in which the molecules under con-
are caused by collisions. Then the effect of a collision is
sideration pass at distances less than the kinetic theory
diameter, = 5A. Then an uncertainty in position of 1A
* The material in this paper is a part of the author's thesis sub- will lead to no great ambiguity in magnitude or type
mitted in partial fulfillment of the requirements for the degree of
Doctor of Philosophy at Harvard University. of intermolecular forces. The corresponding uncertainty
f National Research Council Predoctoral Fellow. in velocity is
% Now at Bell Telephone Laboratories, Murray Hill, New Jersey.
Av/v= 0.3/(molecular weight)1
1
H. A. Lorentz, Proc. Amst. Acad. Sci. 8, 59 (1906). (1)
2
W. Lenz, Zeits. f. Physik. 80, 423 (1933).
3
V. F. Weisskopf, Phys. Zeits. 34, 1 (1933). which is a small percentage for most molecules. Calcu-
1
E. Lindholm, Ark. Mat. Astron. Och Fys. 32, 17 (1945).
5
H . Margenau and W. W. Watson, Rev. Mod. Phys. 8, 22 lations carried out by Lindholm10 for many optical
(1936).
6
problems, not using this assumption, have indicated
P. Debye, Polar Molecules (Chemical Catalog Company, New that the assumption is normally valid, in consideration
York, 1929), Chapter V.
7
W . Kauzmann, Rev. Mod. Phys. 14, 12 (1942).
8 »H. M. Foley, Phys. Rev. 69, 616 (1946).
J. H. Van Vleck and V. F. Weisskopf, Rev. Mod. Phys. 17, 227 10
(1945). E. Lindholm, Dissertation, Uppsala, 1942.
647
648 P. W. ANDERSON
of the large quantum numbers of the partial waves tions come out of the theory actually does appear when
whose contributions were important. the classical path method, or equivalent absorption
The problem is reduced, by this assumption, to methods, are used; 8 we have been able to clarify this
finding the spectrum radiated by a molecule, whose point to some extent, but our work on this point will
unperturbed Hamiltonian we designate by H0. This not be reported in this paper.
molecule undergoes random perturbations due to col- It is interesting to rewrite (2) for the case in which
lisions, representable by a time-dependent interaction the standard Fourier integral of Foley and his prede-
Hamiltonian Hi(t). The problem can be solved following cessors is valid. This is the case in which the perturbing
Foley's method, by the use of general radiation theory; Hamiltonian Hi(t) is either diagonal already, or can be
or it is possible to extend the correspondence-principle made so in a simple manner (following Foley) by re-
radiation theory of Klein and Pauli to our problem by placing transitions to high-energy levels by their
analogy. Either method leads directly to the following second-order effects; that is, we replace the so-called
formula, which we present without proof:** Van der Waals forces, a second-order phenomenon, by
a first-order Hamiltonian. This, too, can be shown to
be valid from the general equations of motion and the
/(«) = const. Xoi4Tr\ dte^'fitit) radiation theory. Then Eq. (3) integrates explicitly; a
('•/: typical matrix element is
X
f dt'e "WO • (2) „W = M»„(0) exp \/ih\ {Em-En)t
Here w is the angular frequency for which we desire

the spectral intensity I(w); p 0 is an assumed initial - f Z(m\H1\m)-(n\H1\n)^\\
density matrix for molecules in the gas, in fact essen-
tially the density matrix for the time / = — co ; /j,z(t) is (5)
the Heisenberg time-dependent operator (see reference
12) for the z-component of the dipole moment (if we The unperturbed frequency is u„m, given by
consider radiation polarized in the z-direction); and
the "Tr" means that one must take the diagonal sum tiUnm = Em — En.
of the indicated matrix product. nz{t), of course, satis-
Em is the mth eigenvalue of the unperturbed Hamil-
fies the Heisenberg time equation
tonian H0. Equation (5) defines the relative phase-shift
ihii=[Hii — nH2, (3) r]nm for the two levels n and m. We can then use the
where Boltzmann density matrix
H= H0+H1(t). (4) (»| pB\m) = Snm exp(-En/kT)

Finally, we have omitted all numerical factors other X[Znexp(-En/kT)r\ (6)
than the frequency-dependent one co4. These factors can
easily be shown to come correctly from the general for po, and I in (2) separates into the following con-
theory. tributions Imn for each of the various spectral lines m
The formula (2) gives the intensity for spontaneous and n:
emission. The approach through spontaneous emission
is chosen in order to avoid the use of a time-dependent 7m„(co) = c o 4 p x p ( - £ „ / * r )
density matrix.tf The intensity for absorption must be
l co 12
derived from (2) by use of the Einstein relations."
Some confusion as to whether the correct Einstein rela- /
(ftexpj—«[(&> — C d m „ ) H - > ! . n n ( 0 0)
** The intensity in classical theory is given by the average over- This is the form of the Fourier integral used by Weiss-
all states of
kopf, etc.JJ
w4 I n,(t)e*"dt
It It is interesting to note that the derivation of (7) employed
The corresponding quantum-mechanical quantity is obtained by by Foley fails to lead to the correct external frequency factor oi'.
replacing all observables (in this case ,i (t)) by quantum-mechanical If carefully followed, Foley's theory gives a factor here of w2conm2,
operators, in multiplying by the density matrix, and taking the which is by no means correct at low or zero frequencies, where the
trace. Thus Eq. (21) is reasonable; a rigorous proof can be given. line-breadths are comparable to umn. Even if Foley's "adiabatic
For this procedure see W. Pauli, Handbuch der Physik, 2. Aufl., assumption" (Hi diagonal) were correct at low frequencies, these
Band 24, 1. Teil, p. 149. factors would be in error. The reason is that the general theory
ft See, for instance, R. Karplus and J. Schwinger, Phys. Rev. must be retained, at least until the point where a partial integration
73, 1(120 (1948). can be performed analogously to that which one would use in the
" A . Einstein, Verh. d. D. Phys. Ges. 18, 318 (1916); Phys. classical theory to go from the Fourier integral of t i e vector
Zeits. 18, 121 (1917). potential A (~/i) to that of «(() itself.
7
PRESSURE BROADENING 649
C. Introduction of the Time-Development This operator is the solution of (9') with initial con-
Matrix T dition T(t) = 1; it answers the question "what happens
in the time-interval t-*l'f". In addition, we use the fact
I t is well known that it is possible to write the time-
that the density matrix, as a function of time, trans-
dependent operators p(l) in Eq. (2) in the following
forms according to T(i) but inversely to other matrices.
manner :12
Then (11) becomes:
U.(t)=U-Kt)»oU(t), (8)
where
ihid/dÛîHe+HÛ. (9) 7(o)) = const. Xw 2 J dt I die
If MO is the value of n at a time U, then U(t0) = 1 is the
initial condition on Eq. (9). The time-development XY.abcde(a\p{t)Ib)(b\no\c) exp(—KOÔ
operator U is not as useful to us as another operator T, X(c\T-\t-+t')\d)(d\IM>\e)
which is defined in much the same way. However, in
deriving T one uses a Heisenberg representation in rela- Xexp(-*fcW')(e| T(t-*t') | a). (13)
tion to the unperturbed energy H0. Therefore the equa-
tion for M(0 is In order to use the considerable simplification made
available by the Wiener-Khintchine relations in evalu-
MW = T - 1 W exp(-ff 0 */tft) w exp{H0t/ik)T(i), (8') ating our formulas, 13 it is expedient to re-express them
in correlation-function form. By means of the simple
and that for T is
substitution
ihf(t) = H/(t)T. (9') t' = t+T,
jf?i' is the interaction Hamiltonian with the time- we obtain
dependence due to ffo inserted:
(m| H^ \ n) = (m] H^t) | n) exp[(£„-E n )t/ih~]. (10) I(u) = w4(const.)Y.de J dr exp\j(u—ude)T]<pde(T), (14)
The diagonal elements of T give the phase-shifts due to

the interaction Hi, while the off-diagonal elements—or where
their squared absolute values—give the transition
probabilities due to this interaction. In terms of eigen- <Pde(r)= I dt Y,abc e x p [ — i(0)bc+0>de)t]
states of H0, (8') is
HM(0|»)=£(w|r-M£) X(a|p(0|i)(»|Mo|c)(c|r-'(«-»H-r)|<i)
Xexp(-io>klt)(k\no\l)(l\T\n). (8") X(d\^\e)(e\T(t^t+T)\a). (15)
We may now insert this explicit form for p.(t) into the We observe that (15) has the form of an average over
Fourier integral formula (2), and obtain the following the time (. I t consists of three essentially different terms:
complicated expression: p(0, which depends on the state of the molecule a t /;
the T(t—H-\-T) factors, which depend on the collisions
1(a) = COnSt. X « 4
Yâbcd^fM I Po| *)
which occur between t and / + T ; and the phase-factor.
At this point we first limit ourselves to the "impact"
type of theory: we assume that the collisions are short
x (^"'(Jir-HOIcXclMoM) compared to the time between them. Under this
assumption, it is easy to show that unless wbc+Vde is less
Xexp(io,cdt)(d\T{t)\e) than 1/T, the inverse of the time between collisions,
(15) will essentially vanish. For this reason, and for
X J* *'e--''(e|r-KOI/)(/lMo|g) other reasons which we shall not consider here, it is
considered legitimate to drop from (15) all terms for
~" Xexp(-i«fot')(g\T(t')\a). (ID which ubc+oide^Q- This is equivalent to assuming that
all lines considered are either not resolved at all com-
D . Re-expression in Correlation-Function Form, pared to the line-breadth—in which case we treat them
and Simplification Using the Assumption of as degenerate—or are well separated.***
Short Collisions I t is also legitimate, since the phase and T-factors
Formula (10) can be simplified greatly by the insertion are independent of the actual state of the molecule, to
of a new time-development matrix take the average of pit) over the history of the molecule
T{t-^t') = T{t')T-l{t). 13
(12)
See, for instance, M. C. Wang and G. E. Uhlenbeck, Rev.
Mod. Phys. 17, 323 (1945).
ls
Bom u. Jordan, Elementary Quantenmechanik (Verlag J. *** In Section H we will discuss further the terms which are here
Springer, Berlin, 1930). dropped.
8
650 P. W. ANDERSON
separately. Clearly, we obtain for pAv the Boltzmann We divide the types of collisions into classes according
distribution (6). Let us now introduce a degenerate to the paths followed by the molecules; then, corre-
index m. Each state will be designated by a non- sponding to each "element of cross section" dak we can
degenerate index (a, b, c, d- • •) and a degenerate index assume that there is a definite phase-shift JJ4. Then the
(m, m'• • •). Then, letting the "initial" state be desig- probability that a collision in d<xk occurs in a given, short
nated b y i, the final by / , (14) and (15) are time dt is given by
p(diri,) — nvdldok, (21)

I = Hi,t I dr exp[i(wu-u)T'](i\ PB\i)<pif(T), (16)
where n is the number of molecules per cc, and v is their
velocity (assuming, as is always justifiable for col-
where lision-broadening problems, that all molecules have the
average velocity v).
<wM = Zm,».'.»•",».'" I dt(i, m\/i\f, m') Consider the integrand of the expression (18) for
<p(r):
X(f,m'\T-\t-^t+T)\f,m") f(t, r) = exp {i[r,it(t+ r) - r,S)1 \ • (22)
If in the time dr after r a collision of type k occurs, we
X(/, m"Mi, m'")(i, m"'\T(t^t+r)\i, m), (17)
have
or f(t,T+dT) = exp(ii)t)f(l,T), ,„,..
/(/, r+dr)-f(t, T) = [exp(i„ t )-l]/(/, r). (Zd)
*v(r) = 7>r f dty.V{T~\t~^t+T)Y Now let us take the integral, which is equivalent to
averaging over all times t:
XM"(JX*-»H-T)«J, (17') <p(T+dr) — <fi(T) — d<p(r)

= ([exp(i7)i) — X]f{t, T)Average over t- (24)
where the superscripts designate "Teilmatrices" for the This is the point a t which the impact theory assumption
given non-degenerate indices. is vital. We must be able to say that the time dT is short
compared to the interval between collisions, and so
E. Solution for the Correlation Function in the contains a t most one collision; on the other hand, dr
Standard Optical Case must be quite long compared with the duration of a
collision, so that there is no correlation between what
To illustrate our method of finding the correlation happens in the interval dr and what happens in the
function and to furnish some introduction to the line- preceding or following times. Then we can take the
shapes to be expected, we shall re-derive the impact averages in (24) separately, and have
theory solution of the standard Fourier integral (7).
This formula, re-written in correlation-function form b y d<p(r) = <p(T){exp(ilJlc) — leverage over collisions
means of (13), and exclusive of numerical factors, is
= 1 Tjodr I <fo-t(exp(ttjfc) —1) I P M , (25)
7(w)= i </Texpp(&) V /-ai)T]^(T),
(18) so that
<p(r) = e-<v>", (26)
<P(T)= J dtexp[i£i)if(t+T)-r)i/(t)l}.
where
a= i da(\ — e*') = <rr+ JV,.

The assumption of the impact theory is that the col- - /
lisions are short compared to the time between them.
Then I t is easy to see that
<P(-T) = <P*{T)
Vif(t+T)-Vif(t)= £ Ve, (19)
collisions from as is required for the reality of I(w); then we get for the
/ tol+r
intensity distribution
where rjc are the phase-shifts due to the various col-
lisions occuring in the interval t—*t+r: 2{nw,)
7(o)) = const.X- (27)
(co—on/—»tw,-)2+ (nv<Tr)2
v.= j dC(»l-ffi'»l*)-(/|i/i'W|/)]. (20)
This is a dispersion line-form with half-width
2ir(Av)i=nv<7r, shift 2xAy</=wWi. (28)
9
F. Solution for the Correlation Function in the The two terms in <r (34) correspond to the 7 and S of
Classical Debye Case the Van Vleck-Weisskopf paper.
Since the classical Debye problem of the spectrum G. The General Quantum-Mechanical Case
of a rotating dipole perturbed by collisions has fre-
quently been considered by absorption methods, 6 - 8 in I t is possible to do the quantum-mechanical non-
particular the very general method in the appendix of resonant, or Debye, case by a generalization of the
the paper by Van Vleck and Weisskopf, it is of interest above arguments similar to that employed in the paper
to show how easily this problem is solved by correlation by Van Vleck and Weisskopf. However, the arguments
function methods. Here we can see that the spectrum is for this and for the general quantum-mechanical case
involving spacial degeneracy are quite equivalent. We
shall give only the latter, regarding the Debye case as an
1(a) = I dTeioT<p(T),
obvious special application.
(29) We want to find the breadth of the line corresponding
to a dipole transition between two states i and / , of
V{T)= f dt^(t+r),xz{t), angular momenta _/',• and jf, with jf=ji±l or /.-.ftt
Equation (17) for the correlation function applies here,
where y.z(t) is the z-component of the rotator's dipole with the spectrum given by (16). Before proceeding
moment at a time t. Again we consider the function farther, we must prove a theorem about the expression
(17), or rather about the following part of it:
/(/,T)=M,(H-TW0, (30) Z~',*<(*n\Ti-1(.tr->t+T)\m'Xm'\pL,v\m')
2
/ ( / , r+dr)-f{t, T-) = M (COS,KH-T-NT) average over /
— C O S ^ ( / + T ) ) COSI^(/), = (m\ (Ti~1(t-*l+T)lX,«Tt(.M+T))!iversisc over I K " )
(36)
where ^,(<) = /icos^(/), ^ being the angle of the dipole
relative to the z-axis. Suppose we apply a rotation of the coordinate axes to
Our basic assumptions will be (a) that the prob- (36) in the sense that we do all of our computations
ability of a collision taking the rotator from the ele- with respect to a new set of axes x', y', z' = R{x, y, z).
ment of solid angle dot into doi' in a time dr is We must use the fact that, since in the course of the
time average we consider collisions from all conceivable
nvd(T„-+a>dT, (31) directions and of all possible types in equal number,
the affect of the (Ti~1*Tf) factor is essentially an iso-
and (b) that this probability is obviously independent tropic one. (This is equivalent to the step (32)—+(33).)
of the azimuthal angle <p between da and da', and there- Mathematically,
fore is a function only of the polar angle 6 between
these two directions. We know that R(T~ V , 7 % = ((RT-1) (.KM.) (RT))„
cosl/(da') — cos\[/(da) cos6-\-sin\p(da) sinS cos<p. (32) = (T-i(Rix,)T)to
Since, upon averaging the difference of cosines in (30) = \i'x(T~1iiXT)ta+ x 2 ' „ ( r - V«^)A.
over all possible collisions, the last term in (32) will
vanish because of the factor cosp, we can use +x^ 2 (r-v,r) s „ (37)
cost(t+ r+dr) - cosMH- r) = cosiKi-r- T) ( C O S 0 - 1). (33) where the X's are the direction cosines of the new axis
Then we actually perform the average of (30) over all system with respect to the old. Thus the quantity (36)
times, using again our impact theory assumption of has the transformation property of a vector com-
short collisions. I t is clear that the result is: ponent. But the i, f matrix elements of any vector
component are determined, exclusive of a constant
factor, by the transformation property. 14 Thus we can
d<p(r) = \nvdT I i<r„->„<(cos0— 1) M r ) , say
(m| (Ti-l(t-^t+TWT'(t->t+T))Av overl\m'")
p(T) = e-*"", (34)
= F{r){m\^'\m'"), (38)
where F(j) is a scalar.
<r= I do-(l — cosfl) (real).
We are now in a position to set up the differential
This gives, with the correct external factor a, the Debye ftf The case of higher multipole transitions is easily worked out
line form: and, interestingly, leads usually to a different breadth.
11
E. U. Condon and G. H. Shortley, Theory of Atomic Spectra
I(a) = nvaa/[_a2-\-(nvo-)r]. (35) (Cambridge University Press, Cambridge, 1935), Chapter III.
0
652 P. W. A N D E R S O N
equation for <p(r), or, equivalently, for F(j), since tically identical, except, of course, that the radiation
occurs according to the operator
<p(T) = T.m„n.:\{m\^'\m'")\*F(T),
M2 (l)+M*(2). (44)
dVi.r) = T.m,^-\(m\^\m'")\2dF{T),
The two molecules of interest are called (1) and (2).
( r" I t is clear that it is legitimate to use only the first
half of (44), regarding, therefore, molecule (1) as
the radiator, (2) as a perturber (the line-shape for
= Tr\ J dW'lTît-^t+T+dT)^'
molecule (2) will be included in the implicit average
over all molecules in the gas). Then, in place of the
X T'(t-*t+ T+dr) - Ti'1(T^t+ T)fiV transformation property "vector component," we use
essentially "vector"
molecule (1) " s c a l a f ^ r n o l e c u l e (2) • Without
X P N + T ) ] . (39)
going through the reasoning for the case of two mole-
We introduce a matrix Td=T(t+ T^>l+T+dr) (the cules, which differs from the preceding only in that the
notation Td indicates that this matrix is not a dif- Boltzmann factors for molecule (2) must be included
ferential) ; then and that T-matrices involving any transition whatever
T(t-+l+T+dT) = T{t-^t+T)Ti. (40) for molecule 2 are of interest, we present the correct
Again we can make use of the reasoning of Section E, result for a:
(24)—->(25): we use the assumption of short collisions to a= I da{[_ £ (mia2\Ti-1(da)\mib2)(ps)b2
enable us to take independent averages of the happen- •/ mi, mi'
mi", mi'"
ings in t—*t-\-r and H-i—>t-\-r-\-dr. Then we introduce Mb,
Eq. (40), Eq. (38) for F{T), and get for (39):
X « I M / 7 I » i " ) (wi"621 Tf{da) I mi'"at)
= F ( T ) r r [ < M / « r „ W M . t f Z Y - M . < 0 ] ) A v . over all dr\. (41) X(«i"W|«i)]

Now we introduce the probability assumption (21) for x [ E IMM/'ki')!2]-1-!}- (45)
types of collisions: mimi'
p(da in dr) = nvdadr. 02, bi designate the various states of the second mole-
cule. Two possible cases occur in which this expression
If a collision of type da occurs in dr, the T matrix T4 can be simplified to a usable form. First we have the case
will be characteristic of the collision da. In fact, T is in which no transitions occur in molecule (2) quantum
obtained by integrating Eq. (9') from t — — » to + » , numbers, except perhaps for degenerate ones; second,
using for ffi'(0 the time-dependent interaction Hamil- that case in which transitions in molecule (2) quantum
tonian for just one collision of type da occurring in the numbers can occur only simultaneously with those of
whole range of t. The initial condition is T(— co) = l. molecule (1), so that T' and Tf axe. still essentially
This integration is legitimate because of our assumption diagonal in molecule (2) because they are so in molecule
that dr can be chosen long compared to the duration (1) (this is the interesting case corresponding to resonant
of one collision. We call the resulting T matrix T{da); interactions). In either case, a reduces to the following
then the average in (41) gives the following equation simple form:
for?.
cr = 2jcro2,
[dF(T)/dT~\=-nwF{T), F{T) = <T™\ (42)
aa%= (PB)«2
/•[TriT'idayW'T'idaW') 1
a= da\ 1 . (43) r [Tr(Ti'"-iday-»uizifT'''"(da)n/i) 1
J L Trfa.W) J X da\ 1 . (46)
It is obvious that the shift in frequency and the line-
breadth follow from the real and imaginary parts of The trace includes a summation over mi, the molecule
the number a, precisely according to (28). Thus the (2) degenerate index, and 2/02+1 is the degeneracy of
problem of the line-shape is solved in principle. the a,2 level.
Before proceeding further with the evaluation of a,
the "collision cross section for line-broadening," it is H. Reduction of a to Computable Form
necessary to generalize formula (43) to a case which we
1. Use of Rotational Invariance to Relate Matrix Elements
shall frequently treat—that in which the interaction
to "Collision Axes"
Hamiltonian Hi'(t) is a matrix involving the quantum-
mechanical states of both the perturbing molecule and In the form (43) it is still practically impossible to
the radiating molecule. The line of reasoning is prac- compute a, since we must find the ^-matrices not only
11
for every impact parameter, or distance of closest tions with the assumption that J'Hi'dt/h is small. We
approach of the two molecules during collision, but for use the following procedure:
every set of direction angles of the colliding molecules
relative to the polarization direction. I t is possible to T=T0+T1+T2+---, 7-0=1,
use the fact that the fdo implies an average over all
conceivable sets of direction angles to remove this r' #i'(0
latter difficulty. We do not do this rigorously, since it ihTi = HiT„; 7\(0= df, (49)
involves some rather long group theory. The following J^„ ih
argument, however, indicates the method.
The integrand of (43) is, as it stands, rotation invari- ' Hi(t') r"Hi'(t")
ant, since it is a trace. Essentially, this means that the ihf^H/Tu T2(t) = df - ', etc.
component of M we deal with is immaterial. Suppose we •J-^ ih J ih
can compute the /"-matrices for one specific set of direc-
tion angles (for instance, suppose we know them for a If we could assume that 777(0 commuted with Hi'(t")
system quantized along b, the impact parameter). at all times, the result of this process would be simply
Then it is possible to leave these /--matrices as they the scalar solution of the linear differential Eq. (9'):
stand and to take the direction average by rotating the
z-axis, i.e., the component of n which is used in (43).
Since (43) is quadratic in \il, it is allowable merely to T(t) (50)
average over the x, y, and z-directions. Then we get
simply
On the other hand, it is clear from the definition (10)
of Hi that it does not commute with itself at all times,
•z>G'.W)=iE*».f 2Trbdb but only because 77i and H0 do not commute. Hi(t) is a
function of the coordinates only, in general, and will
always commute with itself.
An investigation was made into the size and meaning
of the "non^commuting" terms by which (50) differs
fix, y, z fix, y, z '«], (47) from the accurate solution of (49). Those terms in-
volving high frequency elements of Hi ("high" com-
where T'(b) designates the /"-matrix in the "collision- pared to the rate of change of 77i) were found to be
axis" coordinate system. precisely equivalent to the terms obtained in (50) by
Since a great deal of further manipulation with (47) including, in Hi itself, the 2nd-, 3rd-, etc., order forces
is necessary, we shall use an alternative form for it, due to the high frequency elements of 77/. On the
based on the identity between the vector matrix and other hand, such terms are small if they involve low
the Wignerian coefficients.12 frequency elements of 77i'.
Therefore it seems permissible to ignore the non-
(TO I H.V | n) = (jfljun | j'/l/iO) (48) commuting terms and to use (50) for T, including at
least the second-order (Van der Waals') forces in 77i.
and similarly for l/VZO^zbiju,,), with 0 replaced by ± 1 . These terms have not been computed accurately ac-
Then it is easy to show that (47) is the same as cording to (49), because they appear first in the suc-
cessive terms T2 (shift) and 7% (broadening). The Van
(j/ljim\jflfiM) der Waals' procedure introduces the second-order forces
r= f 2-wbdb £ (jAjtm'\j/ln'M) in the same manner as the first-order ones in 7 \ and T2
»^n M,m,m'
M, m, i (2ji+l) where they are much easier to handle. I t is an inter-
esting check on our method to note that the second
XlSmm^^-(m\T^(b)\m')(ix'\Tf(b)\ixU. (47') order effects of high levels would derive naturally in the
correct form from the general theory.
A similar formula can be written in the two molecule For simplicity we limit ourselves to the first three
case of Eq. (46). terms of (50). These are
2. Introduction of an Approximate Form for T. T(b)^T0+Ti+T2,

(51)
T0=l; Tx=-iP; T2=P2/2,
For the purpose of computing a from (47), an ex-
pression for the T-matrices in terms of the interatomic where
forces must still be found. This cannot be done exactly. l r"
However, a device which has been frequently used is P= Hi '{i)dt. (51')
the following expansion of T in successive approxima-
12
654 W. ANDERSON
*
/ \
\ •
APPROXIMATION NO. 1
APPROXIMATION NO.2
paper.16 Also using the fact that Tr1= — Ti, we get
' \ APPROXIMATION NO.3
.
/
1
\
\
PROBABLE SHAPE OF
CURVE: a AND b r h (m\I
(mlP^m)
1 ' \| (2 POSSIBILITIES)
5x-i E —r (54b)
! ! \ L"—it 2ji "—'r 2jf+
\
1 1 ,--- S -~A Referring to the definition (51) of P, it appears that
3< this, the first "shift" term in a, is just the first term in
j J b" \
1 | \2 the simple phase-shift approach, averaged over all
directions.
There are two possibilities for obtaining second-order
1 \ |s a (u)-i ^ ^ ^ ^
terms in the S sum. We have the choice of taking
To(XT/ or vice versa, or of taking the terms involving
FIG. 1. 7VX7Y. The former pair of terms follow precisely as
did the Si terms; we call these (S2) outer.
3. Use of the Approximation for T in Computing a and
r Hzv<-»|») (M|ZY|M)
Line-Broadening
Let us repeat the formula (47) here, in a slightly
(St)o
r
L i
2- ;
2ji+l
hZ
2//+1 J
1
(54c')
different form:
a= f 2irbdbS{b),
4 (i,m\P*\i,m)
2ji+l
(f,»\I»\f,*y
2jf+l J'
The second type of term, however, must be left in its
(52)
(JAj m I JAliM) (jAjiM\ JAP'M) original form; we call the sum of these (.SVJmiddie:
s(b)= z 1
m, m', M 2jd-l
(S*) middle — E (jAjim\JAvM)
2ji~\~l M, m, m'
X («„•*-»'- H T'-Kb) |m'W| T'{b) |„).
X (jAjtm' | JAv-'M) iim \ P | im') (f/ \P\fr). (54c")
We are going to expand S(b) in powers of the P-matrix
of Eq. (51'): Further terms are prohibitively difficult; in any case,
the "non-commuting" part of T can no longer be
S=S0+S1+S2+---. (53) neglected in Ts.ttt
In order to get terms of the zero'th order in P, we must 4. The Interpolation Process
pick Ti=Toi, T'=Tj, and we get
Since the expansion (53) of S(b) is valid only for
So=0. (54a) small P, and P increases very strongly with decreasing b
(/-3 for first-order, r - 6 for second-order forces), ob-
The first-order terms are obtained by including either
viously the first two non-vanishing terms of 5, derived
T*= To, Tf= Tx or vice versa. The result is
above, are of no help for collisions for which b is small.
(JAjwljAiiM) For this range of b we must use physical reasoning,
S1=- £ . (JAJ,-m'\JA^M) based on the fact that "strong" collisions such as these
M, m, m' 2yt~f- 1 must have an effect equivalent to complete interruption
of the radiation. The T-matrix for complete interruption
x[(«|r1i-Mw)^+(M'|r1/|M)5„m-]- 15
G. Racah, Phys. Rev. 62, 438 (1942).
t B I t is clear that in (54c) the degenerate and non-degenerate
This is greatly simplified by the fact that the Wignerian levels play essentially different parts. One can interpret these
coefficient (jAji>n\JAl*M) vanishes unless m=n+M; terms by saying that transitions to non-degenerate levels act like
complete interruptions of the radiation, while transitions among
thus if M=M', m=m', or vice versa, and Si is simply degenerate levels take on the character of phase-shifts to a greater
or lesser extent, whence the terms (54c"). This phenomenon, and
5i=-E E |07ii,H;7W)l2 the question of when a level is or is not "degenerate," are allied
to the general problem of coherence of radiation of different fre-
M m or n
quencies which crops up in many fields.
x[(«|r1^|m)+(M|r1/|M)]. In this case, the problem is fairly simple. It is easy to see that
the different behavior of degenerate and non-degenerate levels
follows from the step of dropping the terms of finite frequency in
The sums over M are easily performed, one by the Section D, Eqs. (13) and (14). We noted there that this step was
unitarity of the Wignerian coefficients, and one by this allowable if the frequency difference of the levels was much
and the "permutation of indices" property of these greater than 1/T. In addition, (54c") can be shown to vanish for
first-order dipole-dipole interactions in any case, so that ob-
coefficients which appears most easily from Racah's servable effects of this cause seldom appear.
13
can have two forms: For the pure phase-shift case, that tion will be completely interrupted. Then
of an arbitrary phase-factor, averaging to zero; for the
r2=b2+v2fit (59)
transition case, that of complete off-diagonality,
and
meaning that the molecule has certainly changed its
state. In either case S(b) contains only the 5„,m'5w<
factors, and a simple summation shows that (a\P\b) = K f At exp(iuabt)(b2+ vW)-*°' " 3 . (60)
•'-»
S(b small) = 1 or averages to 1. (55)
Now we define
The problem of interpolating between the region of
large b, where S(b) is small and satisfies (54), and that x—vt/b, k = bwab/v, (61)
of small b, or (55), is insoluble in principle. However, following Foley; 9 then (60) becomes
several simple models and some general reasoning have
led us to conclude that a definite upper limit for the k f" eihx
real part of a, and a probable lower limit, can be (a|P|J) = dx . (60')
assigned on the basis of two rather arbitrary choices of b2vorbh-L„ (l+x2)3orl
S{b), while a third choice gives a good working approxi-
These integrals can easily be done by contour integra-
mation.
tion; however, the main features of the two solutions
Figure 1 illustrates the various possibilities. Our
are the same. Both integrals have nearly the same
three approximate "ideas" for curves are labeled "ap-
value as at £ = 0 ("fast" collisions or degenerate levels)
proximation # 1 , 2 , 3 . " # 1 is essentially the S(b) curve
for a fairly large region of the parameter k (relating
used in the phase-shift theory.
speed of collision to frequency) approaching k=l.
5#i(6) = l-cos(25 2 (6))». (56) Then there is quite a rapid falling off until at k = 4 or 5
(60') essentially vanishes.
It is easy to show that because of the factor b in (52a) We may interpret this according to the uncertainty
this is an upper limit for all conceivable shapes. One can principle. As long as the collisions are sufficiently fast
look upon the true curve as an average of several curves (&<1), the energy-difference corresponding to oiab acts
like # 1 , with different asymptotic strengths; then it as though it were negligible: AEAt>h implies that there
might well resemble curve (a) or (b). # 2 is the approxi- is a region of "vagueness in energy" of the width h/Al,
mation obtained by using S2(b) from (54c) to the point with At the time in which a collision occurs. Since the
at which S= 1; then we use exactly (55). # 3 is a rather P-matrix element gives, by (51), the probability of
arbitrarily set lower limit occurrence of transitions between the states a and b,
these transitions will occur as though these states were
5# a (6) = l - e x p [ - 2 5 2 ( 6 ) ] . (56')
resonant throughout the region k<l. This is a point
We will find that the three approximations are never which is frequently ignored in the theory; qualitatively,
more than 20 percent from each other, which is usually it was previously made by Foley. We shall find oppor-
less than experimental error. Where the experiments are tunity to use the "region of resonance" idea repeatedly
more accurate, # 2 does indeed seem to be the best in the applications of our theory. On the other hand,
choice. for large k—slow collisions—no non-adiabatic transi-
tions will occur, since the motion is adiabatic.
5. Some Remarks on the Matrix "P"
SUMMARY OF THE THEORY
Before concluding the theoretical part of this work
it is well to make some remarks about the magnitude At this point it will clarify the exposition to present
of the a, b elements of P as a function of the separation a summary of the theory up to this point. We have,
in energy of the levels a, b. An element of P is given by essentially, found the line shape of a pressure-broadened
spectral line under the following assumptions:
(a) that the relative motion of the molecules is classical; this
{a\P\b)= f d/expCWXalffiMI*). (57) enables us to write down the generalized Fourier integral (2).
(b) that the time between collisions is long compared to the
duration of the collisions; this enables us to separate the Fourier
Hi(t) will always be either a first- or second-order dipole- integral into component parts referring to the separate spectral
dipole interaction; therefore it will have a form like lines ((16) and (17)), to find a general expression for the correlation
function (42), and to write down the broadening cross-section a
H1(t) = Kr-s or Kr~\ (58) occurring in this expression in terms of collision operators T (43).
(c) that the collision operators are smooth functions of the
We can always assume that the paths followed by the collision parameter b between large b (where we have the ex-
pansion (51)) and small b, where complete interruption occurs.
colliding molecules are straight lines, since one can Then we are able to apply the interpolation process, and obtain
show that if the molecules collide with sufficient strength for u the expression (52), 5(&) given by the formulas (56) fitted
to have curved paths it is quite certain that the radia- to the large —6 expansion (54).
14
656 P. W. ANDERSON
TABLE I. Comparison of theory and experiment. function with respect to inversion.
Theory Experiment ^+=^+K+4'-K, ^ - = ^ + K - ^ - K . (1)
Approx. # i #2 #3 Here -\-K and —K denote the "single-ended" states,

&m= 0.86 0.77 0.68 cm"1'/atmos. 0.74 cm - 1 /atmos. whose degeneracy is raised by the well-known inversion
or "tunnel effect."
TABLE II. Comparison of theory with experiment-—relative First we consider the interaction between two am-
breadths in the ammonia spectrum. monia molecules (first-order only). The Hamiltonian of
this interaction is
Experi- Quasi-
Theoretical mental empirical
breadths: data: formula: _(tfi-r)(»2-r)
•r)-|
J = K= this thesis Bleaney Bleaney H -- tfl-tf2- — r (2)
(a) U» M W) M
2 1 2.6 2.6 2.7
3 1 2.3 2.4 2.2 where yi and y 2 are the dipole moment matrices of the
3 2 3.3 3.2 3.4
3 3 4.5 4.5 4.5 two molecules considered, and r is the distance between
4 4 4.6 4.5 4.6 the molecules. Equation (2) can easily be expanded in
5 1 1.8 1.8 1.6 terms of direction-angles in a system of reference in
5 2 2.5 2.6 2.5
5 3 3.3 3.3 3.3 which the impact parameter b is the polar axis of
5 5 4.8 4.7 4.65 spherical coordinates for each molecule. Then 0i, <pi
6 3 2.8 3.1 3.1 and 02, <pi are the polar angles of HI and M2 referred to b,
6 4 3.5 3.6 3.6
6 6 4.8 4.7 4.7 while ^=tan _ 1 [(rf)/(&)] is the angle which r makes
7 5 3.7 4.1 3.8 with b. The result for H is
7 6 4.3 3.9 4.3
8 7 4.3 4.1 4.35
10 9 4.5 4.2 4.5 H=pt/r3[cos8i cos0 2 (l — 3 cosV)+sin<?i sin# 2
11 9 4.1 2.9 4.2 X (cos(pi— (pi) — sinhp cospi cos<p2)
+ terms in cosi/' sini/']]. (3)
Data are half-widths in c m _ 1 X 1 0 - 1 at 0.5 mm Hg pressure.
We must obtain the matrix elements of the com-
The procedure to be followed in computing line binations of functions of 6 and (p which occur here.
shapes is this: given an assumed type of intermolecular One theorem simplifies this problem greatly. All matrix
forces, one computes the matrix P (51') in the collision- elements vanish between states which are both + or
parameter coordinate system. The interaction energy both —, while the ( + | u | — ) elements are precisely
Hi is given by (10). The elements of this matrix in- those of the ordinary symmetrical top for states of the
volving the initial and final states are then averaged correct / , K and M. The proof is obvious upon observ-
over directions according to (54b) and (54c' and c"). ing that
The 5 2 averages (54c) are employed in the approxima-
tions (56) and the <r-integral (52) to obtain a broadening (K\v\K) = -(-K\v\-K); (K\9\-K)9&. (4)
cross-section, which gives the line-breadth from (28); The latter holds true within the order of the tunnel
a shift, if present, can be computed similarly from .Si frequency compared to a vibrational frequency, or 10~3.
(54b), (52), and (28). Then (1) leads directly to our theorem.
PART II. APPLICATIONS OF THE THEORY We present here a tabulation of the required matrix
TO SPECTRA elements: 16
A. Cases Involving P e r m a n e n t Dipole (JKM\ cosfl \J KM) = KM/J{J+1)

Interactions
(/ KMlsinBe^lJK Mil)
1. Ammonia Self-Broadening
= =FK/J(J+ \){{J±M+ 1)(/TM))!,
We shall begin with various problems involving per-
manent dipole interactions, since there is much experi- (JKM\cos6\J+\KM)
mental work on them and since, in general, the
"difficult" term (^middle vanishes in these problems. ![(/+l)2-A2][(/+l)'-Jlf2]}*
Self-broadening in the inversion spectrum of ammonia
has been studied most extensively. (/+l)[(2/+l)(2/+3)]*
The states of the ammonia molecule, exclusive of
[(/ 2 -A 2 )(/ 2 -ilf 2 );p
vibrational and electronic quantum numbers, are {JKM\cosB\J-\KM)=-
specified by four quantum numbers: / ( t o t a l angular 7[(2/+l)(2/-l)]»
momentum), i f (equatorial), A s y m m e t r i c top quantum
number), and ± , denoting the symmetry of the wave- 16
R. de L. Kronig, and I. Rabi, Phys. Rev. 29, 262 (1927).
15
(JKM\suite**\J-l KM± 1) At this point it is very easy to calculate the correct S2

2
sums. We present merely the result, which includes the
[_(J -K*){J^FM-\)(J^FM)J identical sums for initial and final states. For the simple
~ /[(2/-l)(2/+l)]» ' Stark effect we get the following S 2 sum, applicable to
collisions of the radiating molecule, quantum numbers
( / if M | sinte^* | J + 1 K M± 1) Ji and Ki, with molecules of any J2 and K2 (n2 in the
general formula (46) of Part I for the two-molecule case
{[(/+l)2-if2](/+l±Af)(/+2±jW)j»
being here J 2 , K2).
(/+l)[(2/+l)(2/+3)]*
KSK*
S2(J2, K2)—~
Of course, these elements are to be understood, ac- o j W / ^ + l X ^ / d - l )
cording to our theorem above, as always off-diagonal
in the index ± , which is not explicitly written in. (Stark effect). (8j
There are two kinds of matrix elements of the "inter-
action matrix" P which do not vanish, for all practical For the two rotational resonance 5 2 's, in case the
purposes: (a) those corresponding to a simple "first- radiating molecule J\, Ki collides with a molecule with
order Stark effect" interaction involving the J—*J / 2 = / i ± l , we get
elements of (5), and (b) those corresponding to rota-
8 u4
tional resonance and involving / — > / ± l elements of (5). 52(/i-l,X2) = -
All others are negligible because of the large energy- 9 bVh2
changes (order of 10 cm - 1 ) involved; we neglect the
second order effects and vibrational resonances, the (.J1*-K1*)(.Ji2-K*i)
latter because few vibrationally excited molecules are X- -(rot. res.) (9a)
/12(2/1+1)(2J1-1)
present.
However, it is necessary to observe that there is no
resonance effect as far as the inversion frequency is 52(/1+I,A:2)=-
concerned. The parameter k of Section H-5, Part I, is 9 bW
small for the inversion frequency, and therefore col-
KJi+iy-KSjLiJi+D'-Kt'i
lisions of + molecules with + molecules or of — with X- • (9b)
2
—, in which each type must undergo a -\—>— transition (/1+l) (2/1+l)(2/1+3)
(total energy change about 1.6 cm - 1 ) are as effective in
For collisions with molecules of / 2 = / i ± l , one must
broadening as the truly resonant I, _ , collisions. add (9a) or (9b) and (8) together to get the total 5 2 sum.
This applies as far as both effects (a) and (b) are con- These are (S2) outer sums j no (»S2) inner sums enter in this
cerned. The parameter k is actually problem because the elements of y are all off-diagonal
in the non-degenerate index ± .
bu 6(A) X10~ 8 X 3X1.6 X10 18 X2ir We must now obtain the quantity a (or rather the
&=—= =J(A)X0.035. separate cr(72, K2) to average over J2 and K2 by Part I,
v 8X10 4
Eq. (46)). Without bothering to do the integrals," we
(5)
present here the three possible o-'s for our three approxi-
We shall never consider impact parameters b greater mations of Part I, Section H-4. We use the notation
than ISA. A k of 0.5 may lead to an error of a few per-
cent. S2(J2,K2) = A*/bi- (10)
These considerations lead us to the following pro- Then the three approximations give
cedure: we shall calculate the P-matrix (Part I, Eq.
(51)) ignoring the exponential time factors entirely, and #1. <n(/ 2 , K2) = pB(J2, 2r 2 )2rr(l.lL4),
then substitute the resulting P-matrix into the sum S 2 #2. <ri(Ji,Ki) = PB(Ji,Ki)2irA, (11)
(I, Eq. (54c)). Since we are ignoring the time-factor, it #3. <r3(/2, K2) = PB(J2, K2)2w(0.885A),
is permissible to perform the time-integral of (3)
previously to taking the matrix elements. We use the
formulas: <r(J2, K2) = 2irApB{J2, K2)XC,
H = J 2 + vV cost =b/r sin^ = vt/r, (6) where C has the values
and the result is C=l.ll #1,
C=1.00 #2, (11')
Exit 2M2
-[_— COS01 COS02 C= 0.885 #3.
/: h b2vh
-sinfli sin02 sinpi sin<s2]. (7) n H . Jensen, Zeits. f. Physik. 80, 448 (1933).
16
658 P. W. AND ERSON
Then we have Eq. (46) of Part I computed correction for the rotational resonance effect,
from Eq. (9). These values are normalized as explained
<r= X) <r(Ji> #2), above. Column (d) gives Bieaney and Penrose's experi-
mental data, and column (e) gives predictions from a
and theoretical formula of Bieaney based on the arbitrary
(Av) cm~'= (nv/2irc)<r (12) assumption that the broadening effect of a collision
which determine the line-breadth from Eqs. (8)-(ll). depends on the maximum interaction energy of the two
Two types of experimental information are available. molecules. This formula is
The most extensively measured single line in the am-
monia inversion spectrum is the Ji = 3, Ki = 3 line.18'19 (AiO#~ (*?/(/(./+1))*.
For this line we can neglect, with one or two percent It does not seem very well based since it takes no
error, the effect of the rotational resonance sums (9a) account of rotational resonance. All breadths are in
and (9b). Then we can compare our theoretical values 10™» cm -1 at 0.5 mm pressure.
from (8)—(12) with the experimental results.§ The The computational error in our approximate method
result shown in Table I (all breadths are half-widths in for the rotational resonance is about 0.1 crcr'XIO -4 .
cm -1 /atmos. a t 0°C). This shows the agreement of our Bieaney states that his experimental error is of the order
theory with experiment as far as absolute values of the of 0.2 cnr'XIO - 4 . Within these errors, our agreement
broadening are concerned. Certainly the experimental is excellent, except for the last four lines, which seem to
value lies within our computational error, and it is allow no reasonable explanation. The fortuitous agree-
interesting that the working approximation # 2 is ment of Bleaney's formula (e) with experiment is
fairly close. interesting, but not, apparently, significant.
The second type of experimental data available is
that on the relative breadths for all the lines in the B. Other Cases: HCN and HC1 Vibrational
ammonia spectrum, taken by Bieaney and Penrose.20 Band Spectra
In Table II we present our theoretical results on relative
21 16
values. In order to avoid the ambiguity due to the Lindholm ' has taken a great deal of experimental
interpolation procedure, we have normalized the theo- data on line-breadths in the vibrational bands of HC1
retical value for the 33 line to fit the experimental value and HCN. He has also made theoretical calculations of
9
precisely. Columns (a) and (b) give the J and K values the line-breadths for these spectra, as has Foley.
for the various lines; column (c) gives our values using Before giving our own theory, it will make our procedure
the Stark eflect, from Eq. (8), plus an approximately somewhat clearer to indicate how these older theories
worked.##
For cases in which resonances occur, as in the mole-
THEORETICAL CURVES
I THIS PAPER cules under consideration, the standard practice has
been to find a mean square direction-averaged energy
EXPERIMENTAL DATA perturbation due to resonance, for instance22
UNCORRECTED, LINDHOLM.
FIG. 7
' O SAME CORRECTED FOR
\ EXPERIMENTAL SMEARING J2
\ «F%)K/-^/-D =
-0)'f ( 2 / - l ) ( 2 / + 3 ) J R3
(13)
in the case of rotational resonance of dumbell molecules.

Then this energy change is inserted as the diagonal B(t)
in the simple phase-shift theory. This is for precise
resonance; if imprecise resonance occurs, the "quasi-
resonance" type of calculation such as the following is
used. If two levels are separated by a small difference
in energy A, and an interaction V'« is assumed between
them, the energy E of the perturbed levels is computed
from the secular equation
(A/2)-£ Vik
= 0. (14)
FIG. 2. Comparison of theoretical and experimental results. Vik (-A/2)-£
18 21
B. Bieaney and R. P. Penrose, Proc. Roy. Soc. A189, 358 E. Lindholm, Zeits. i. Physik 109, 223 (1938).
(1947). # # Lindholm in reference 10 himself stated that he found this
» C. H. Townes, Phys. Rev. 70, 166 (1946). type of theory very bad in atomic resonance problems and that he
§ Bleaney's experimental value seems the soundest. used it only because his own rigorous theory was too difficult in
,0 these cases.
B. Bieaney and R. R. Penrose, Proc. Phys. Soc. London 59, 22
424 (1947); 60, 540 (1948). H. Margenau, Rev. Mod. Phys. 11, 1 (1939).
17
Then (4) Second-order Keesom alignment forces, operative primarily

£=±((A/2) 2 +(F r t ) 2 )». when the colliding molecule is not in resonance with either state
of the radiator.
In the application of the method used by these
Effect (1) is ignored. Effect (2) we can compute on
authors, a mean square perturbation such as (13) is
much the same basis we have already used for am-
introduced as Vu,, and that solution of (14) chosen
monia: in fact, we can take over the S 2 sums (9) bodily
which reduces to the correct level at Va = 0. Then this
from the ammonia problem, if we specialize to K=0
E is used in the phase-shift method.
and realize we must take only half of the sum there used,
Two objections can be made to this idea, considered since only one state of the two involved in the radiation
simply as an interpolation method between the first is undergoing the first-order perturbation.
order, resonance region and the second-order, r~e force
region for V. In the first place, the criterion for the 4 M4 (/i+l)2
point at which the change in type of forces occurs is S2(J2 = J1+1) = ,
whether V=A, apparently independent of the speed of 9JVA2(2/1+l)(2/1-r-3)
collisions, which enters into the criterion for our own
theory. This can be shown to be not a serious dis- 52(/2=/i-l) =
crepancy, using the fact that on both theories a col- 9bVhi(2J1-l)(2J1+l)
lision must be both long and strong enough to cause a
The effect (3) is not a precise resonance, because the
certain P-matrix element before it has an appreciable
moment of inertia of the H C N molecule depends on
effect. However, the two criteria are not identical; the vibrational quantum number. If we express the rota-
quantitative results of the two theories differ con- tional energy approximately as
siderably.
The second objection is more fundamental. A par- £«(/) = £„/(/+1). (16)
ticular choice of the solution of (14) is made; thus in Then 23
general it appears that a center-frequency shift should
Bv=lA87&-0.0093(vl+i)
be expected, larger the closer to resonance the levels +0.0007(» 2 +l)-0.0108(i>3+i)
come. However, experimentally and theoretically it can
be shown that there is no shift due to precisely resonant B is therefore about 0.04 c m - 1 different in excited and
collisions: both signs in (14) occur when A = 0. This normal states. Since the quantum exchanged in the
contradiction is a true defect in the quasi-resonance rotational resonance is about 2BJ wave numbers, the
theory. total amount by which the excited and ground states
We are not able to present a good theory of the inter- fail to be in resonance is approximately
polation region between first order (r~3) forces and r - 6 Ao)(off resonance) = 0 . 0 8 / c m - 1 . (16)
forces, or even, in the particular case of the rotator, of
the second-order, r~e, region. Therefore, since the Then our parameter k is
answers for HC1 depend strongly on how these regions k=bAa/v=2.9XiO-"b(A)J. (17')
are treated, we do not present detailed work on this
molecule. We are able to attain fair (30 percent) agree- The maximum collision parameters for this problem are
ment with Lindholm's observed widths for this spec- about 30A; then
trum, but no better agreement than that of his quasi- £<0.09/ (17")
resonance theory. However, his theory predicts much
greater breadths for the P than the R branch, which are and it is only for values of J~ 10 that an appreciable
not observed; our method of treating near-resonance correction for failure to resonate should appear. We
could never lead to such a great difference, and thus shall ignore this correction entirely, both because it
agrees better with experiment. should not be large for the interesting lines and because
We have been able to make fairly quantitative calcu- it is extremely difficult to compute. Then for the effect
lations for the H C N molecule. This molecule is a (3) we also use Eqs. (15), except that / 1 must be
dumbell rotator with a large dipole moment 2.65X 10~18 replaced by / i + l .
esu. The observed spectra are the P-branches of the two Lindholm finds a large correction for quasi-resonance
bands at 11500 crer 1 ( » i = l , v3=3) and 12500 cm" 1 in this case, not depending much on / . We do not agree
(»8=4). There are several types of interactions to be with this result. In addition, this is the point in which
considered. his HC1 treatment differs greatly from ours: we find
that if (as is necessary for most lines) a correction is to
(1) Vibrational resonance; this is neglected because the dipole be applied for this effect, it should be fairly symmetric
moment for this interaction is quite small. between the P and R branches of the band.
(2) Rotational resonance of the ground state ( / = / i ) with unex-
cited colliding molecules, having / = / i ± l .
23
(3) Rotational resonance of the excited state ( / = / i — 1 ) with G. Herzberg, Infrared and Raman Spectra (D. Van Nostrand
unexcited colliding molecules having J—Ji, Ji=2. New York, 1945), p. 393.
18
660 P. W. ANDERSON
Effect (4) we have not been able to treat rigorously TABLE III. Comparison of measured and computed diameters.
by our own methods, since the sums required become
Line-breadth
quite difficult for the second-order alignment forces. It (a) (b)
seems likely that the results for these forces obtained Kinetic-
Molecules Measured theory
by the simple direction-averaged phase-shift theory Radiator Perturber diameter A diameter A diameter(A)
should not be very far wrong; in order to include this
NH 3 He 1.5 2.4 3.2
important effect, we have simply borrowed the result NH 3 H, 1.9 3.5 3.4
of Lindholm and Foley for these forces (they agree) NH 3 N2 2.6 6.4 3.4
and added their line-breadths for these forces to our NH 3 02 2.5 4.8 3.5
NH 3 A 2.6 4.6 3.7
own for resonant forces. This simple addition is legiti- H20 Air 3.3 5.6 3.5
mate, since the important contributions these forces
make come from different collisions from those con-
tributing to the resonant line breadth. are the corrected data. In addition, we present three
We use the "working approximation" number (2). theoretical curves for comparison: (1) is our own theory
Then the cross section is given by (11) with C= 1, and for resonance, borrowing the Lindholm-Foley values for
A defined by (10). We use S 2 from (IS). The line the small contribution of alignment forces; (2) is
breadth can then be computed from (12) and Eq. (46) Lindholm's curve for this spectrum; and (3) is Foley's
of P a r t I, for all resonant collisions (effects (2) and (3): published values for the H C N spectrum. Lindholm's
sum over Jz = Jx—2, / x — 1 , J\, / i + l ) with a con- curve is appreciably lower than Foley's for two reasons.
tribution computed from Lindholm's figure (7) added He has a correction for quasi-resonance, which is, in our
in for the alignment forces. opinion, unfounded. Foley need not use this correction
Before comparing our results with Lindholm's experi- in any case, since his theory applied to the 14/J funda-
ments in Fig. 2, we should like to make some comments mental band. In addition, it seems that even taking into
about his experimental data. These were taken in the account this correction Foley's values are some 20
photographic infra-red, by means of densitometer percent higher than Lindholm's; perhaps this is a
analysis of plates. His best data are those on the 11.5 numerical error in Foley's work, since we get agreement
X10 3 c m - 1 band at pressures of 25 cm and 58 cm Hg; with Lindholm's values upon re-computing by their
he also has data on the 12.5X10 3 crrr 1 band at three common method.
pressures, 25, 40, and 58 cm Hg. The most obvious fact Two forms of agreement can be claimed for our
about his original breadth data is that they are not pro- curve: (a) excellent agreement with the experimental
portional to the pressure, as one would expect on any variation with / , a factor which is unchanged by any
theory at these pressures, but that instead they are method of correcting for experimental error; and (b)
given by good agreement quantitatively with the corrected data.
&v = const.-)- (const.) X pressure, (18)
C. Foreign Gas Broadening: Van der
within the experimental fluctuations. One can interpret Waals Interactions
the constant in this expression as an experimental
broadening due to slit width, finite resolving power, Broadening of microwave lines by pressures of
etc.; this would be the case if these effects were sum- foreign gases has been the subject of some experimental
marized by a dispersion form of curve, to be smeared investigations. 20,24 The theoretical approach to the
with the true line breadth. This constant is about 0.13 problem of foreign gas broadening has generally been
c m - 1 in half-width, a very reasonable value for experi- that of computing the Van der Waals interactions be-
mental effects. To get the true width one should sub- tween the molecules concerned and using these to obtain
tract this constant from all data.**** broadening cross sections. We have followed this pro-
cedure with our more accurate theory. In general, it can
In Fig. 2 we present both sets of experimental data:
be shown that the r~e term in the interaction between
the actual observations at 58 cm extrapolated to 1 atm.
two molecules is, for the Debye induction type of
pressure using simple proportionality to pressure, and
forces:
the same with the constant term in (18) subtracted,
which we expect to be the true impact theory line #ind.= (<W/2r 6 )(l-r-3 cos 2 0), (19a)
breadth. The legend explains that the lower (o) values
where a% is the polarizability for the non-polar molecule,
**** In a private communication, Lindholm suggests that while assumed isotropic, m the dipole moment of the polar
there must be some error of the type suggested here, he cannot molecule, and 6 the angle which yi makes with the
agree that it could be quite so large. One may consider that the intermolecular distance r. Very similarly, the London
correction for experimental error is uncertain by perhaps half
its total value, and thus that the data is uncertain by 0.06 cm" 1 / dispersion forces for two interacting non-polar mole-
atmos. Gordy has recently reported microwave measurements on cules, one assumed isotropic, are approximately given
the HCN rotational spectrum (Rev. Mod. Phys. 20, 668 (1948))
and it is to be hoped that the observation of one or two of these
24
lines will definitely settle the question. G. E. Becker and S. H. Autler, Phys. Rev. 70, 300 (1946).
19
by cross sections a calculated by means of the known

Van der Waals forces come out usually to be consider-
ably smaller than the kinetic theory cross sections com-
ffdisp. = ( 3 a ' + a " + ( « " - « ' ) COS20), (19b) puted from viscosity, etc. This means that the forces
which cause broadening must be the higher order and
where a 2 is still the polarizability of the isotropic mole- exchange forces which come into prominence a t the
cule, Ii and Iz are the fundamental frequencies from diameter corresponding to the kinetic theory cross-
one-term dispersion formulas, or the ionization poten- section. As would be expected, the measured broadening
tials, for the two molecules, and a' and a" are the cross sections are at least as great as the kinetic theory
polarizabilities along the two different axes of the ones in most cases.
polarizability ellipsoid of the non-isotropic molecule, Table I I I summarizes the situation very well.
22
assumed to be an ellipsoid of rotation. (A somewhat Column (b) is from Bleaney and Penrose's paper,
similar formula holds in case the anisotropic molecule except the H 2 0 line, which comes from Becker and
24 26
has three different polarizabilities.) 6 is the angle Autler. Column (c) is from Stuart's book.
between the axis of the polarizability ellipsoid and the One more problem was computed. The microwave
intermolecular distance r. spectrum of oxygen has been measured 26 in some detail.
For states separated by microwave frequency dif- This case is even more complicated that the preceding,
ferences, the isotropic parts of (19) make no con- because transitions among states having different total
tribution to broadening, and one need consider only an angular momenta (/) but the same K, or molecular
interaction of the form quantum number, must be considered. J = K + S , where
S is the spin angular momentum of O2, \S\ = 1 . The
ff=const.X(cos29/r6). (20) splitting of the states J=K±\, K is very small, since
this' is the line observed at microwave frequencies, and
52 sums can be computed using this interaction, and thus this type of transition occurs easily upon collision.
inserting appropriate constants for various cases. These Another theorem of Racah was used to simplify this
sums are quite difficult, both because the cos20 term computation. Here the agreement with experiment was
introduces essentially a second-order Legendre poly- somewhat better than would be expected from the
nomial symmetry into the matrix P, and because the kinetic theory diameters and Table I I , but this is not
difficult term (.Sumner does not vanish. They are evalu- significant since the experimental data are only on the
-1
ated easily only by using the group-theoretical methods entire unresolved set of lines near 2 cm .
of Racah; 15 however, using these methods they can I should like to express my gratitude for the wise
be done in general, even for forces of more complicated direction, help, and encouragement given me in my
symmetry. work on this problem by Professor J. H. Van Vleck.
We do not present this evaluation because the 26
H. A. Stuart, Molekttlstruktur (Verlag J. Springer, Berlin,
results, in general, do not have any relation to experi- 1934), Table II.
26
mental results. It is found that the collision broadening J. H. Van Vleck, Phys. Rev. 71, 413 (1947).
20
Theory of Ferroelectric Behavior of Barium Titanate
Ceramic Age 57, 29-34 (1951)
This is my first review paper — a forerunner of the "soft mode" paper of 1958. Bill
Shockley hired me at Bell to work on his ideas on ferroelectricity and I did, but soon
became more interested in generalities, feeling that calculations were premature. He
was unhappy with my work, which seemed to ignore his ideas for detailed
calculations; but he taught me a lot of solid state physics and the background and the
experience were both valuable to me.
21
Theory of
Ferroelectric Behavior of Barium Titanate
By P. W . ANDERSON
I — Review of Experimental Information
Bell Telephone Laboratories
M u r r a y Hill, N . J .
gram in Fig. 1, in which the effect of
T HE purpose of this paper is, first,
to give a review of certain typical
experimental information on fer-
applying a small extra voltage at vari-
ous points of the loop is indicated, of ferroelectrics, about 16 x 1 0 _ 0
coul/cm2. ,
roelectrics, particularly on barium ti- shows that the effective reversible di-
tanate and its sister-structures, which electric constant er(,v- for small fields The shapes of these hysteresis loops,
have absorbed most of the recent ef- is only about 1000, at all points of as in ferromagnetism, are very sensi-
fort in this field; and second, to pre- the loop, and that for small fields there tive to small amounts of impurity and
sent a rough idea of the present the- is little or no hysteresis loss. To those even to the previous thermal and elec-
oretical picture of the atomic mechan- who are familiar with ferromagnetism, trical history of a single specimen.
isms underlying the phenomenon of this will indicate that small fields are Nevertheless, PB and, usually, cny.
ferroelectricity, again concentrating on causing little or no domain motion, roughly are constant for a given sub-
barium titanate. Since it does not seem since £,.,,v. also is about equal to the stance.
necessary to attempt to give complete saturation dielectric constant, which The second similarity between ferro-
references in a paper of this character, represents the dielectric constant of a magnetism and ferroelectricity lies in
particularly because a quite complete sample in which all domains are com- the Curie point phenomenon. Fig. 2
bibliography of the subject is avail- pletely aligned, and cannot be re- shows a plot of €rev. versus the tern-
able in Von Hippel's excellent review aligned by the measuring field.
article1, it should be made clear at the
start that the writer is borrowing ma- On the ferroelectric hysteresis loop,
terial from various sources.** one can define quantities entirely
analagous to those which characterize
The first and most obvious question
to be answered is: what is a ferroelec-
tric, and why is it so called? The name
does not come from any relation to
iron, but from the very close analogy
between the phenomena of ferroelec-
tricity and the better-known ferromag- Fig. 2: Dielectric constant as a
netism. By definition, ferroelectricity function of temperature for a
is the dielectric analogue of ferromag- single crystal of barium titanate.
netism2.
Fig. 1 shows the most striking point perature for a single crystal of BaTi0 3 .
of this analogy: The ferroelectric hy- A very high peak occurs at the "Curie
steresis loop in barium titanate (Ba- point" TL. = 120°; above this peak the
TiOa). Here we are plotting the di- substance behaves like a relatively nor-
electric induction D versus the applied mal dielectric of high dielectric con-
field E, as can be done on a cathode- stant, while below 120°, as shown in
the Fig., there is complicated crystal-
ray oscilloscope with certain rather
direction dependence for £rev. In this
simple circuit arrangements. The full region we have also the field depend-
scale of D on these diagrams roughly ence of the dielectric properties indi-
is 2 x 107 volts/cm, while the full cated by the hysteresis loops of Fig. 1.
scale in E is only about 4 x 103 volts/- Returning to the region above Tc, Fig.
cm. Thus the mean or "true" dielectric Fig. 1: Hysteresis loops in BaTi0 3 .
3 shows the D versus E plot here. At
D magnetic loops: the coercive field temperatures only slightly higher than
constant E = —, corresponding to the EL. which is required to return the in- Tf., shown by the curve labelled T c +
E duction to zero, the remanent polariza- 2°, the hysteresis loop has disappeared,
slope of the loop as a whole, is about Dr flattening out to a single curve which,
10*. tion P r = at zero applied field, nonetheless, still shows some evidence
of dielectric saturation in the curvature
On the other hand, the second dia- 4r
,at high fields. At still higher tempera-
and the saturation polarization, the tures the dielectric constant is lower,
* Presented at Tenth Symposium on Ceramic
Dielectrics, School of Ceramics, Rutgers Uni- polarization P8 of a hypothetical speci- and the characteristic is perfectly linear
versity, New Brunswick, N. J., November 17, men with all domains parallel at no
1950. as in an ordinary dielectric.
** The general rule followed in this paper is field. This latter quantity can be ob-
that if a development is covered in some detail tained by linearly extrapolating the Another characteristic of the region
by Von Hippel, no reference for such is noted.
1
A. Von Hippel, Rev. Mod. Phys. 22, 221 saturation line at high fields back to above the Curie point which is strik-
(July 1950). the E = 0 axis: this is shown in Fig. ingly similar to the behavior of ferro-
- Thus an eleetret, by this definition, is not
a ferroelectric, since very few of the ferro- 1. In BaTi0 3 the saturation polariza- magnets is the Curie-Weiss law. This
magnetic properties have analogues in elec- tion is larger than in the other classes law states that the inverse of the sus-
trets.
(Reprinted from CERAMIC AGE, Newark, N. J., April, 1951)
22
ceptibility (magnetic or electric) is er two shortening, to become the c and
proportional to the temperature, i.e. a axes respectively of the tetragonal
structure. The ratio c to a is 1.01 at
1
— = A + BT (1) room temperature.
£
Recent x-ray results indicate that in
1 addition to this general deformation
and that actually the intercept at - - the relative ionic positions shift slight-
^ ly, the Ti ions moving .044 A in the
c
direction of the c-axis, the O ions mov-
= 0 is approximately Tc, i.e. ing very slightly the other way4. This is
(2) confirmed by the fact that the spon-
c ~ const
— T-Tc / taneous polarization P s has been found,
4. by experiments on single crystals, to
The correctness of this law is dem- lie along the c-axis. It is fairly certain
onstrated in Fig. 4. The small devia- that the positively-charged Ti ions
tions near the Curie point are found Fig. 4: I t as a function of
temperature in BaTiC>3. move in the direction of P s and the
also in ferromagnets, but are not quite negative O ions in the opposite direc-
the same in detail. tion.
tions add up to zero polarization for
Thermally, the electric and magnetic the specimen as a whole. Through One can think of the tetragonal de-
Curie points are not quite so similar beautiful theoretical and experimental formation as an electrostrictive effect.
in structure. Although both show sin- work3 the domains in ferromagnets
gularities in the specific heat at this have come in quite recent years to be -- - - CONST X T / f l C
point, the ferroelectric singularity is
considerably smaller for the BaTi0 3
class than is the usual ferromagnetic
specific heat singularity. Also, while
understood and definitely "pinned
down" theoretically, as well as made
visible experimentally with the pow-
der pattern technique. The opposite
"* ("" -S*)'
!>=
rH ^ k
V
most (not all) known ferromagnetic order is the rule in ferroelectrics: the \
Curie points are thermodynamically domains are easily visible and, in some
classified as second-order, or A-point, cases, have been the first observed evi-
transitions, with no latent heat or dis- dence of ferro-electricity: it has been
continuity in volume or other "first easier to identify a ferroelectric opti-
order" properties, it is possible that the cally by looking for domains than
BaTi0 3 Curie point is a very slightly electrically by finding a hysteresis loop
— although the latter, it must be said,
is a far surer test, since domain-like .
Tc*2" twinning is found to be surprisingly
common. Fig. 6: Proportionality
To explain why the domains are of strain and PB2.
so easily visible we must first go into It is found that an "electrostrictive"
the crystallography of BaTiOs. Above law
the Curie point, this substance crystal-
lizes in the cubic perovskite structure, A ( — ) = const x P- (3)
as shown in Fig. 5. One may think of
the large barium and oxygen ions as
lying on the corners and face-centers is obeyed if we insert for P the satura-
of a cube, respectively, together mak- tion polarization P s . The same law,
Fig. 3: D vs. E for BaTi0 3 above ing up a face-centered structure; the with the same constant, is found to
the Curie point. small titanium ions are then at the hold above the Curie point for polari-
cube centers, surrounded by an octa- zations due to applied fields. There, of
hedron of oxygen ions. At T c and be- course, the effect is true electrostric-
first-order change with a true volume tion. Fig. 6 shows how equation ( 3) is
discontinuity. Further reference will low this structure is found, by x-rays,
to deform into a tetragonal unit cell, obeyed. The deviations at lower tem-
be made later to some of these points. peratures are due not to the failure of
with one of the three cubic axes (arbi-
A third striking experimental sim- the law but to inability to properly
trarily of course) lengthening, the oth-
ilarity between ferromagnetism and saturate the crystal when measuring r^,
ferroelectricity lies in the domain even in the so-called "single-domain"
structure. It is well-known that in or- crystals. The cubic magnetic materials
der to explain the magnetic hysteresis also have this same slight tetragonal
loop Weiss and others postulated that distortion, although in these the effect
a ferromagnetic specimen is divided always is thought of as magnetostric-
into a system of regions, or domains, tion superposed upon a basically cubic
each polarized to saturation in a certain structure, while the BaTiOs effect is
direction. At zero applied field, how- so large that calling it electrostriction
ever, the magnetic domain polariza- is rather a stretch of the imagination.
a
For a review see C. Kittel, Rev. Mod. Because of this tetragonality a polar-
Phys. 21, 541 (1949). ized piece of BaTiOa is optically aniso-
* G. C. Danielson, unpublished.
W. Kaenzig, Phys. Rev. 80, 94 (1950). Fig. 5: Crystal structure of BaTiC>3. tropic, with different indices of refrac-
23
tion in the two different directions; C the latter two quantities are large, the
with a polarizing microscope, then, it strain = A ( — ) = co'nst x P- result is a large and useful effect.
is seen easily that a typical crystal of a In another sense, in an actual un-
BaTi0 3 is subdivided into many do- (4) polarized crystal or ceramic the effect
mains, polarized in various directions, of a field is not at first to give a linear
just as the existence of the hysteresis d c dP piezo-effect, because the deformation
loop would lead one to suspect. A typ- (A ( — ) ) = const x 2P of domains polarized in one direction
ical domain pattern is seen in Fig. 7. dE a dE is cancelled completely by that of do-
Thus the change of strain with field mains polarized in opposite directions.
involves the constant of electrostric- Thus, it is necessary to "pole" a speci-
tion, the saturation polarization, and men for piezo-electric work by the ap-
dP plication of high fields, and according-
the dielectric susceptibility ; since ly to leave the specimen in an at least
dE partially polarized state.
II — Review of the Theory of Ferroelectricity

Fig. 7: A typical d o m a i n pattern In commencing the theoretical part field due to all charges outside the
in BaTiOa. of this paper, one can remark that the sphere, then, is simply one of ordinary
main task of the theoretician has been macroscopic electrostatics; it involves
Precisely as in ferromagnetism, it is to try to de-emphasize the close anal- the fields of: (a) the surface charge
found always that the component of ogy of ferromagnetism and ferroelec- on the sphere left by cutting it out;
polarization perpendicular to a domain tricity, at least as far as the actual basic (b) the surface charges on the dielec-
wall is the same on each side of the mechanism is concerned. tric surfaces ( ( a ) and (b) involve
wall, so that no free charge need form The ideas involved in the present P ) ; and (c) the charge on the con-
in the wall to compensate the surface theoretical picture of the mechanism of denser plates. The result of such a cal-
charge density which would otherwise ferroelectricity are not new and not culation is the Lorentz formula ( 5 ) .
appear. Such a charge density would complicated. We must go back to 4n
have a high electrostatic self-energy. 1908, the year in which P. Weiss first The term P is, as von Hippel
proposed an acceptable, if somewhat 3
Two more lines of experimental
work, not directly connected with the ad hoc, theory of ferromagnetism.6 In has called it, a "bootstraps" term, i.e. a
analogy the writer has been develop- his proposal he cited the familiar for- term by which the polarization is cap-
ing, should be mentioned. The first is mula with which the names of Lor- able of pulling itself up by its own
the search for more ferroelectric ma- entz, Lorenz, Clausius, Mosotti, Debye, bootstraps. In more conventional lan-
terials. It is found that many com- and others have been connected in its
pounds with structures similar to various aspects:
BaTiOa — the " p e r o v s k i t e struc- 4irP + + + + + *
ture" — are also ferroelectric.5 Thus E -| (electric case)
BaTiC>3 is not an isolated phenomenon 3
as are Rochelle salt and KDP, the other F= (5)
two ferroelectric substances known; 4TTM
!p ©
this, with its relatively simple structure, H -1 (magnetic case)
contributes to the interest in work on 3
BaTi0 3 . In this formula, F is the "internal"
field, the field which actually acts to + + + + + + +
Because of the practical importance polarize the individual molecule inside
of this subject, I should also say some- Fig. 8: Internal field in a dielectric.
the material one is discussing. The ex-
thing about the piezoelectric effect in ternal applied field is E or H, the
BaTiO;t. Some confusion has arisen as guage, this is a "cooperative force"
polarization of the substance per unit term. To show how this term acts, let
to whether this effect should be called volume P or M, and the important
electrostriction or a true piezo-effect. us compute the polarization of a mole-
point of the formula is that it does con- cule in the dielectric under an external
In the precise sense of the words, be- tain a contribution of P or M.
low the Curie point the strain is pro- field:
portional to applied field in each sin- The derivation of this formula is a P mol — aF (6)
gle domain of a BaTi0 3 crystal, and matter of simple electro- or magneto-
statics. It can be best understood by This equation defines the polarizability
thus we have to do with a pure piezo- a of a molecule. Then let us find the
effect; however, in two ways the referring to Fig. 8. We take the par-
ticular molecule on which we are try- total polarization of a cc of the sub-
phenomenon is related closely to elec- stance containing n molecules, and fi-
trostriction. First, the actual piezo-elec- ing to compute the acting field, and
surround it with an imaginary sphere nally insert for F equation ( 5 ) :
tric constant is readily derived from
the electrostrictive equation (3) which of macroscopic size. If the symmetry 47rPt0,
we have already mentioned, as follows: of the substance under consideration is P, ot = naF = no (E -) )
as high as cubic (this includes iso- 3
tropy, as in liquids) the field due to or, (7)
'Matthias, Phys. Rev. 76. 175, 430, 1886.
(19491 all the molecules inside this sphere can P no
Shirane et. al., Phys. Rev. 80, 485. (1950) be shown rigorously to cancel out, if
Smolenskii, J. Tech. Phys. USSR 20, 137. the polarization inside the sphere is
(1950)
" P . Weiss, Phys. Zeits. 9, 358 (1908). uniform. The problem of finding the
24
The susceptibility x ' s connected with The explanation of the factor N, how- tion, i.e. in substances in which there
the dielectric constant by ever, had to wait 20 years for Heisen- are no permanent dipoles, but only
berg, who in 1928 showed that a charges elastically bound to equi-
*=l+4,rX (8) roughly similar force resulted from librium positions. This demonstrates
Now it is clear that without the "boot- the quantum-mechanical exchange ef- to us the second i m p o r t a n t dif-
straps" term we get the "naive" or fect.7 ference between ferromagnetism and
Drude formula for susceptibility which ferroelectricity: ferromagnetism always
is normally used in gases: In dielectric substances, on the other involves permanent magnetic dipoles
hand, there is no such limitation on
(9) (the field factor N is not of magneto-
na, as the existence of numerous sub-
Y = na stances with dielectric constants greater static origin, and thus is not affected
than 10 will testify. We see here the by the Onsager calculation) while fer-
4r first major point of difference between roelectricity probably cannot involve
Thus the term has increased the ferroelectricity and ferromagnetism; no permanent electric dipoles. It is pos-
3 such esoteric cooperative force as the sible that ferroelectricity can occur
susceptibility and the dielectric con- exchange force is necessary in the di- only in materials with large polariza-
ATT electric case, since the simple electro- bilities of the induced type.
stant. In fact, if we imagine that — no static "Lorentz force" seems always to Rochelle salt and KDP have long
3 be quite adequate. been thought to be dipolar-type ferro-
is allowed to increase toward the electrics; however, Pirenne10 has re-
On the other hand, in the dielectric cently shown that the isotope effect
value unity, we see that x a n d * in- case the shoe is almost too much on
crease without limit, and a little in the latter substance cannot be ex-
the other foot. Extrapolation from gas plained on the dipolar hypothesis, and
An polarizabilities for dipolar substances that probably KDP is at least an inter-
thought will convince one that if — no by the Curie law led, for many polar mediate, more complicated case than
3 liquids, to values of the polarizability either the harmonic (induced) or the
becomes greater than 1, the polari- which indicated that na should be con- dipolar.
zation will become infinite without siderably greater than the ferroelectric A model of BaTiOs has been pro-
the application of any external field value. That is why Van Vleck named posed which frequently is called the
at all. However, of course equation
"4TT "rattling titanium" hypothesis. It is
(6) breaks down when F and P be-
the problem the catastrophe": assumed that the titanium ion "rattles"
come too large, and in such a region
3 among six equivalent equilibrium po-
we will have merely a finite spon-
the theory seems to lead to the predic- sitions, one near to each of the oxygens
taneous polarization, just as is observed
in ferroelectricity and ferromagnetism. tion, for example, that water at 100°C
This is the phenomenon which has should be ferroelectric, and presumably
therefore frozen or at least of consider-
"Air
ably different physical character from
been called the catastrophe".
the water we know.
3 o" o
In the case of magnetism, the actual The theoreticians were rescued from
their theoretical catastrophe in the
fact is that all known substances at 1930's when Onsager showed that for
Air polar liquids, at least, the original Lor-
normal temperatures have —— n<*
3
< < 1; thus it was necessary for
entz field equation (5) is incorrect.
The reason for this is that while on
the average the polarization of the
-o ^&/
o"
Q^
o"~
6
Weiss to postulate boldly an addition- sphere is uniform, actually at any in- Fig. 9: Models for BaTiO :j .
al cooperative force term N : stant the permanent dipole of the cen-
4w tral molecule affects the distribution in the octahedron. A schematic diagram
F= H+ ( f-N) M in the sphere, leading to a situation in of this situation is given in the first
3 which the field due to the interior of half of Fig. 9- This model leads to
and N had to be taken of order of the sphere does' not cancel out when quite a high polarizability for the Ti
the order of taking averages is correction, which as we shall see later is an
magnitude 104. This completely un- ly handled. He concluded from his unnecessary frill; on the other hand,
substantiated hypothesis was remark- formula that the "catastrophe" could so far as Onsager field effects are con-
ably successful, when combined with not occur at all for permanent dipoles; cerned, the "rattling" ion is simply
the domain idea, in explaining ferro- this may not be quite correct, but it equivalent to a free rotating dipole,
magnetism in considerable detail, since is probable that the Curie temperatures which as we have seen is not likely
the precise behavior of a, and of the would be so low that other interactions to exhibit ferroelectricity. Most theore-
high field equation corresponding to than the electrostatic ones would al- ticians interested in ferroelectricity
(6), was being discovered by Lange- ready have taken over the situation, now feel rhat while the "rattling" model
vin and others at about the same time. and one should more properly speak is a convenient visualization aid, as
7
of phase or order-disorder transitions, used for instance in Von Hippel's ar-
W. Heisenbers, Zeits. fur Phys. 49, 619
(1928). such as occur in the hydrogen halides, ticle,1 the second diagram, represent-
s
The best discussion of these general con- than of Lorentz catastrophes.8 Van ing a pure "induced polarization" mod-
c^pts is g i v n by H. Frohlich. Theory of Di-
electrics (Oxford University Press, London,
Vleck9 has shown, however, that the el, is far more likely to be correct.
19491. Also see J. Pirenne, Helv, Phys. Acta, problem of the Onsager inner field
22, 479 (1949). does not arise at all in the case of solids Another aspect of ferroelectricity in
»J. H. Van Vleck. J. Chem. Phys. 5. 320 the BaTiOa class which shows the su-
(1937). with "harmonic" or induced polariza-
10
J. Pirenne, Physica 15, 1019 (1949). periority of the "induced" model is
25
demonstrated by a quantitative con- However, the measured value of the
sideration of the Curie-Weiss law numerator of (10) is larger than 10,-
above Tc. As we saw in equation (2) 000. This means that the dipolar type "N
and Fig. 4, the susceptibility above the of model is in error numerically by
a factor of more than two orders of \
Curie point is given approximately by
K magnitude.11 V
X= " (10) On the other hand, the induced
T-T c polarization model is capable of giving
Air as large a value of K in (10) as we
Now above the Curie point like. The reason for this is that the
temperature coefficient of induced J CYCLES PER SECOND
is less than one, so that equation (7) polarizability is generally quite small,
holds: of the order of 10" 5 —10" 6 per de- Fig. 10: Dispersion and loss in
gree (the same order as coefficients of BaTi0 3 as a function of frequency.
(7) volume expansion); actually, this co-
efficient is due to the combination of to find an induced type of polarizabil-
1— aa the effects of change of n with temper- ity, coming from ionic motions, which
3
ature with those of a slight change in is large enough so that added to (16)
In all of the permanent dipole-like shape of the potential troughs in which it gives 1:
cases (square well, rattling titanium, the ions find themselves as the lattice
or true rotating dipole) one and only (17)
changes size and the thermal agitation
one temperature dependence of the changes. Both of these effects are of ^ n ( a 0 l , t i c a l ""i~ "ionic motion) > 1
polarizability is to be expected, that the order of magnitude we have men- i —
given by the familiar dipolar "Curie tioned. Then if we postulate a small Such polarizabilities, proceeding en-
lav": negative temperature coefficient 8 for tirely from the harmonic binding of
C na, we get ions (we think particularly of the mo-
n« = — (11) no — (no) ( 1 — 8 (T—T c )) tion of the highly-charged titanium
T T=TC ion), are not at all surprising; LiF,
This can be inserted into equation (13) for instance, has a value as high as
(7), and it is found indeed that an tl.is. Actually, recent computations by
expression of the form (10) results, = ( 1 — 8 (T—T,..)) Slater11 have shown that even this
as follows: Air amount of clonic is not required, due
C to a peculiarity of the Lorentz field in
ihis, substituted into 7, gives the perovskite structure. (The original
3 (14) Lorentz calculation does not hold quite
T
Air 3 exactly for the O ions, but instead
a larger Lorentz field is to be ex-
Air C 4TT
x —1 _ ( i _ 8 ( T — T c ) ) = fZJf
4*-8
pected. )
3 T 3 Thus A number of rather fantastic ex-
but then (15) planations have been suggested to ex-
K = plain the "surprising" phenomenon of
4,r ferroelectricity. As we have seen, it
4^-8
is not necessary to go to any great
and this tells us that by a choice of 8 lengths to understand this phenomen-
which is quite consistent with normal on; what has always seemed to me a
magnitudes we may easily explain the little surprising is that there are not
value of K which is observed. Thus more known ferroelectrics, since a
we have a second very sound and con- combination of circumstances in which
(12) vincing argument that the polarization (17) is obeyed seems not to be a pri-
T —Tc in BaTi0 3 is of induced type. ori unusual. Perhaps, some have sug-
It is a simple matter to test this equa- We may now say a little, about what gested, many well-known substances
tion in the case of BaTiCV, T c is this kind of thinking gives'us in terms are ferroelectric, but do not have Curie
known to be about 400° Kelvin, which of a picture for BaTiO-j. In the first points at measurable temperatures and
gives us place, at optical frequencies this is a have such large coercive fields that the
3 fairly normal material, having the polarization is to all intents and pur-
slightly high index of refraction 2.4. poses completely frozen in and thus
K = T c ~ 100 unobservable.
Air The application of equation (7) to
n2 — €„pticai gives us One final point on the two models
" This discussion is similar to that given by
Slater, Phys. Key. 78. 748 (1950), but, as Air (16) of ferroelectricity, in relation to some
Slater points out. very similar conclusions naoptical ~ -60 recent experimental evidence, should
were reached by Devonshire, Phil. Mag. 40. be made. Many investigators have
1040 (1949), Ginsburg, J. Exp. Tn. Phys. 3
USSR 12, 239 (1948), and Richardson and
This part of the polarizability, is, as measured the frequency response of
myself (unpublished). The two latter were
more concerned with the size of the specific is well-known, due entirely to the mo- the dielectric constant of BaTi0 3 at
heat anomaly than with the Curie law r the room temperature, and all have found
two phenomena are, however, easily shown to tions of the electrons, since the ionic
be closely related by a simple thermodynamic
motions cannot follow optical frequen- more or less the behavior of Fig. 10.1
argument, and in fact the specific heat anom- At first glance it appears that here is
aly is several orders of magnitude too small cies. Thus it is only necessary for us
to be explained by a dipolar mechanism.
26
a typical relaxation spectrum, indicat- son12 in England have thrown con- checked by making measurements at
ing that some kind of potential bar- siderable doubt on this interpretation. high frequencies on single crystals. (2)
rier must be surmounted in order for These workers have measured the The second possibility is that the losses
the dielectric polarization to change; losses as a function of temperature, at have something to do with domain ef-
the polarizable entities thus cannot fol- frequencies near the relaxation fre- fects; for instance, they may be domain
low rapid motion of the field. This quency. Some of their results are wall motion relaxation such as is ob-
seems at first a very strong argument shown in Fig. 11. We see that the served in some ferromagnets. This may
for the "rattling titanium" model of losses drop off very sharply above the be wrong in BaTi0 3 , but is appearing
Fig. 9, since this model provides an Curie point; other work, on (Ba-Sr) more and more likely in some of the
immediate visualization for this po- TiC>3 mixtures, indicates that even the newer materials, in which something
tential barrier. In fact, a reasonable remaining losses seem to be illusory, very like small-signal hysteresis losses
size for the barrier between Ti posi- and thus that BaTiC>3 is essentially a seem to be present. Here also there are
tions gives a very good value for the lossless material above the Curie point. indications, however, that the losses are
relaxation frequency. However, recent not ordinary relaxation losses.
measurements by Powles and Jack- This behavior cannot be explained
on the relaxation idea; a relaxation fre- In this second, theoretical, part of
quency such as is given by the "rattling this paper I have tried to present the
Ti" model cannot disappear so abrupt- idea that BaTi0 3 and its sister ma-
ly simply because of the disappearance terials are simply victims of the long-
of spontaneous polarization. The in- predicted and rather well-understood
duced polarization model correctly pre- Lorentz "Air catastrophe." The atomic
dicts that there should be no large
losses at any frequency lower than the "T
fundamental infrared frequency, which mechanism leading to the ferroelectric
should appear at a few tenths of a phenomena in these crystals is thus ac-
millimeter wavelength. The observed cidental, in the sense that quantitative
\ losses below the Curie point have been prediction is rather hopeless, but cer-
t ascribed to two possible phenomena: tainly not surprising. The electronic
T in DEGREES C (1) Slater's suggestion (made in in- and ionic polarizability, added together,
Fig. 1 1 : High-frequency loss in formal discussion) is that these losses pass the "critical value" at the Curie
BaTiC>3 as a function of are acoustic radiation losses from the point. The results of this occurrence
temperature. small particles of the ceramics upon are the various experimentally observed
12
which all thse measurements have phenomena discussed in the first por-
Proc. Inst. Elec. Eng. part 3, p. 383
(1949). been made. This suggestion can be tion of the paper.
27
Use of Stochastic Methods in Line-Broadening Problems
Kyoto Lectures, March 1954
These lectures, given at the very young Yukawa Institute as a final thanks to Japan, for
my Fulbright visit in 1953-54, are a summary of my approach to line-broadening
problems, the field of my thesis and of much of my early work at Bell. During my stay
with Kubo in Tokyo he began to go on with these ideas to develop the correlation
function methods which became so important in condensed matter theory. (As did Lax
at Bell, but not with so much effect.) My paper with Weiss in Rev. Mod. Phys., in spite
of its appearance in a review journal, was not a review. Several reviews appeared on
my approach to pressure broadening but none were by me.
29
Exploring the Variety of Random
Documents with Different Content
began to fall at Zhibu on the 3rd of October, the fall would
commence at the Tépe, more than two hundred miles higher up
along the windings of the river, at least three days before, if we
take the current at three miles an hour. My statement, therefore,
that the river begins decidedly to fall at the confluence at the very
end of September, has been singularly confirmed. But that there
is also some truth with regard to the long continuance of the
highest level is evident from the conflicting observations of the
party. (See Baikie’s Journal, p. 217.) Indeed the sailor-master
insisted that the river had fallen long before; and all the people
were puzzled about it. From all this I must conclude that my
statement with regard to the river, instead of having been
considerably modified by the expedition, has been confirmed by
their experience in all its principal points. We shall see the same
difficulty recur with regard to a maximum level preserved for forty
days by the western river, although the time when it begins to fall
is entirely different; and as to the latter river, not only I, but the
natives also were mistaken with respect to its presumed time of
falling. The same is the case with the (river) Shári, and is natural
enough, considering the extensive inundations with which the rise
of these African rivers is attended. This state of the rivers in the
tropical climes is so irregular, that Leo Africanus has made quite
the same observation. (L. i. c. 28, “Descrizione dell’ Africa.”)
[65] I leave this passage as it stood in my journal, although it
describes a state of things which now, in 1857, belongs to the
past. This stronghold also has at length been taken by the
intruders, and the seat of happiness and independence converted
into a region of desolation. In 1853, two years after my journey
to Ádamáwa, Mohammed Láwl left his residence with a great
host, having sworn not to return before he had reduced Bágelé.
After a siege of almost two months, with the assistance of a few
muskets, he succeeded in conquering the mountaineers, and
reducing them to slavery. The chief of the pagans of the Bágelé,
who belong to the Bátta tribe, in the height of his power
exercised paramount authority over the neighbouring tribes, and
is said to have even had the “jus primæ noctis.”
[66] Ribágo, sometimes contracted to the form Ribáwo, means
“a governor’s country-seat.”
[67] With regard to the Fúlbe, the prayers of dhohor (“zúhura,”
or “sallifánna”) may rightly be called midday prayers, as they are
accustomed to pray as soon as the zawál has been observed. But
in general it would be wrong to call dhohor noon, as is very often
done; for none of the other Mohammedans in this part of the
world will say his dhohor prayer before two o’clock p.m. at the
very earliest, and generally not before three o’clock.
[68] Adamáwa is certainly not quite identical with Fúmbiná, as
it denotes only those regions of the latter which have been
conquered by the Fúlbe, while many parts are as yet unsubdued.
[69] With regard to salt, I will observe that the greater part of
it is brought from Búmánda, on the Bénuwé, near Hamárruwa,
where it seems to be obtained from the soil in the same way as I
shall describe the salt-boiling in Fóga, although in Búmánda there
is no valley-formation, and Mr. Vogel, who lately visited this place,
may be right in stating that the salt is merely obtained from ashes
by burning the grass which grows in that locality.
[70] It is a great pity that the members of the Bénuwé
expedition were not able to measure the elevation of the river at
the furthest point reached. My thermometer for measuring the
boiling-point of water was so deranged, that my observation at
the Tépe is without any value. Till further observations have been
made, I think it may be assumed to be from 800 to 850 feet.
[71] It would be rather more appropriate to give the name of
Lower Bénuwé to that part of the river below, and that of Upper
Bénuwé to the part above the confluence, than to call Upper
Bénuwé the part of the river visited by Dr. Baikie.
[72] This name is evidently connected with that of the
Balanites, which they call “tanní”; and several Negro nations
compare the date with the fruit of that tree.
[73] Mr. Vogel, who has succeeded in obtaining a sight of this
animal, found that it is a Mammal like the Manatus Senegalensis.
The South African rivers also have these Mammals, and the ayú is
not less frequent in the Ísu near Timbúktu than it is in the
Bénuwé.
[74] Súmmo, situated between Holma and Song.
[75] The numbers “three” (tan) and “four” (nan) seem to point
to the Fulfúlde as well as to the Kaffir languages.
[76] It is probable that this tribe is indicated by the ‫ مكبا‬of
Makrízi (Hamaker, Spec. Catal. p. 206), although there are several
other localities of the same name.
[77] Probably their real name is Tiká. See Appendix.
[78] The termination nchí is nothing but the Sónghay word ki,
which in several dialects is pronounced as chí, and means
“language.” On account of this termination being added to the
original name, I have purposely not marked the accents in this
list. The languages thus marked are spoken only partly in
Ádamáwa, the tribes to whom they are peculiar being for the
greatest part independent.
[79] In the following sketch, made just at the moment, I aimed
only at giving the outlines of the mount, without any pretension
to represent the country around. The foreground, therefore, is
left quite level.
[80] Perhaps this was a sign of mourning.
[81] The marriage (nigá) ceremonies in this country fill a whole
week. The first day is dedicated to the feasting on the favourite
“nákia,” the paste mentioned before; the second to the “tíggra,” a
dried paste made of millet, with an immense quantity of pepper;
the third to the “ngáji,” the common dish made of sorghum, with
a little fish sauce, if possible; the fourth day is called “líktere,” I
think from the taking away the emblems of the virginal state of
the bride, “larússa”; the fifth, the bride is placed on a mat or
búshi, from which she rises seven times, and kneels down as
often; this is called “búshiro,” or “búchiro genátsin”; the next day,
which must be a Friday, her female friends wash her head while
singing, and in the evening she is placed upon a horse and
brought to the house of the bridegroom, where the final act of
the nigrá is accomplished. The Kanúri are very peculiar in the
distinction of a marriage with a virgin, “féro,” or “féro kuyánga,”
or a widow, or “kámo záwar.”
[82] Between Yédi and the Tsád, the following places are
situated—Léga, a considerable town surrounded by a wall;
Díbbuwa, Jíggeri, Manawáze, Górdiná, and Mógolám.
[83] Mr. Vogel, who likewise visited this spot in 1854, found the
plain elevated 920 feet above the level of the sea, while the two
mounts attained the respective heights of 1,300 and 1,600 feet.
INDEX.
Ádamáwa, 401, 428

Afalésselez, 117
Ágades, 175, 203
Ahmed, 371
ʿAin Shershára, Ruins near, 31
ʿAin Zára, 39
Ájirí, 349
Alantíka, 451
Aláune, 343
Asben, 121, 139
Asïu, 127
Aúderas barbarity, 171
Bághzen, 167
Baháushe slave, 313
Bárakat, 109
Bel-Ghét, 269
Bélem, 445
Bénuwé, 451
Benzári, 323
Berbers, 103
Beshér, 353
Bóghel Valley, 173
Bokhári, 323
Bórnu, 333
Búndi, 331
Búwa, 351
Chad, or Tsád, Lake, 386

Chémia, 225
Cheréka, Mount, 165
Cinyps, 37
Damerghú, 241
Dan Íbra, 237
Démmo, 581
Díkowa, 549
Éderi, 67
Enshéd eʾ Sufét, 15
Fáro, 451
Fénorang Valley, 129
Fódet, 149
Fugábú, 527
Fúgo Mozári, 409
Fúmbiná, 469
Gámerghú District, 405
Gazáwa, 257
Gébi, 133
Gérki, 317
Ghaladíma, 331
Gharíya, 59
Gharíya eʾ Sherkíya, 61
Ghát, 101
Ghurián, 43
Gílmirám, 245
Gozenákko, 249
Gúmrek, Lake, 235
Hadánarang, 105
Háj Beshír, 373
Háj Hassan, 395
Hanshír, 22
Hatíta, 89
Háusa, 273
Ikadémmelrang, 125
Imghád, 107
Itísan and Kél-gerés, 157
Jebel Durmán, 47
Jebel Msíd, 25, 33
Jerma Kadím, 71
Kánem, 501, 541

Kanó, 285
Kasr Dawán, 33
Kasr Dóga, 31
Kasr Ghurián, 21
Kasr Kérker, 36
Kasr Teghrínna, 21
Kátsena, 277
Kél-gerés, 157
Kél-owí, 153
Kikla, 18
Komádugu, The Jungles of, 345
Kúka Mairuá, 315
Kúkawa, 353, 369, 379, 491
Kurúlu, 439
Kusáda, 283
Laháula, 425
Lake Chad, or Tsád, 386
Leptis Khoms, 37
Mábaní, 407
Maduwári, 391
Mʿallem Dalíli, 445
Mándará, 561
Mánga warriors, 327
Marghí, 410
Márte, 547
Máshena, 325
Mbutúdi, 433
Melágo, 449
Mejenín, 40
Meselláta, 35
Mizda, 45
Molghoy, 413
Múbi, 479
Múglebú, 479
Múrzuk, 75
Ngórnu, 387
Rabda, 19
Rálle, 95
Ribágo, 459
Salla-léja, 187
Sámmit, 237
Shʿabet eʾ talha, 65
Shitáti, 526
Shúwa, 557
Sókoto, 497
Soy, 367
Sulléri, 475
Taboníye, 57
Tagáma, 233
Taganáma, 325
Tarhóna, 29
Tasáwa, 78, 251
Tébu Merchants, 225
Tekút, Mount, 21
Teléshera, 221
Terguláwen, 231
Tíggeda, 169
Tintagh-odé, 145
Tin-téggana, 215
Tin-téllust, 151, 213
Titíwi, 395
Tripoli, 7, 8
Tsád, Lake, 386
Tunis, 1
Úba, 481
Ugréfe, 71
Ujé Kasúkulá, 409
Um eʾ Zerzán, 15
Unán Valley, 227
Wadáÿ, 497
Wádi, 339
Wady Aberjúsh, 85
Wady Eláwen, 87
Wady Haera, 41
Wady Kérdemín, 18
Wady Rán, 23
Wady Rummána, 23
Wady Shʿabet, 65
Wady Sháti, 66
Wady Tagíje, 53
Wándalá Mountains, 421
Wáza, 605
Welád Slimán, 518
Wúliya, 583
Yó, 505
Yóla, 461
*** END OF THE PROJECT GUTENBERG EBOOK TRAVELS AND
DISCOVERIES IN NORTH AND CENTRAL AFRICA ***
Updated editions will replace the previous one—the old editions

will be renamed.
Creating the works from print editions not protected by U.S.

copyright law means that no one owns a United States
copyright in these works, so the Foundation (and you!) can copy
and distribute it in the United States without permission and
without paying copyright royalties. Special rules, set forth in the
General Terms of Use part of this license, apply to copying and
distributing Project Gutenberg™ electronic works to protect the
PROJECT GUTENBERG™ concept and trademark. Project
Gutenberg is a registered trademark, and may not be used if
you charge for an eBook, except by following the terms of the
trademark license, including paying royalties for use of the
Project Gutenberg trademark. If you do not charge anything for
copies of this eBook, complying with the trademark license is
very easy. You may use this eBook for nearly any purpose such
as creation of derivative works, reports, performances and
research. Project Gutenberg eBooks may be modified and
printed and given away—you may do practically ANYTHING in
the United States with eBooks not protected by U.S. copyright
law. Redistribution is subject to the trademark license, especially
commercial redistribution.
START: FULL LICENSE

THE FULL PROJECT GUTENBERG LICENSE
PLEASE READ THIS BEFORE YOU DISTRIBUTE OR USE THIS WORK
To protect the Project Gutenberg™ mission of promoting the

free distribution of electronic works, by using or distributing this
work (or any other work associated in any way with the phrase
“Project Gutenberg”), you agree to comply with all the terms of
the Full Project Gutenberg™ License available with this file or
online at www.gutenberg.org/license.
Section 1. General Terms of Use and

Redistributing Project Gutenberg™
electronic works
1.A. By reading or using any part of this Project Gutenberg™
electronic work, you indicate that you have read, understand,
agree to and accept all the terms of this license and intellectual
property (trademark/copyright) agreement. If you do not agree
to abide by all the terms of this agreement, you must cease
using and return or destroy all copies of Project Gutenberg™
electronic works in your possession. If you paid a fee for
obtaining a copy of or access to a Project Gutenberg™
electronic work and you do not agree to be bound by the terms
of this agreement, you may obtain a refund from the person or
entity to whom you paid the fee as set forth in paragraph 1.E.8.
1.B. “Project Gutenberg” is a registered trademark. It may only

be used on or associated in any way with an electronic work by
people who agree to be bound by the terms of this agreement.
There are a few things that you can do with most Project
Gutenberg™ electronic works even without complying with the
full terms of this agreement. See paragraph 1.C below. There
are a lot of things you can do with Project Gutenberg™
electronic works if you follow the terms of this agreement and
help preserve free future access to Project Gutenberg™
electronic works. See paragraph 1.E below.
1.C. The Project Gutenberg Literary Archive Foundation (“the
Foundation” or PGLAF), owns a compilation copyright in the
collection of Project Gutenberg™ electronic works. Nearly all the
individual works in the collection are in the public domain in the
United States. If an individual work is unprotected by copyright
law in the United States and you are located in the United
States, we do not claim a right to prevent you from copying,
distributing, performing, displaying or creating derivative works
based on the work as long as all references to Project
Gutenberg are removed. Of course, we hope that you will
support the Project Gutenberg™ mission of promoting free
access to electronic works by freely sharing Project Gutenberg™
works in compliance with the terms of this agreement for
keeping the Project Gutenberg™ name associated with the
work. You can easily comply with the terms of this agreement
by keeping this work in the same format with its attached full
Project Gutenberg™ License when you share it without charge
with others.
1.D. The copyright laws of the place where you are located also
govern what you can do with this work. Copyright laws in most
countries are in a constant state of change. If you are outside
the United States, check the laws of your country in addition to
the terms of this agreement before downloading, copying,
displaying, performing, distributing or creating derivative works
based on this work or any other Project Gutenberg™ work. The
Foundation makes no representations concerning the copyright
status of any work in any country other than the United States.
1.E. Unless you have removed all references to Project

Gutenberg:
1.E.1. The following sentence, with active links to, or other

immediate access to, the full Project Gutenberg™ License must
appear prominently whenever any copy of a Project
Gutenberg™ work (any work on which the phrase “Project
Gutenberg” appears, or with which the phrase “Project
Gutenberg” is associated) is accessed, displayed, performed,
viewed, copied or distributed:
This eBook is for the use of anyone anywhere in the United

States and most other parts of the world at no cost and
with almost no restrictions whatsoever. You may copy it,
give it away or re-use it under the terms of the Project
Gutenberg License included with this eBook or online at
www.gutenberg.org. If you are not located in the United
States, you will have to check the laws of the country
where you are located before using this eBook.
1.E.2. If an individual Project Gutenberg™ electronic work is

derived from texts not protected by U.S. copyright law (does not
contain a notice indicating that it is posted with permission of
the copyright holder), the work can be copied and distributed to
anyone in the United States without paying any fees or charges.
If you are redistributing or providing access to a work with the
phrase “Project Gutenberg” associated with or appearing on the
work, you must comply either with the requirements of
paragraphs 1.E.1 through 1.E.7 or obtain permission for the use
of the work and the Project Gutenberg™ trademark as set forth
in paragraphs 1.E.8 or 1.E.9.
1.E.3. If an individual Project Gutenberg™ electronic work is

posted with the permission of the copyright holder, your use and
distribution must comply with both paragraphs 1.E.1 through
1.E.7 and any additional terms imposed by the copyright holder.
Additional terms will be linked to the Project Gutenberg™
License for all works posted with the permission of the copyright
holder found at the beginning of this work.
1.E.4. Do not unlink or detach or remove the full Project

Gutenberg™ License terms from this work, or any files
containing a part of this work or any other work associated with
Project Gutenberg™.
1.E.5. Do not copy, display, perform, distribute or redistribute

this electronic work, or any part of this electronic work, without
prominently displaying the sentence set forth in paragraph 1.E.1
with active links or immediate access to the full terms of the
Project Gutenberg™ License.
1.E.6. You may convert to and distribute this work in any binary,
compressed, marked up, nonproprietary or proprietary form,
including any word processing or hypertext form. However, if
you provide access to or distribute copies of a Project
Gutenberg™ work in a format other than “Plain Vanilla ASCII” or
other format used in the official version posted on the official
Project Gutenberg™ website (www.gutenberg.org), you must,
at no additional cost, fee or expense to the user, provide a copy,
a means of exporting a copy, or a means of obtaining a copy
upon request, of the work in its original “Plain Vanilla ASCII” or
other form. Any alternate format must include the full Project
Gutenberg™ License as specified in paragraph 1.E.1.
1.E.7. Do not charge a fee for access to, viewing, displaying,

performing, copying or distributing any Project Gutenberg™
works unless you comply with paragraph 1.E.8 or 1.E.9.
1.E.8. You may charge a reasonable fee for copies of or

providing access to or distributing Project Gutenberg™
electronic works provided that:
• You pay a royalty fee of 20% of the gross profits you

derive from the use of Project Gutenberg™ works
calculated using the method you already use to calculate
your applicable taxes. The fee is owed to the owner of the
Project Gutenberg™ trademark, but he has agreed to
donate royalties under this paragraph to the Project
Gutenberg Literary Archive Foundation. Royalty payments
must be paid within 60 days following each date on which
you prepare (or are legally required to prepare) your
periodic tax returns. Royalty payments should be clearly
marked as such and sent to the Project Gutenberg Literary
Archive Foundation at the address specified in Section 4,
“Information about donations to the Project Gutenberg
Literary Archive Foundation.”
• You provide a full refund of any money paid by a user who

notifies you in writing (or by e-mail) within 30 days of
receipt that s/he does not agree to the terms of the full
Project Gutenberg™ License. You must require such a user
to return or destroy all copies of the works possessed in a
physical medium and discontinue all use of and all access to
other copies of Project Gutenberg™ works.
• You provide, in accordance with paragraph 1.F.3, a full

refund of any money paid for a work or a replacement
copy, if a defect in the electronic work is discovered and
reported to you within 90 days of receipt of the work.
• You comply with all other terms of this agreement for free
distribution of Project Gutenberg™ works.
1.E.9. If you wish to charge a fee or distribute a Project

Gutenberg™ electronic work or group of works on different
terms than are set forth in this agreement, you must obtain
permission in writing from the Project Gutenberg Literary
Archive Foundation, the manager of the Project Gutenberg™
trademark. Contact the Foundation as set forth in Section 3
below.
1.F.
1.F.1. Project Gutenberg volunteers and employees expend

considerable effort to identify, do copyright research on,
transcribe and proofread works not protected by U.S. copyright
law in creating the Project Gutenberg™ collection. Despite these
efforts, Project Gutenberg™ electronic works, and the medium
on which they may be stored, may contain “Defects,” such as,
but not limited to, incomplete, inaccurate or corrupt data,
transcription errors, a copyright or other intellectual property
infringement, a defective or damaged disk or other medium, a
computer virus, or computer codes that damage or cannot be
read by your equipment.
1.F.2. LIMITED WARRANTY, DISCLAIMER OF DAMAGES - Except

for the “Right of Replacement or Refund” described in
paragraph 1.F.3, the Project Gutenberg Literary Archive
Foundation, the owner of the Project Gutenberg™ trademark,
and any other party distributing a Project Gutenberg™ electronic
work under this agreement, disclaim all liability to you for
damages, costs and expenses, including legal fees. YOU AGREE
THAT YOU HAVE NO REMEDIES FOR NEGLIGENCE, STRICT
LIABILITY, BREACH OF WARRANTY OR BREACH OF CONTRACT
EXCEPT THOSE PROVIDED IN PARAGRAPH 1.F.3. YOU AGREE
THAT THE FOUNDATION, THE TRADEMARK OWNER, AND ANY
DISTRIBUTOR UNDER THIS AGREEMENT WILL NOT BE LIABLE
TO YOU FOR ACTUAL, DIRECT, INDIRECT, CONSEQUENTIAL,
PUNITIVE OR INCIDENTAL DAMAGES EVEN IF YOU GIVE
NOTICE OF THE POSSIBILITY OF SUCH DAMAGE.
1.F.3. LIMITED RIGHT OF REPLACEMENT OR REFUND - If you

discover a defect in this electronic work within 90 days of
receiving it, you can receive a refund of the money (if any) you
paid for it by sending a written explanation to the person you
received the work from. If you received the work on a physical
medium, you must return the medium with your written
explanation. The person or entity that provided you with the
defective work may elect to provide a replacement copy in lieu
of a refund. If you received the work electronically, the person
or entity providing it to you may choose to give you a second
opportunity to receive the work electronically in lieu of a refund.
If the second copy is also defective, you may demand a refund
in writing without further opportunities to fix the problem.
1.F.4. Except for the limited right of replacement or refund set

forth in paragraph 1.F.3, this work is provided to you ‘AS-IS’,
WITH NO OTHER WARRANTIES OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO WARRANTIES OF
MERCHANTABILITY OR FITNESS FOR ANY PURPOSE.
1.F.5. Some states do not allow disclaimers of certain implied

warranties or the exclusion or limitation of certain types of
damages. If any disclaimer or limitation set forth in this
agreement violates the law of the state applicable to this
agreement, the agreement shall be interpreted to make the
maximum disclaimer or limitation permitted by the applicable
state law. The invalidity or unenforceability of any provision of
this agreement shall not void the remaining provisions.
1.F.6. INDEMNITY - You agree to indemnify and hold the

Foundation, the trademark owner, any agent or employee of the
Foundation, anyone providing copies of Project Gutenberg™
electronic works in accordance with this agreement, and any
volunteers associated with the production, promotion and
distribution of Project Gutenberg™ electronic works, harmless
from all liability, costs and expenses, including legal fees, that
arise directly or indirectly from any of the following which you
do or cause to occur: (a) distribution of this or any Project
Gutenberg™ work, (b) alteration, modification, or additions or
deletions to any Project Gutenberg™ work, and (c) any Defect
you cause.
Section 2. Information about the Mission

of Project Gutenberg™
Project Gutenberg™ is synonymous with the free distribution of
electronic works in formats readable by the widest variety of
computers including obsolete, old, middle-aged and new
computers. It exists because of the efforts of hundreds of
volunteers and donations from people in all walks of life.
Volunteers and financial support to provide volunteers with the

assistance they need are critical to reaching Project
Gutenberg™’s goals and ensuring that the Project Gutenberg™
collection will remain freely available for generations to come. In
2001, the Project Gutenberg Literary Archive Foundation was
created to provide a secure and permanent future for Project
Gutenberg™ and future generations. To learn more about the
Project Gutenberg Literary Archive Foundation and how your
efforts and donations can help, see Sections 3 and 4 and the
Foundation information page at www.gutenberg.org.
Section 3. Information about the Project

Gutenberg Literary Archive Foundation
The Project Gutenberg Literary Archive Foundation is a non-
profit 501(c)(3) educational corporation organized under the
laws of the state of Mississippi and granted tax exempt status
by the Internal Revenue Service. The Foundation’s EIN or
federal tax identification number is 64-6221541. Contributions
to the Project Gutenberg Literary Archive Foundation are tax
deductible to the full extent permitted by U.S. federal laws and
your state’s laws.
The Foundation’s business office is located at 809 North 1500

West, Salt Lake City, UT 84116, (801) 596-1887. Email contact
links and up to date contact information can be found at the
Foundation’s website and official page at
www.gutenberg.org/contact
Section 4. Information about Donations to
the Project Gutenberg Literary Archive
Foundation
Project Gutenberg™ depends upon and cannot survive without
widespread public support and donations to carry out its mission
of increasing the number of public domain and licensed works
that can be freely distributed in machine-readable form
accessible by the widest array of equipment including outdated
equipment. Many small donations ($1 to $5,000) are particularly
important to maintaining tax exempt status with the IRS.
The Foundation is committed to complying with the laws

regulating charities and charitable donations in all 50 states of
the United States. Compliance requirements are not uniform
and it takes a considerable effort, much paperwork and many
fees to meet and keep up with these requirements. We do not
solicit donations in locations where we have not received written
confirmation of compliance. To SEND DONATIONS or determine
the status of compliance for any particular state visit
www.gutenberg.org/donate.
While we cannot and do not solicit contributions from states

where we have not met the solicitation requirements, we know
of no prohibition against accepting unsolicited donations from
donors in such states who approach us with offers to donate.
International donations are gratefully accepted, but we cannot

make any statements concerning tax treatment of donations
received from outside the United States. U.S. laws alone swamp
our small staff.
Please check the Project Gutenberg web pages for current

donation methods and addresses. Donations are accepted in a
number of other ways including checks, online payments and
credit card donations. To donate, please visit:
www.gutenberg.org/donate.
Section 5. General Information About

Project Gutenberg™ electronic works
Professor Michael S. Hart was the originator of the Project
Gutenberg™ concept of a library of electronic works that could
be freely shared with anyone. For forty years, he produced and
distributed Project Gutenberg™ eBooks with only a loose
network of volunteer support.
Project Gutenberg™ eBooks are often created from several

printed editions, all of which are confirmed as not protected by
copyright in the U.S. unless a copyright notice is included. Thus,
we do not necessarily keep eBooks in compliance with any
particular paper edition.
Most people start at our website which has the main PG search
facility: www.gutenberg.org.
This website includes information about Project Gutenberg™,

including how to make donations to the Project Gutenberg
Literary Archive Foundation, how to help produce our new
eBooks, and how to subscribe to our email newsletter to hear
about new eBooks.

Instant Download A Career In Theoretical Physics 2nd Edition P. W. Anderson PDF All Chapters

Uploaded by

Copyright:

Available Formats

Instant Download A Career In Theoretical Physics 2nd Edition P. W. Anderson PDF All Chapters

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Instant Download A Career In Theoretical Physics 2nd Edition P. W. Anderson PDF All Chapters

Uploaded by

Copyright:

Available Formats

Download the full version of the ebook at ebookfinal.

A Career In Theoretical Physics 2nd Edition P. W.

Download more ebook instantly today at https://ebookfinal.com

Theoretical Nuclear And Subnuclear Physics 2nd Edition

Physics and Theoretical Computer Science From Numbers and

The Ghost in the Atom A Discussion of the Mysteries of

Condensed Matter Physics 2nd Edition Michael P. Marder

Musculoskeletal Imaging Cases Cases in Radiology Mark W.

Equations in Mathematical Physics A practical course 1st

Differentiable Manifolds A Theoretical Physics Approach

Strings Branes and Gravity 1st Edition Theoretical

Vol. 21 Spectroscopy with Coherent Radiation — Selected Papers of Norman F. Ramsey

For information on Vols. 1-20, please visit http://www.worldscibooks.com/series/wsscp_series.shtml

YJ? World Scientific

American Association for the Advancement of Science (Science)

Library of Congress Cataloguing-in-Publication Data

All rights reserved.

Oxford, December 2, 1993

Preface to the Second Edition v

Introductory Essay to the First Edition vii

1. Pressure Broadening in the Microwave and Infrared Regions 5

2. Theory of Ferroelectric Behavior of Barium Titanate 21

3. Use of Stochastic Methods in Line-Broadening Problems 29

4. An Approximate Quantum Theory of the Antiferromagnetic Ground State 51

5. Qualitative Considerations on the Statistics of the Phase Transition in

6. Ordering and Antiferromagnetism in Ferrites 71

7. Absence of Diffusion in Certain Random Lattices 79

8. New Method in the Theory of Superconductivity 95

10. Theory of Dirty Superconductors 113

11. Calculation of the Superconducting State Parameters with Retarded

12. Generalized Bardeen-Cooper-Schrieffer States and the Proposed

13. Localized Magnetic States in Metals 155

14. Plasmons, Gauge Invariance, and Mass 169

16. Probable Observation of the Josephson Superconducting

17. Image of the Phonon Spectrum in the Tunneling Characteristic between

18. Exchange in Magnetic Insulators 189

20. Coherent Matter Field Phenomena in Superfluids 219

21. Considerations on the Flow of Superfluid Helium 241

22. Multiple-Scattering Theory and Resonances in Transition Metals 255

24. The Kondo Effect I 297

The Kondo Effect II

Kondo Effect III: The Wilderness — Mainly Theoretical 312

Kondo Effect IV: Out of the Wilderness 318

25. Superconductivity in the Past and the Future 325

26. Macroscopic Coherence and Superfluidity 343

28. Space-Time and Scaling Techniques in the Kondo Problem 361

29. Anomalous Low-temperature Thermal Properties of Glasses and

30. Comments on the Maximum Superconducting Transition Temperature 377

Comment on "Model for an Exciton Mechanism of Superconductivity" 389

31. Resonating Valence Bonds: A New Kind of Insulator? 393

32. Anisotropic Superfluidity in 3 He: A Possible Interpretation of

33. Conference Summary 409

34. Asymptotically Exact Methods in the Kondo Problem 417

36. Conductivity from Charge or Spin Density Waves 469

37. Uses of Solid State Analogies in Elementary Particle Theory 477

38. Possible Consequences of Negative U Centers in Amorphous Materials 503

39. Theory of Spin Glasses 509

40. Solution of "Solvable Model of a Spin Glass" 521

42. Scaling Theory of Localization: Absence of Quantum Diffusion in

43. Some General Thoughts about Broken Symmetry 543