Notice: We are in the process of migrating Oral History Interview metadata to this new version of our website.
During this migration, the following fields associated with interviews may be incomplete: Institutions, Additional Persons, and Subjects. Our Browse Subjects feature is also affected by this migration.
We encourage researchers to utilize the full-text search on this page to navigate our oral histories or to use our catalog to locate oral history interviews by keyword.
Please contact [email protected] with any feedback.
This transcript may not be quoted, reproduced or redistributed in whole or in part by any means except with the written permission of the American Institute of Physics.
This transcript is based on a tape-recorded interview deposited at the Center for History of Physics of the American Institute of Physics. The AIP's interviews have generally been transcribed from tape, edited by the interviewer for clarity, and then further edited by the interviewee. If this interview is important to you, you should consult earlier versions of the transcript or listen to the original tape. For many interviews, the AIP retains substantial files with further information about the interviewee and the interview itself. Please contact us for information about accessing these materials.
Please bear in mind that: 1) This material is a transcript of the spoken word rather than a literary product; 2) An interview must be read with the awareness that different people's memories about an event will often differ, and that memories can change with time for many reasons including subsequent experiences, interactions with others, and one's feelings about an event. Disclaimer: This transcript was scanned from a typescript, introducing occasional spelling errors. The original typescript is available.
In footnotes or endnotes please cite AIP interviews like this:
Interview of Philip W. Anderson by Lillian Hoddeson on 1988 May 10,
Niels Bohr Library & Archives, American Institute of Physics,
College Park, MD USA,
For multiple citations, "AIP" is the preferred abbreviation for the location.
Topics discussed include: family background, education at Harvard, Japan, interactions with Japanese physicists, Bell Labs, solid state physics, Mott phenomenon, Green's function, and Helium-3.
This is Lillian Hoddeson talking to Dr. Philip Anderson about his career in solid state physics. I’d like to put a little bit of early background on this tape. You were born December 13, 1923, in Indianapolis.
Would you tell a little bit about your early childhood, and in particular something about how you were first exposed to scientific things, or first became interested in science and math, anything like that that comes to mind.
Well, my family was kind of a scientific family. My father was a professor himself. He had been the son of an itinerant fire-breathing preacher, who was so fire-breathing he couldn’t find a congregation, and so he became a farmer, had four children all of whom became academics. Both my father and his brother became professors of plant pathology, because they were influenced strongly by a very good professor of botany at Wabash College, which was the local college in Crawfordsville, Indiana, where almost all my forebears come from. And both of them went on to graduate school, in biology, and became plant pathologists. Both of them ran experiment stations in two different states, one in Connecticut, one for University of Illinois. He married my mother, who was a Wabash girl, the daughter of a professor of mathematics, as a matter of fact, at Wabash College, and her brother was also a professor at Wabash College. A very academic family. Her brother had been quite a brilliant student and had even won a Rhodes Scholarship early in his career. He had had rheumatic fever and had a heart weakness, and for that and other reasons, which we never discussed in the family, returned from what had been a very eminent career to Wabash College to teach. So right from the start, there was an interest in science and intellectual things in the family. My mother was a very intelligent woman herself. It was not that usual in those days to have gone to college, especially since the science professors at Wabash College didn’t necessarily run to sending daughters of the family to college. She did a year at Northwestern and finished her degree at University of Illinois. Was eminent in such things as the Association of University Women, League of Women Voters and things like that. She was quite a powerful character, very very much a social person at the University of Illinois. She was a member of very interesting groups of people. In particular, there was an organization known as the Saturday Hikers, which you've probably heard of. You've never heard of it?
Oh, it was a very —
— where did they hike?
It was around the University of Illinois for some 30 or 40 years. It may even still exist. In the late fifties, it really meant University of Illinois because the provost and the president and everyone else were members, such people of eminence as Hod Gray in the department of Law and Clarence Berdahl in the department of political science, and Bill Oldfather was kind of the leader of the group. He was in the department of classics, one of Illinois' most eminent people. And Wheeler Loomis was always a member. My father was kind of almost a charter member, but there was also a kind of ladies auxiliary that my mother was very busy in, and they would very often go out on Sundays with the same men who’d gone out on Saturdays, and you may ask, where did they hike? But in fact, you could go 20 or 30 miles from Urbana in almost any direction, to the Sangamon, even a few less miles than that, and be in quite nice woods, as close as Homer, something like that. You can do a lot of picnicking around on the Sangamon. They’d find a pasture and set up a softball diamond or go swimming, skinny dipping in the Salt Fork, or in the strip-mine pools, what is now Kickapoo State Park. One time, these people bought a large fraction of what is now Kickapoo State Park and eventually gave it to the state as part of the state park. We called it the Pollywogs. So in this group there were a lot of scientists, of course, in particular Worth Rodebush was a member. He was a physical chemist. Wheeler, Gerald Almy was a member. Sid Dancoff was a hiker. Bartlett and so — Not a very attractive character but somebody that we all kind of tolerated, who was difficult to get along with, but we tolerated him. A lot of quite important people in the history of science at Illinois were part of it. That’s just an example of the kinds of things my mother did. She was quite a gal. Not easy to get along with necessarily, but very much, as people like that are, really a social leader, very much involved in society. So it was kind of inevitable that if I was going to be a scientist, I would be interested. My father encouraged me with a chemistry set in the basement.
We’re now talking about roughly what age?
Oh, I started having a chemistry set at the age of 9 or 10.
Still in elementary school.
And this was in Urbana?
When did you move there?
Oh, I moved to Urbana at the age of five days. My father — see, my mother had had a hard time with my sister, four years before, when my father had just been beginning as a tenured professor at Illinois. I guess they built their house in 1919. And she associated the hard time with the hospital in Illinois, that strange state, so she insisted on going to the hospital in her own state, Indiana. But then as soon as things were all right, she transported me across the state line.
Oh, I see. You were effectively born in Urbana.
Effectively born in Urbana. Yes, he was a professor at Illinois, by this time. My mother was three years younger. He must have been 35. It was a time when people waited until they got tenure to build a house before they had their families. I think it took him eight years to get his degree, during which time he was teaching continuously and maybe even had the title of professor, and he got his degree and became a real professor. Very different kind of society in science in those days. One tends to forget that scientists mostly came from small Midwestern colleges, instead of mostly coming from the Chinese ghettoes or from the Jewish bourgeois of New York City and San Francisco. In those days they mostly came from small colleges like Wabash. I think it’s been actually documented that there was an enormous number of them from that kind of background. Anyhow, it was almost inevitable.
And I understand you went to Uni High.
Yes. Let’s see, there’s one thing — in grade school, my mother was determined (my birthday was December) that I wouldn’t have to wait till I was nearly seven to get in grade school, so at the age of 5-3/4 she sent me to a private school called the Forestry School, where I didn’t learn to write properly, and it still shows, but otherwise it was a nice school, and at age seven I went into the public school and was notoriously a bad boy for a year or two.
Just the usual gifted child syndrome?
No, it was the gifted child syndrome with maybe a little bit extra. I stayed home a great deal; either faked or had respiratory infections and things like that. And my mother was very sympathetic. The story is told; that Mother was once heard to say that it wasn’t until I got to Uni High that she was quite sure whether or not I was bright or just a quiet child. Certainly the school management thought I was not bright, was just a problem and a trouble maker, I think right up to the sixth grade. And I didn’t pass the exams to get into Uni High until I was a “sub-freshman.” Then after that I never had any more problems in school. I guess one other thing that was important to me was the one family sabbatical in which we went to Europe, and Uni High was very good about that. They let me continue in the freshman year of high school even though I’d already been a sub-freshman; and do my studies as I traveled and that worked out perfectly well. I never felt any —
Did you have any problems with it?
No. Again later on there were discipline problems. I guess I was notorious for disrupting the ancient history class when it got boring, and there was one Latin class —
The problems had to do with you getting bored.
Yes, mostly. Well, I was not that good at English. Nobody taught us how to write. That was always a problem. And I can’t memorize. When it was a matter of memorizing long poems, I kind of bombed out. The poem I chose to memorize was Walt Whitman’s — “When Lilacs Last,” so I was interested — I was always very interested in literature. I read Proust during this period. I read a lot of things of Jules Romains. This enormous series on the World War. I loved Thomas Mann, was a great fan of Thomas Mann. So in a way really I was more interested in literature than in science, although I loved math and did very well in it. Didn’t like physics. We had a kind of a tinker of a physics professor who was interested in gadgets and not in quantitative things, and I was quantitative. So actually I went to Harvard thinking I was going to be a mathematician. And fortunately, I don’t know whether it was Wheeler Loomis or Gerald Almy said to my mother, and I did what my mother said — you know, I was very rebellious and very difficult but I always did what they said. Fortunately, Gerald Almy said, “You can’t replace math and you can’t replace physics, if you feel you want to do it, so you have to take the first year of physics.” I had the first year physics course under Furry and Furry said, “Half of you will be gone by the end of the year,” and it was a challenge, so I stayed the course and actually had a rather serious problem at the beginning, didn’t — I had a terrible problem getting into a feeling for vectors and statics. And then suddenly it fell into place, and I just never had any problem with physics after that, and I relaxed and worked like mad on the massive history course, European history course that they give you, which I also loved. “Frisky” Merriman gave it and I fought like mad to get into the discussion sessions which the good students got into, things like that. And I managed by the skin of my teeth, because I could do maps and 15 percent of every exam was a map.
I’ve heard that Furry was a wonderful undergraduate instructor.
Yes, he really was.
Yes, much better than Schwinger ever was, I’m sure.
So then what happened?
Well, there was a National Scholarship, of course, and the reason for the National Scholarship was Uni High. They’d had Jim Tobin and Pierre Noyes’ brother Richard in ‘36, I guess so they were conscious of the National Scholarships, and so all of a sudden, with only a year to go, Sanford (principal of Uni High) called my parents up and said, “Has Phil taken Latin?” Did you have to take two years of Latin?
Well, I hadn’t. So Bill Oldfather of the classics department spent three months that summer tutoring me in first year Latin. And then I got through second year Latin, and we ruined one substitute after another. The famous Latin teacher we had had died and we had Latin substitutes, and they had terrible discipline problems. Jim Tobin’s younger brother and I actually ended up in opposite corners in back, completely forbidden to move about the room. That was the only other problem I had at Uni High. It wasn’t a problem I had with Uni High; it was a problem Uni High had with me really.
So you learned two years of Latin at Uni High, one in the summer.
One in the summer, yes.
When you graduated, it was 1940. The war was just about to start. So that means that your undergraduate years were at Harvard during the period when many of the professors were gone.
I tried to get into ROTC. I wear glasses. I couldn’t. I tried to get into some kind of program for navigators. If you wore glasses, you couldn’t. So I just took the accelerated program.
My family, incidentally, was very very involved politically, in the American Committee to — it was one of the famous big American committees involved in supporting the Allies. They were very active in that. You know, we were, the faculty group, the Hikers group, pretty much a left wing group. We had all the usual — well, I don’t think it was like the corresponding group at Harvard would be, but there were certainly a number of token Communists. Certainly it was very far left, and very much concerned about the war. Everyone went through that, I guess, at my age. Not everyone. But we had much more of a feeling for it than most people, in the Midwest, perhaps not in the East. Certainly not more than Jewish families, except as I say we knew Jewish people.
Was Harvard very different for you?
When the war was going on —
— oh, no —
Can you comment on it?
No, I think the first year was — no, they just ignored the war. Yes. After Pearl Harbor, very rapidly things changed. After Pearl Harbor we were all encouraged to take something called engineering physics.
Was that a program in your first year?
No, it was in my second year. So in February I started on an engineering physics course or program which involved a lot of —
– this is February after Pearl Harbor?
Yes, which involved a lot of courses in microwaves, courses in amplifiers, courses in electronics. Nobody was telling us that we should have stayed right in nuclear physics. Fortunately, because I think I managed very much better than people who went into physics, who knew that and stayed in physics and went off to Los Alamos, in terms of postwar development. People like Henry Silsbee, for instance, went off to Los Alamos, and then were sent to the South Pacific for a year and a half, whereas I went to electronic physics, went to NRL, and people at NRL were back in graduate school in October, ‘45. But anyway, we accelerated. I had only two or three real physics courses. Had time to have electricity and magnetism from van Vleck who was kind of a — well, he was very good.
And he stay there throughout the war?
Well, he was director of the group at the Harvard branch of the Radiation Lab, what was it called? There was a Harvard equivalent. He did a lot of war work, especially on — you know, he was ignored when he said, “You’d better not use K band.” He did a lot of war work on that and on other things. He gave marvelous lectures in the magnetism course, marvelous because you had to figure it out for yourself, and we had a study group that involved people like Tom Kuhn and Forster, Henry Silsbee, and a man who disappeared later, Bob Houston, who was as bright as the rest of us. And most of these people got summas and later had distinguished careers one way or another. Then there was a mathematical physics course I took which was totally disorganized. Kind of a kook thing, Herbert Jehle, you may have run into —
He gave the course in a most peculiar way, kind of just throwing the whole of mathematical and theoretical physics in front of you, in the form of Morse and Feshbach (book) and chatted about this and that, and then he gave this open book exam, which really was very educational. The open book exam was graded on the basis of products rather than sums – products of some inverse power of the time taken, some positive small power of the number of problems attempted some other power of the number of problems gotten right, and so on. I guess there was one other factor. And difficulty, I guess, was another factor. Half the grades were A plus and the other half of the grades failed. And actually, you know, the department didn’t believe those grades, but actually all the grades of A plus all went to people who really were extremely good physicists in later life, and all the grades of Fail went to people who didn’t amount to much in later life. It was a remarkable predictor. Jehle himself became a nut on the subject, some kind of molecular theory, how proteins go together, which was the wrong molecular theory, the right idea but the wrong embodiment of it, and lost tenure, and everyone asked — it was later on, I guess, when his tenure came up, somebody came around and said privately, “Did you think he was a good teacher?” “Well, yes, but he was kind of crazy.” But that was the only really enjoyable physics experience. There was an awful professor named R.W.P. King in engineering who taught antennas. It really was ruinous, a most awful man. He had the intellect to understand the ber and bei functions, the Bessel functions of fractional order, and just about that, so he taught the whole course on the ber and bei functions (Bessel functions of fractional order) and on his own particular theory of how antennas worked. Actually it turned out, that was what I did for my war work, was antennas.
So you actually did do some.
Yes, I did.
We’ll get around to that.
Yes, I’m sorry, I’m being too slow.
No. So we’re now at about 1943?
And starting war work, with a BS?
With a BS. Went to NRL. Rapidly assigned to the antenna group in the department that was working on IFF. (Identification, friend or foe), and also on jamming counter-measures. I guess we did some good work, although I find it very hard to believe, knowing how much more competent Rad Lab and Bell Labs and places like that were. I think we did some good work in testing various equipment. We could tell when the equipment shook apart on our tables as well as anyone else could. And I designed a couple of antennas. I was totally incompetent in electronics.
Where were you based, all in one place?
Yes. At NRL. Anacostia. We worked very hard, six days a week. We had a little shack on the top of a five story building, where the temperature reached — you know, it was Washington, there was no air conditioning, and the shack was a wooden shack on top of a building, and it was a swamp, the Anacostia swamp. It was really something. I remember my first assignment was, go out and test antennas on the pier, on the Potomac River, on a July day. We could barely walk on the pier because all the tar on the pier had melted. But I guess I did eventually design an antenna that actually went into a piece of equipment, that didn’t appear until after the war. The only thing we did that was at all interesting was to design the antennas for the invasion fleet. There was a thing not very well known called a glide bomb that the Germans designed. It was very clever and worked remarkably well, and was used in the Mediterranean. They sank a couple of cruisers with it. It was the very first guided missile as far as I know. It was guided simply by radio signals from an airplane that flew and circled around up there, and it eventually went down the smokestack of a cruiser. It was very effective. And actually one of the members of our group of electronics specialists, a man from North Carolina named Tommy, had been on one of those cruisers and he said, yes, they were very effective. But apparently, for reasons probably having to do with Hitler saying no, no, I don’t think it’s ever got into the archives, they were not used against the invasion fleet. They may have been tried and they may have known that we had counter-measures for them. Anyhow we built jamming receivers; our group built the jamming antennas. They used, the wavelength band was very inconvenient, 10 meters to about 1 meter, so that you had to use the whole ship practically as the antenna, and put these giant funny things on the ships. I think the transmitters they made were so good that it probably didn’t matter how bad the antennas were. We actually installed them on a number of the old battleships. They came up the Potomac and we installed them. Unfortunately in the midst of installing them I came down with scarlet fever. So I had to leave. I participated for a few days. That was the only genuine thing I guess I did for the war effort, and whether it’s genuine or not, nobody knows. Anyhow they did not use glide bombs on the invasion fleet, and they didn’t sink anything. Or they may have just kept the planes far enough away. They may have tried.
At what point then did you go back to Cambridge?
October ‘45. It was very very quick. With ONR, again, ONR had good people in it. Manny Piore may already have been in ONR. And one of our people was the son of Admiral Strauss, and I wouldn’t be surprised if he hadn’t pulled strings. Somebody pulled strings and said, this one group of people who were drafted and worked in the war effort are going to be sent straight back to graduate school rather than to replace troops in Europe or the South Pacific, which most everyone else did. So I came right back, and Van arranged — I had seen Van at a meeting where he’d been talking about the millimeter wave stuff, the fact that K band wouldn’t work, and about wave properties, microwave propagation in general, as the war was winding down. They were very good, NRL, various senior people who came in, and we then began to start doing experiments on getting propagation through the exhausts of rockets and things like that, and were allowed to simply go off in a corner and read up physics. And Van, I saw at this meeting, and he said, “Do you want to come back to Harvard?” and I said, “Yes,” and he said, “Fine, and try to make it by the first semester,” although courses had started.
So then at that time people were coming back.
Who did you study with?
Well, we had a course sequence at that time. You had to pass a certain number of courses. And Schwinger, did Schwinger arrive the next year or that year? I’m not sure. He must have arrived either in the middle of the semester or the next year. But we had a quantum mechanics sequence, Furry again did that. We had an experimental physics sequence that was great fun that Oldenberg did. Used a lot of the original equipment, like Bainbridge’s original mass spectrometer and so on. I enjoyed doing a thesis for that on cosmic rays and auroras and things like that. It was all wrong but it was interesting to do. There was this awful course on analysis that D.V. Widder gave, that managed to get us through complex analysis without ever doing a contour integral. I was always in trouble on Green’s functions because I never understood contour integrals. He did it all with Mittag-Leffler series and things like that, and doubly periodic functions and it was awful. Schwinger came, and I basically took Schwinger simultaneously with the big course in quantum mechanics, simultaneously with statistical mechanics, just wrote notes, because he talked very fast, and studied them afterwards. I guess everyone had that experience.
What was Schwinger teaching?
Schwinger taught modern physics for three semesters, three successive semesters, essentially everything Schwinger had been doing in nuclear physics before and during the war, simultaneously with lecturing on electrodynamics for advanced people, which I didn’t, I couldn’t understand. I went to some of them and listened. Oh, Bridgman did thermodynamics. That was an awful course. You were required to take thermodynamics. Bridgman could put people to sleep in ten minutes flat. Everyone in the room was asleep. He had the most soporific voice. You should record this for insomniacs. It was unbelievable. He wasn’t interested in thermodynamics at all. He spent the entire course doing identities. I’ve always hated thermodynamics, actually, as a consequence of that, and did, well, not hate it, but felt negative about thermodynamics, whereas the next year, Furry did statistical mechanics and that was a marvelous course. He started out by explaining that there is no material in statistical mechanics. It’s all a bunch of examples about two equations, which is true. Then he said, “Well, since that’s the case, I’ll just do kinetic theory,” and he told us all about how he did isotope separation by thermal diffusion.
He taught you that?
He taught essentially thermal diffusion and kinetic theory, for about the second half of the course. But you know, we learned enough statistical mechanics. He taught us about steepest descents, which was very valuable, and free energies and all of that. Actually he could have told us more about free energies. The way I teach it now, I teach it equally fast, and I do it from free energies. He didn’t tell us enough about free energies, but he told us a lot about Woodwind models and things like that. It was very useful. There was this grueling Schwinger thing going on all the time. Every course was fun, except for Wilder. He was a test. Tom Kuhn and Ken Case and I would all do the homework independently if possible, except in one or two cases where I finally gave up and asked Ken what such and such a problem was about, and he would give me a hint. I think he’s a nice person. Ken Case got straight five pluses on all the homework. I got five plus minus two pluses. Tom got maybe five plus minus, about three pluses, and this was the competition. And otherwise, you know, everything was very easy. We would be playing bridge, drank when we could, sang Lehren songs, and did double crostics. Had a wonderful time. Till it came to prelims. Oh, that first semester, I took van Vleck on classical mechanics, which again was a marvelous course, and I got a very good grade in it. Then there was a course on elastic theory. Westergard from Illinois was supposed to be teaching, and Timoshenko was there, also a famous elasticist, but Westergard never arrived. He was apparently designing bombs. Timoshenko was a very poor teacher, and besides he was just substituting, and I got a B in that. I hated the course. That’s all irrelevant but anyhow I’ve had a negative thing about engineers ever since. That and Mr. R.W.P. King. There were two very good engineers in the undergraduate course, H.R. Menlo gave a marvelous course, an undergraduate course on electronics and how amplifiers work and so on. I’ve always appreciated that kind of hands-on engineer, as compared with the hands-on engineer who puts everything on the computer and plugs in his grand bei-functions, I really can’t abide. He doesn’t understand what he’s doing. Menlo was just a unique intuitionist, and there was some other old guy, they must have been giants in early electronics and tube theory, gave another course and I loved that, but that King course was a mess. Good lab. Very grueling lab. You had to write everything down, and you write on the right line.
OK, so where are we now? We’re up to your prelims and then —
Prelims I had a terrible time with, because Van assumed that since I’d passed mechanics I knew what Euler’s equations were, and I didn’t. I didn’t know spinning tops. So he just fought and fought trying to get spinning tops out of me, and I didn’t know spinning tops from a hole in the ground, and this was awful for Van, because he’d been brought up on action and angle variables. Everything else, I did just fine, so they asked me if I didn’t want to be an experimentalist, and I said, “Horrors, no, you should see what I did with the circuits at NRL.” And so they agreed and said, “OK, I guess you can be a theorist.” And Van gave me this problem.
Pressure broadening. Now, I wasn’t very serious, yet. I wasn’t very serious yet because — well, you know, I was really very young. You realize I still wasn’t — well, I guess I was 22, but I’d been off to the war, and got out of college at 19, and college had been accelerated, and I had a lot of fun to get out of my system, so the two years were mostly fun, and maybe that next year was a little bit of fun. And well — the other thing was, Schwinger a little bit frightened me, but also I just didn’t like that kind of atmosphere, and had a feeling that kind of physics was just, in the first place, very competitive. There we were fighting for five pluses. There were a lot of people who were at least as competent as I was at that kind of thing. And besides Schwinger had this long line of people outside his office. I liked Van. I knew Van. I felt I could cope with Van. I liked Furry, but Furry — I probably would have worked with Furry, but actually I think he even put me off a little bit, he really said, which is true, and that he was kind of in psychological, he didn’t say he was in psychological problems, but he said, “Are you quite sure?” — maybe I wasn’t quite sure. The other thing was that I really —
He was in trouble?
Well, he was unhappy, very very unhappy because he was competing against Schwinger, and Schwinger is a force of nature, and when you compete with forces of nature you are very likely to lose. I’m not sure he wasn’t a better physicist than Schwinger. That’s, I think, absolute heresy. But he was unable to do things as spectacularly as Schwinger did them. He certainly was a better teacher. But he very rapidly became not one, because I think he lost confidence in himself. Anyhow, the other thing was, I was interested — I’d been confronted with this idea of coherent spectroscopy in the war, in working with microwaves, and realizing from the conference on microwaves, that we were actually seeing lines, spectral lines in the atmosphere. Cleeton was even the boss of my section in NRL, Cleeton who discovered microwave spectroscopy. And Van — and you know, there was — marginally I knew some of the people who were working on NMR, and NMR was being discovered, and I went to the APS meeting where NMR was announced, and there was a marvelous summer course, or actually semester course from Gorter on relaxation which I liked very much. I kind of liked this whole business of the contact between classical coherence spectroscopy and electron... (off tape) Anyhow, I was interested in that kind of thing, and Van said, “Well, here’s a problem for you. There’s this man Lindholm in Sweden who claims he can do spectral line breadths in the infra-red, there’s a lot of literature, and there’s Henry Foley who did his thesis at Columbia on spectral line broadening, and there’s a lot of earlier work by Weisskopf and Kuhn and people. (crosstalk) But in fact — oh yes, this is the other Kuhn. He’s still around. One of these people who did a lot of very unexpected stuff. He discovered what is known as the Huckel theory, really the Kuhn theory perhaps, things like that. I don’t know why Kuhn isn’t more famous than he is. He also discovered all the business of origin of life independently from Manfred Eigen. Anyhow, Van’s point was that now that you have microwave spectroscopy, you can study line breadths in incredible detail. It used to be that to get the breadth of an optical line, you had to pump in incredible amounts of something to broaden it, and even then there was all kinds of difficulty with self-reversal and all that. Now you can just, a little bit of gas, you can tune your spectrometer through the line, and get the entire shaped line, beautiful. So, couldn’t these people do it? And I kept reading these things, and also I managed to get married and have a baby in this period, summer of ‘47. A year and a half after I got out of the Navy. And did a lot of commuting back and forth to New York, where Joyce was. Then she came up and worked for Harvard University, and I remember I’ve never been very favorable about the working conditions in the staffs of departments and universities since. They’re mean. They’re real mean. I don’t know whether you ever encountered it. They’re awful. I always have a lot of sympathy for secretaries and bookkeepers and things like that. And then she went home, and worked as a teacher until the baby became so obvious that she ran afoul of the nepotism laws, worked as a teacher as Miss Gothwaite and her very sophisticated students, towards the end of the semester, came to realize that she wasn’t a Miss because Susan was on the way. Anyhow, now it was a lot more serious, so about that time I began to think very hard about my problem, and began to have fun with it. There was one paper where Van asked me to go to the library and learn about interatomic forces, and another when he asked me to think about group theory, and then I went and learned about group theory. He’d given a group theory course which was very hard because he’s not the greatest explicator, but you could certainly teach yourself group theory yourself through Van’s classes.
Did he understand it, do you think, himself?
Oh, incredibly well. But by intuition. The groups, the Darstellung were just in his hat. He did things in ways which no one else could figure out, but he was a fantastic group theorist. Oh yes. He understood it all. Of course he did. He practically put group theory into quantum mechanics. But particularly discrete groups. Continuous groups and Lie theory and so on, well, it didn’t come into physics until later. But the Lorentz group, yes, to some extent. And I gradually began to realize the problem was hard, and different, and to understand what Lindholm had done, and then I began to do some things which I think were fairly important and fairly original. And I didn’t know how original and important they were, but they were. And I really had a hell of a good time.
This is before your PhD.
Before my PhD. Well, there were several things. There was a kind of a generalization of relaxation theory, a tensorial generalization of relaxation theory, and there was a generalization of what had been called the Fourier integral formulation of line broadening theory, something that Weisskopf had derived one way, and Lorentz I think had derived another way, and then there were some wiggles and tricks of my own. There were probably three major things that went into my thesis, and I don’t know to what extent — I knew it was good but I didn’t know to what extent it was good, and what was original. At this point, I don’t think I’m going to go into any of the papers as such. One thing was, Einstein — there’s Einstein who had done the beginnings of fluctuation dissipation theory. In a sense, he said you can get spontaneous emission from dissipation and vice versa. What I realized was, I could go the other way. I could take spontaneous emission, and get absorption from it, get dissipation. The spontaneous emission I could calculate by simply allowing the molecule to bump around through the molecules of the gas, and emit as it encountered all these collisions. And so I just took a Fourier analysis of what was happening to the dipole moment of the molecule. This is a derivation of this old Fourier integral representation, but it’s a derivation that starts with the microscopic and starts from something which might be generalized to the whole box of gas, the whole solid state sample. Weisskopf’s derivation is also very beautiful. His derivation was essentially a WKB derivation. He took this thing as a classical or semi-classical quantum mechanical orbit. And in a sense that’s more general because he’s saying this is a quantum mechanical theory which we’re doing, but I don’t think he realized the generality of what he’d done — and maybe I didn’t realize it too quickly. And I think that may have been one of the first statements that you could take a correlation function and make a spectrum out of it. This formula existed but it wasn’t seen that that was what it was, the Fourier transform of the correlation function of the dipole moment.
Well, it wasn’t that recent, ‘38 or ‘39. And something like it existed, and Debye’s derivation of the relaxation was something like it, but he was really working only at zero frequency. How does the thing relax? And I was saying, the whole damn spectrum comes out of this. It wasn’t the fluctuation dissipation theorem, because I hadn’t calculated my spectrum right, but it was basically a fluctuation dissipation theorem that I was presenting — it was kind of the first point at which anyone used the fluctuation dissipation theorem to get a spectrum. That’s one thing. One thing was this semi-classical, was much less spectacular, really, in a fundamental sense, but maybe more so in — oh, the point about this, the way I did it was to do it really with full indexes, which was the Schwinger way, to write it out in terms of traces of things, which was the Schwinger point of view, Schwinger training, to think, well, if you can do something like this, you can write it in fully invariant quantum mechanical representation. So I did that, which was essential for the spectral broadening problem, because of course these are complicated sets of levels, with rotation series of sequences and so on, vibration sequences, and you had to worry about what if you applied an electric field and spread the lines out, so it actually was a very complicated problem, what’s the, how does the pressure broadening change? You know, in the old days pressure broadening was just the breadth, and when you broadened out the spectrum, you didn’t worry about the fact that the different components had different breadths. Microwave region, you are worried about what happens if I change from the line as a whole; if you separate out the Zeeman components, which is easy as pie, what are the components? So I was doing a much harder problem, and I needed all those indices, and I needed the full generality. It was kind of using Schwinger – Van Vleck didn’t approve of it and didn’t like it very much, but there’s a lot of other stuff in the thesis, and a few papers — the Weisskopf formula is to some extent talking about this fluctuation dissipation theory. Anyhow, the other thing was this classical path method I used to actually do the collisions, which still exists in chemistry, and still exists in laser atomic physics. The point is to treat the Hamiltonian, as, take two molecules, pass them past each other, do the interior of the molecules fully quantum mechanically, but the paths of the two molecules are classical variables, so you have a time-dependent Hamiltonian, and you solve that kind of Hamiltonian problem and find the asymptotic values of the density matrix before and after. And I used some tricks in that which were just my own bells and whistles, which aren’t necessary, but this I believe is the first time that people did that. Ted Holstein picked it up for doing resonance transfer, and nowadays people do all kinds of excitation transfer problems with the classical path method. I just talked to a chemist four days ago who said, “Yes, I know about your thesis because of the classical path method. “The other thing was this — well, I enjoyed it — was this full tensorial equation of motion method for solving the problem, as we had it, of kind of getting the linear coefficient for the time evolution of the relaxation, which was a inverse relaxation time, and actually of course the relaxation rate is a tensorial quantity, so I did a lot of interesting rotation group stuff with that. That’s just what I enjoyed doing, a tricky and amusing little bit of mathematical fiddling, which maybe came in useful later, but I think — well, two things. One was really being able to do the problem by putting all these things together, and therefore really kind of going from the atomic forces all the way to the line breadth, which had never been done before, and doing it with reasonable rigor, simply as though you felt that you were actually going to solve the problem, and people weren’t doing many body problems that way at that time. People didn’t take, you know, real big boxes of gas and do things from first principles and get answers. So it was kind of a very early piece of many body physics. You know, you can find where the errors were, and where one had to make them. So, well, it lived a long time. It’s still being used. It kind of set me up.
Yes, and in a certain direction.
In a certain direction.
Which you followed for quite a long time.
Well, the next move is to Bell in 1949.
Yes. It was hard to get to Bell. Van Vleck had to fight to get me there.
Yes. I was going to go to Pullman State College in Washington.
Yes. I had practically accepted the job. In fact, I had accepted the job. And Van asked me where I was going, and he said, “No, no, no, where do you want to go?” And so he argued and I argued quite a bit with Bill Shockley, and he kind of said, “You must go to Bell.”
Van said, not —
Van said, “You must take him.”
Oh, I see, and Shockley didn’t want to?
Shockley was over his “nose count”.
Oh, I see.
And actually the only compromise he made with his management was to say, “You’ll have to fire a man in order to get this one,” which was kind of sad, because the man they fired was also a very good person, John Richardson. I didn’t know about this bargain. It got me off on the wrong foot at Bell, to some extent. Well, in the same month, Gregory Wannier, Bernd Matthias, Jack Galt and I arrived at Bell Labs, and they had only two slots. And Bernd got in specially because of ferro-electrics, and me they got in specially because Van said, “You must take him.” They tried to hire me temporarily, and I said no, oh, it was quite a time. Then Shockley was very good about it, luckily, you know, unusual for Shockley. Shockley’s kind of a Dr. Jekyll and Mr. Hyde. We were poor, and we drove down there in our new car with all our belongings in the back seat, and our daughter, and they put us up for two or three days until we found a place to live.
At their house.
Goodness. And how was it?
Well, he’s a funny man, you know, doing all this — it was very strange. Jean was marvelous at the time, the wife he later divorced and bad mouthed about how her genes weren’t satisfactory. But Bill was very strange. Funny. Cutting, the way he could be. Well, that story’s in the interview so I don’t need to go into it too much. You’re interested in the physics. Bill wanted me to do ferroelectrics.
He hired you to do ferroelectrics.
Yes. He said, “You do ferroelectrics.”
That was of interest to the phone system?
Well, (crosstalk) Matthias was coming with barium titanium, crystals of barium titanate, that only he knew how to grow. And yes, it was of interest because it was a possible memory, and they were already interested in memories. A possible switch, a possible nonlinear element. Also, they were very much interested in piezo-electricity of course. I think already Mason had patents, was getting patents on ferroelectric microphones, ceramic microphones, and the ceramic transducer. Barium titanate is still used as transducers. So of course the phone company was interested in it. And you know at that time one of the things we did was to read ASTOUNDING SCIENCE FICTION. Even some of us wrote for it. Chan Davis wrote for it.
ASTOUNDING SCIENCE FICTION. One of the things I left behind at Harvard was a collection of all the early ASTOUNDING SCIENCE FICTIONS, which I think would be worth a fortune if I still had it. That was the heyday, the great period of ASTOUNDING, great innovators in science fiction, people like Lester Del Ray, Asimov and so on. And John Pierce wrote for ASTOUNDING, you know, as J.J. Coupling. He was an executive director already, I believe, or a director at Bell Labs. And he wrote some science fiction. But he also wrote, he was the Asimov of his time, J.J. Coupling, he wrote science popularization. He wrote a very enthusiastic article about ferroelectrics for ASTOUNDING, as an energy storage device. You didn’t know that about him? It’s important — well, because of course his career has always been kind of science fiction based. Especially the satellite. He even invented it before A.C. Clarke. It was in his science fiction already, about the same time.
So who was there then?
Well, that’s the reason I went. I had been interviewed, had been very impressed, and had been very poor because I didn’t know anything about the problem, interviewed before I began to solve it. Then I was interviewed after, and that was all right. But I’d been interviewed by Bardeen. Herring hadn’t been there, the first time.
You would go there and then talk to various people?
Yes, you’d talk (crosstalk)
— and then they would register their opinion of you?
They’d register — this is the standard interview thing. You give a talk, and — it still goes on. It should have been frightening. I wasn’t frightened, eventually. But I suddenly became frightened the first time — I realized I didn’t know anything about the subject. I didn’t talk to Pierce, but I talked to Bardeen, Shockley, and who else was there?
Was Kittel there?
Kittel was there the second time. He was very impressive.
Bozorth was of course there, but I did not talk to him. Brattain was of course there. I may have talked to him. Pearson was there. I did talk to him.
Wooldridge would have been there but I didn’t. Harvey Fletcher was the director of physics, and of course one talked to him, but he was kind of a nonentity. Stan Morgan was not a nonentity, Addison White was not a nonentity, and I did talk to them.
Townes, yes, I talked to Townes, of course. In fact Townes was there, I talked to him both times. Townes was of course the major interaction, because of course that was his field, and he was very interested, that I agreed with his data, not with Bleaney, the second time around. That may have been what got me the job, that I said Bleaney’s data were wrong on two lines.
So the first time, you didn’t pass —
No, no, and quite right. They just said, “Come back later when you’ve worked on your problem.”
Oh, I see. They didn’t say no, they just said, come back later.
Come back later.
The second time, I didn’t pass at a lot of places. I didn’t pass at Brookhaven. I didn’t pass at GE. I had a marvelous interview at GE. Holloman and Fisher were very interested in having me, but I did pass at Westinghouse, because Ted Holstein wanted badly to have me, but then they put me in this new solid state group and it looked like a bummer and I didn’t want to go. The reason why Goudsmit later told me he didn’t want me was because when he asked me what did I want to do —
Goudsmit was there?
At Brookhaven. I knew him from Harvard — he gave a course on dynamics. That’s one of the reasons I didn’t know Bulan’s equations because his course on dynamics had been entirely a course on variation theory, and he didn’t do the equations. He did bugs on strings, and Brachistrone problems, things like that.
We were on Goudsmit.
Goudsmit, yes. Goudsmit gave this marvelous summer course, and then of course also talked out, we spent a night, practically, and a case of beer, listening to Goudsmit talk out his book Alsos. In the course, he was wonderful. I liked him. We got along just fine. But he asked me, “Why do you want to come to Brookhaven? Specifically what have you got in mind? What do you want to do on your problem?” I said, “I don’t have any problem. I solved it. What do you want me to work on?” He didn’t believe me. Of course he didn’t believe me. He didn’t believe that my problem was clearly over and I needed to learn some other things. For one thing, I had just audited the Van Vleck solid state course, and it seemed to me very diffuse and boring. I didn’t know much solid state physics. Had a lot to learn. And so I didn’t really have a broad enough attitude toward physics to know what I wanted to do next, and I really did want to go someplace where there were lots of problems and lots of interesting things going on. And yes, there were things I needed to do to complete my thesis. There were things I did do. And I wrote several papers. There was pressure broadening, foreign gas broadening, that I wrote a paper on. There was plenty more I could do in broadening, but I wanted to learn about other problems. Basically I had done so much. I really felt I wanted to learn other problems. So I was happy enough, when he said “Work on ferroelectrics,” I was happy enough to learn about ferroelectrics. And I spent about a year doing some things that — well, no, a few months doing some things that were more or less along Bill’s lines, and Bill taught me a hell of a lot of solid state physics in a very rapid hurry.
He was a good teacher?
He was fantastic. Very good. Well, he had a very good and simple way of looking at things. Never had any trouble with space lattices since talking with Bill about it. He could put a space lattice on the board like nothing. In fact, you know, he had a kind of intuition for lattice structures. He was very quick about it. At that time, he had this intuition about bands, or very soon thereafter, this intuition about tubes in band theory, which was very good. That’s the kind of thing that he could do. There’s a Shockley-Bardeen paper about deformation potentials which was marvelous. But he taught me how to do Coulomb lattice sums, what is that called, the Ewald method. It’s a beautiful way of expressing the Ewald method, he had beautiful ways of doing all kinds of things. And I talked to a number of other people. Betty Wood was a good crystallographer, was willing to tell you how her work on crystallography went together, and Alan Holden the same. I was put in a room with Gregory Wannier and learned all about Bands and Wannier, was interested of course in Statmech and Onsager and was trying to figure out how Onsager’s solution worked, so I learned a lot about statistical mechanics. Conyers explained to me about free energies. I was busy learning a lot of solid state physics very very fast. And Bill told me about local fields. Then I began to dig into local field literature, and then I found what Van had done about local fields, and what Onsager had done about local fields, and I was fascinated by local fields. And that’s what the ferroelectric stuff was about, of course. Bill was saying, “Well, it’s just the old Lorenz catastrophe happening. It’s not anything fancy like dipole moments and hydrogen bonds and so on, it’s just the dielectric constants, the polarizability reaches 3 over 4 pi and there you are.” And so we had the Lorenz catastrophe theory, and what I had to do was to figure out, just do the calculations for it. He said, “Just try to figure out why and how the effected factor charges are and why they are what they are.” John Richardson had started out on this program, and stopped because he had felt that he didn’t really understand the problem, from a fundamental point of view, and I started out on it and stopped for the same reason. And meanwhile, John, Gregory and I got interested in the general problem of phase transitions, we wrote a little note about how basically the Tisza and Landau theories weren’t really solving phase transitions, because Onsager showed that phase transitions weren’t simple mean field phase transitions, which may have been the first time anyone ever said it in the literature. That was Gregory’s thing, not mine. I just sat there and listened. And Richardson and I asked, how can we really show — second thing, how can we really show that ferroelectric theory, is just the Lorenz catastrophe and is not the usual kind of thermo-dynamic phase transition. We were trying to make some kind of connection between them. So really Richardson was doing just as much interesting stuff as I was, but he was fired and I wasn’t because he hadn’t done what Shockley wanted him to do. He was a very easygoing, marvelous craggy man, sweet, and never held it against me for a moment, although all his friends did, that were close to Richardson, they were very nervous with me for a while. But I didn’t know that it was my fault that Richardson was being fired, until much later. But then, well, I did get interested in local fields and in phase transitions.
Well, because of ferroelectrics, and that led to quite a number of other things, but it didn’t lead to a vindication of Shockley’s local field ideas, in the form he liked — except it did, in fact, although Shockley wouldn’t accept it, John Richardson’s and my unpublished notes on specific heat gave the answer. And then later on, there’s a paper on ferroelectrics, this review paper that appeared in a peculiar place, CERAMICS AGE, on ferroelectric behavior. That did in fact, gave Shockley’s point of view, and claimed that it really was a Lorentz catastrophe kind of thing, but this didn’t really make Shockley happy, and he wanted to fire me. In the meantime, I was thinking about —
You mean, after only a year and a half to two years, he wanted to fire you?
No, one year.
Very quickly he wanted to fire me.
Because I had not been doing what Richardson had been not doing, which was trying to calculate local fields. The minute I tried to calculate local fields, it looked to me as though the local fields were much too big, and that if you took it naively, the phase transition should have been taking place at thousands of degrees Kelvin, and so I started doing some quantum mechanics, and thought, why were the affected charges too small? I guess Bill didn’t really disapprove of that. I wrote a memorandum. I also began to think about phase transitions in general, and why do they occur where they are, and why for instance isn’t water a ferroelectric, and then I came across Onsager’s stuff about water, so I began to think very generally about phase transitions. And of course that was a little out of my depth. I did compute a number of things. One was that, already in my notebooks I think at this point or within a year, I first did a little kind of Gedanken theory of ferroelectricity, as a model theory, and then in my notebooks, I put down the note that there should be an infra-red mode that was going unstable, which was the soft mode theory, which I didn’t publish until ten years later, perhaps nine years later, which became very famous for reasons I still don’t understand. So I had the soft mode theory in there. I also began to realize that there were very interesting problems about, how do you go from the simple mean field where it’s catastrophe, to real phase transitions like dipole ferroelectrics, dipole ferromagnets, was a field theory and so on? That was the way of Onsager’s solution of the Ising model, which the Ising model was a different kind of phase transition, and that there were various things involved. One was fluctuations and one was the peculiarities of the dipole interaction. I kind of put all this down on paper in ‘57, in a special paper I gave in Russia, that didn’t solve the phase transition problem at that point. I guess all of this is prelude to the fact that Jim Talman and I, in about 1954-55, spent at least a year, he was a summer student and then I went on for a year, trying to use the ferroelectric model, which is the Landau-Ginsburg theory, to solve the general theory of phase transitions, from more or less the point of view that Ken Wilson eventually solved it later.
So you were on the right track.
I was on the right track. I did know that in five dimensions, it did not diverge. I did know the problem was that the diagrams diverged in dimensionality. I did know that the problem was a breakdown of conventional Gaussian fluctuation theory. But I didn’t see continuing in d, damnit! Well, anyhow, that’s what I didn’t do. But I did get very interested in local fields, which has always been a fascinating problem to me well, and, I did enough so that they couldn’t say I didn’t have a respectable product. Also, there were various things that I did on ferroelectricity, most of which, insofar as it got published, got published as appendices to other people’s experimental papers, because I was already busy on doing the Bell Labs thing of being a consultant primarily to experimentalists.
Right. You never finished telling me what happened when Shockley tried to fire you?
Oh, at the same time, about six months after I got there, Charlie Kittel came into the office with Gregory and me and said, “What about antiferro-magnetism?” And slapped Neel’s paper, and some early papers by Shull, and Van Vleck’s paper about antiferromagnetism, down on our desks and said, “Let’s think about antiferromagnetism, it looks very important.” And we did, we started a series of papers. Actually, if you look at some of these papers on antiferromagnetism, their original titles were “Antiferromagnetism I, Antiferromagnetism II, Antiferromagnetism III,” and one is by Anderson, and one is by Wannier and Anderson. So Wannier did this paper, big paper on the Triangular ISing Model which I kind of forgotten about, but it’s very important because it’s the first time anybody ever exactly calculated the zero point June 13, 2008 entropy. Do you realize, that’s the first paper on frustration as a problem, as a real problem in statistical mechanics, which later on became a big theme of solid state physics. The second paper on frustration also happened in that period. We had summer students. About ‘52 I think I had a summer student named Frank Stern. He had already done my mean field theory of antiferromagnetism, and I decided it would be amusing to apply this to the pure antiferromagnetic case of a face-centered cubic lattice, which I already knew was a frustrated lattice. I didn’t make him publish it, it appeared as memos — most of my papers in this period did — Frank Stern calculated the fluctuations of the face-centered Heisenberg model, and proved that if you had only nearest neighbor interactions, it was not stable. That was a theme that somehow grew, in Gregory’s work, and we played a lot with that triangular ISing model, because it had this fascinating property of frustration. Then what about why all the numbers come out funny in antiferromagnetism? That’s another question Charlie asked. The big question is, what is this, what kinds of interactions are going on? So we very quickly got into two of those things. I did the Neel mean field theory for it, the real case, and showed that that gave the numbers for manganese oxide, and I looked at Kramers’ old paper on super exchange, and wrote a paper saying, here’s how super exchange works, in this particular case, which says that when the manganese is on the opposite sides of the oxygen, there are stronger interactions than when the manganese is at 90 degrees, and Cliff Shull came to the labs about that time, and gave an auditorium talk, a good talk on his antiferromagnetic data, and I immediately realized — what?
Where was he based at that time?
He was based at Oak Ridge.
Oh, he was at Oak Ridge. That’s right.
He was at Oak Ridge, and discovered the use of neutrons in connection with antiferromagnetism, and he gives these marvelous talks about Iron and about manganese oxide and so on, and MnO had this strange magnetic structure which he couldn’t super exchange interpret. Well, I said that my theory of super exchange said it’s going to be this way, and then in three months, I was giving an invited paper at the 1951 APS meeting, and I was untouchable, by Shockley. But while I was doing this, Charlie Kittel fought the management like mad. Very hard. He said I had to be kept because what I was doing in antiferromagnetism was very important. So he and Stan Morgan saved my skin. But of course, the degree of approval was always reflected in your raise, and it was a year and a half before I got a raise, exactly at the cost of living. And I didn’t know this. In those days, it was thought much better, at Bell Laboratories, that young people not know, unless they have to be told to get lost, that they not know the exact figure. All this business of leveling with the employee on exactly how he’s doing is for the birds. They do much better if they think they’re doing all right. They always over-estimate themselves, and that’s very important. So I didn’t realize that I was on the verge of being fired. (Note: The voices at the start of this tape are so blurred, probably something about your tape recorder, that I am missing a great deal; unless it clears up this is going to be a very defective transcript. EE) ...(chatter)...Where were we? (crosstalk) I think you can tell from what I said that I didn’t know that I was not really highly regarded, and I was working on antiferromagnetism, and interested also in other things. Also, Bell Labs was becoming less interested in ferroelectrics, although it had some people working still in ferroelectric memories, and there was a very useful experimental physicist, Walter Merz who had been hired in addition to Matthias to work on ferroelectrics. One of my publications should be Walter Merz’s paper because it contains my account of my free energy theory for ferroelectronics. [At this point I seem to have been enumerating all the papers I was working on in that period (1950-53), approx.). I must have had a bibliography in front of me, to which I refer from time to time. I will try to give an ordered account true to what I can make out of my words in ‘88.)] – An experimental paper by Walter Merz could count as one of my publications since it contains an appendix which is entirely the heuristic free energy I developed for Barium Titanate. Some work was in theoretical magnetism, with the mean field theory of the oxide antiferromagnets and the simple superexchange theory which predicted the angle-dependence of the exchange integrals in MnO for instance. Gregory and I, as I said, were thinking in terms of a series of papers on antiferromagnetism and his paper on the triangular Ising model was numbered “I”, mine on superexchange “II”. And then I got involved with the problem of the ground state, of which more later. Charlie Kittel, as I said, had a lot to do with that interest. On the nature of ferroelectricity — the problem was the “effective charges” which move with the ions, which are not equal to the nominal charges on the ions and can be deduced from the infrared dielectric constant — basically, to confirm Shockley’s point of view by the kind of calculation he wanted to do one needed to understand these microscopically, and it was too early (relative to our understanding of electronic structure) to calculate effective charges. I produced a theory of them (which, from the standpoint of 40 years later, was actually surprisingly successful and had the right physics in it, but the numerical values depended on wave functions we didn’t know and were way off) and wrote an internal memo about it. Later, Overhauser, and then Jim Phillips, worked on the problem and did better, but still not up to present-day accuracy. The stat mech problem of the phase transition was very worthwhile, in that it started me on the whole problem of phase transitions, and (as Wilson later showed) the simple ferroelectric model is a good starting point for that. I’m trying to divide this period into before I went to Japan and after. These are very different In terms of how things were for me. Before, I saw myself as a very junior person, and I felt a little guilty about not working on ferroelectricity but on antiferromagnetism and — see later — line-broadening problems in magnetic resonance. Antiferromagnetism started, as I said, with the “frustration” problem of competing interaction (there was, much later (‘56), a paper on frustration in the spinel lattice) which was stimulated by the early contact with Shull. Kittel, probably, interested me in the classic papers of Bethe and Hulthen about the antiferromagnetic linear chain, and at some point, I acquired van Vleck’s copy of Hulthen’s thesis on this. (At the same time Galt was working on ferromagnetic domains and domain walls, and Charlie on ferromagnetic resonance and with Herring on spin waves in metals). This all led to very stimulating discussions of magnetism. [The next bit is completely garbled in the transcript, and I will try to reproduce something like what I must have said — because this is probably my most important work, at least at that period]. It all started from the general idea of spin waves as classical vibrations of the magnetization. (That’s not the way they appeared in ferromagnetism, they were invented by Bloch as quantum objects, but there was a paper by Heller and Kramers which derived them classically.) I started out to find the spin waves for an antiferromagnet, and they were quite different from the ferromagnetic case. Then I noticed a little paper by Martin Klein, who later became a very distinguished historian of science, in which he studied the classical-quantum correspondence for ferromagnetic spin waves, taking into account the quantum zero-point fluctuations — which in that case just lead to obvious corrections, if one realizes that a spin S actually corresponds to a classical value of square root of S(S+1). But when I did the antiferromagnetic case, the zero-point fluctuations were very much larger — in fact always diverging at long wavelengths. In one dimension, the total of the fluctuations was infinite, which accounts for the fact that the one-dimensional chain in not ordered even in the ground state. The total remains finite in two and three dimensions, though two always diverges at finite T. The numerical values for these cases agreed with other estimates — they gave good numbers (for which I had calculated bounds in another paper.) I then began to think about why things worked out that way, and I recalled something from a lecture of Schwinger’s in which he proved that the ground states of systems ALWAYS are eigenstates of conserved quantum numbers — he was, of course, doing nuclear physics, for which this “theorem” is really true. That is, they have to be eigenstates of the rotational quantum number — and hence have to be isotropic, if the eigenvalue is zero as it is for the Heisenberg antiferromagnet. (Schwinger was proving generally that the nucleus can’t have an electric dipole moment, which I noticed because I knew perfectly well that molecules do have dipole mements). It takes a Schwinger to convince you that what you know is true isn’t true. But, in fact, in an eigenstate the ammonia molecule doesn’t have a constant dipole moment because of the inversion — it is only as the molecule gets heavier that the inversion becomes too slow. Well, I had thought about it because of my pressure broadening, and why does the ammonia molecule have a dipole moment or why doesn’t it? Or in what sense does it have a dipole moment? Of course it doesn’t in the ground state because of the inversion. You make a heavier molecule, it ferromagnetic in this peculiar state? He said, well, he understands about the classical things like —
— the ‘37 theory?
Yes. ‘37 theory. And of course said Landau always argued —
You say it was equivalent?
It was almost equivalent. Tisza somehow didn’t do it quite as well... made a mess of it really but it was equivalent. Really, we had it from Tisza rather than from Landau. We had the sense of order parameters. And then we asked the question, what happens when the order parameter isn’t in classical theory anymore? What could you possibly do there? And of course at this point, this was before we knew neutron diffraction was the proper measurement. It wasn’t something tangible, like you could reach for it. And so I just set out to solve that, and we eventually came to the conclusion that the dynamics of the order parameter for the antiferromagnet could be reduced to the dynamics of the classical rotor, and it really was the same basically. And that it had a moment of inertia and all that stuff. And a classical microscopic moment of inertia would take a very long time to fluctuate out of its classically oriented direction. So this was the first conscious realization of this possibility. It involved the quasi degeneracy of the state. Of course Debye had done all that, really, but he hadn’t said it and thought about it. He hadn’t realized he was doing something non-classical. He hadn’t thought — it was very strange to have a solid body because a solid body sticks to a position and all particles have no fixed positions. So he hadn’t thought about it. Of course, the physics is there, but the realization that he’d done something was not there. For heaven’s sake, here I’d discovered something that I think is important, because I said, you know, if you’ve got the appropriate symmetry then you necessarily have these spin waves. That is in the paper.
Was it appreciated at that time?
No. It wasn’t appreciated at all.
Nobody at all.
Almost no one at all. Nobody found it interesting. Conyers, I suspect, Conyers understands everything, but he doesn’t publicize much. I think he understood it. Marshall. I guess Kubo must have understood it.
I know he understood it, as a matter of fact. He understood that it was important. I saw him, when I first was — he said he was among the first group to visit Russia — the first group of magneticians that went into Russia after the thaw of ‘56. Kubo listened, he went and talked about, the spin wave theory but had not done the zero point energy and after he saw my work he tried to do a better job. He was very impressed by it. He really understood it. We talked it over. And in ‘56, Landau said, “But the antiferromagnet doesn’t have an ordered ground state. It can’t have an ordered ground state, because of the standard spin wave theory,” and Kubo said, “But it does and can, because of Anderson’s spin waves.” And Landau said, “No, no, no, no,” and didn’t accept it. But then he did accept it when it came along in superconductivity. And some of the things that Landau said to me after my talk, we talked about the spin model and superconductivity. I think I now understand, having referred back to this original conversation with Kubo. Because Kubo argued about it with Landau. Interesting sidelight.
When did you finish the work?
‘52 or ‘53. There were a couple of other things. And, well, Stern’s thesis about — which was a question of what kinds of ground states can you have in the face centered lattice, which is a peculiar frustrated lattice. And does it go unstable, when it goes, the two dimensional. The answer is, yes. So, that was broken symmetry, and that was going to be very important later on. The other thing — well, Kubo —
Let’s just stay on broken symmetry for just one more question. I gather this was the first time you ever hit on it, in your work.
How did you feel about it?
I felt it was important.
It was important, but I mean, were you very surprised, were you scared?
Oh, no —
— you just did it.
What I really thought was that I understood this thing, now, finally, but that of course all kinds of brilliant people like Weisskopf must already have understood. I never told you Wiesskopf and Schwinger were on my thesis committee, along with Van Vleck.
That’s another thing I learned, 30 years later. 30 years later, Weisskopf told me that the reason that Van Vleck put those two on my thesis committee was that he was very proud of my thesis. And I didn’t know that. I’d never heard that. The only other positive thing he said about it was this brusque statement, “Where are you going? Where do you want to go?” And questioned me. But he didn’t ever tell me that. He was absolutely quiet about the fact that he was proud of my thesis. But he fought hard to get me in....
So, where are we now?
I was sure Weisskopf understood this, because I thought Weisskopf was infinitely wise, and you know, I didn’t really think Van but I thought Weisskopf would, and I thought Kramers would — oh, that’s the other person in this, Kramers.
Kramers called me in, when he was visiting the Institute, and spent the afternoon talking about the problem and the solution. And what he thought was important. Not because, he said, you know, he didn’t anticipate broken symmetry and field theory, or anything like that, but he said, “You know, I must tell you — ”
Another question. You said that Richardson was fired at the time that you joined —
No, he was fired, —
A little bit later?
— he was marked to be fired.
Oh, he wasn’t fired.
He wasn’t, he was, you know, we always found people jobs first. Still do. We never fired anybody without a safety net. There was a recent program called SPAC, something, that did that. But until that program, which didn’t happen until many years later, we never fired anyone from Bell Labs without finding him a job first. And he got a good job. He had a very distinguished career. He didn’t hold any grudge. (crosstalk...)
We were talking about line breadths.
Japan, OK, we have one or two things before Japan and then we go to Japan. Line breadths. (break...) All right, so now we’re continuing after a break for lunch, and there are a couple of comments. We want to talk about line broadening?
The line broadening.
Well, Kubo brings that up. The other thing I was doing was continuing thinking about line broadening, because that had been of course my original interest, and because the Bell group was very busy doing various new kinds of coherent spectroscopy and paramagnetic resonance. Kittel and Bill Yeager had begun the field of ferromagnetic resonance, and Jack Galt was working on ferromagnetic resonance in insulators. I gradually gravitated toward being the house theorist, other than Charlie Kittel, for that general area as well as NMR and picked up problems, aside from the consulting, picked up problems in that area as I saw them, things like spin lattice relaxation problems in the presence of exchange, paramagnetic resonance line breadths in diluted crystals, which turns out to be a problem equivalent to something that already existed in ordinary gas line broadening theory. So it was fairly natural to begin to think about the narrowing that Van Vleck had already pointed out due to exchange and I began to see ways that that could be done, again using this kind of general correlation function approach, that I’d used in generalizing the correlation function approach I’d used in my thesis. And that was the source of the exchange narrowing paper with Peter Weiss. It was not done cooperatively with Peter Weiss at all. It’s just that Peter had done something similar, nowhere near as complete a job of picking up on some of the same ideas, and so when I saw the preprint, I said, “Why compete? Let’s get together and write the paper together.” Now, both of these problems were ones that were interesting to Kubo, and Kubo and I —
Are we making the transition to Japan now?
No, not yet.
Before we do that, I just want to question you on one thing. You mentioned — you used the term “house theoretician.” In that capacity, were you in very close communication with the experimentalists?
Did they come and see you, tell you their problems, and can you please find us a theory? How did that work?
Well, the way it works is much more informal than that. I was doing it to a great extent with Walter Merz already in ferroelectrics. You — well, in the first place, you’re in the same group. The sociology of Bell Labs is, if you’re in the same department, you tend to have lunch together, you tend to have a lot of contacts. Without any pressure applied, but there is a tendency for theorists to talk to experimentalists. I would wander, I got in the habit —
— not in most institutions.
Not in most institutions, but it became the custom at Bell, and various people, I think John Bardeen may have been one of the very first to, well, not the first, but one of the first to do this kind of thing —
(crosstalk) — actually collaborate —
— sat in the same office with Pearson and Bardeen who wrote this long paper. Brattain and Bardeen were really in kind of a dialogue of theory and experiment. This is more or less the way I worked with Merz initially. The management certainly was conscious of this, as an appropriate thing to do, and the experimentalists were conscious of it as an appropriate thing to do. They would come wandering into your office and say, “Here, look at this data.” And on the most preliminary kinds of data. I think I’m a little unique in that I also had a tendency to wander into people’s labs and watch what was going on there. With Merz, with Matts, later on with, to a less extent with Galt, perhaps. At this time somewhere in here, they acquired a big magnet, which no one was particularly in charge of. It was kind of a general facility, and we were going to do magnetic resonance experiments with the big magnet, and we met as a committee to decide what was to be done, and the theorists could direct experiments to be done, as well as the experimentalists. So I actually had two experiments done in this magnet by Bill Yeager and somebody else. They were good experiments, and much later it turned out the results would have been significant if we’d understood them. We didn’t. And the deficiency was in theory. There is magnetic resonance on hematite and magnetic resonance on the iron impurities in barium titanate, and the hematite thing was an important early experiment in weak ferromagnetism. But yes, the contacts were very close. Yeager’s magnetic resonance in alpha Fe2O3 with Merz. There was a guy doing domain motion, Bob Miller, preliminary ideas on domain motion of barium titanate. Both Merz and he were seeing some things going on in domain motion. And I had an original idea, I had a suggestion as to what was causing this, and it was published as a memo. There’s a lot of publication, as you see, just as memos circulated to the group as a whole. It was much more common then to do that than now. There’s no way you can get any credit at Bell Labs for any work publishing something just as a memo. But then, you know, solid state physics was 50 percent Bell Labs. At least we had that feeling and maybe we were right. And who needed to publish outside Bell Labs? Most of the relevant people were there, who knew what was going on. I’m not so sure that attitude was correct, but that was our attitude. For instance, the —
Wasn’t there also a feeling that patents needed protection?
No. No. But everything had to be released, if it were going to be published, released in a very formal way, much more formal than it is now. Spin Relaxation to the Exchange” that’s the exchange reservoir. Bloembergen had the same idea and published it. I actually sent this paper to Bloembergen before he published it. I don’t know, he probably didn’t read it. I now know that one should be much less devious about these things. I think Bloembergen got more stuff than he could read. I think I actually talked to him about it, too, and then it was published, with or without — I don’t remember whether there was acknowledgement. But I don’t think that in the Bell Labs hierarchy, I lost anything by having had that idea and then letting Bloembergen publish it. Later on, I would have (...) So, yes, in the first place, we were much more internalized. In the second place, my collaborators were all, almost all experimentalists, in-so-far as I collaborated at all. An apparent collaboration, like with Weiss, wasn’t a real collaboration, it was just that we wrote similar papers at the same time and put them together.
Later on you did some collaboration.
Right. But we’re talking about this period now.
This period. So — well, ....Kubo visited the labs a couple of times, and then we had this interaction, I guess we both gave papers at a notorious meeting 1952, was it. It’s in January, 1953, issue of REVIEWS OF MODERN PHYSICS, big meeting on magnetism, where a lot of interesting things happened.
I think I’ve heard of it.
Kubo gave the — I gave the exchange narrowing paper, I think, and he gave the spin wave paper. We talked about line broadening, I remember later — I don’t remember what the occasion of the meeting was. I think it was some meeting on statistical mechanics. I remember attending it in Pittsburgh with Kubo. That may have been later, because at this point he had his senior with him. Kubo and us were trying to persuade me to go to Japan. I remember, at that time, saying to Kubo, “Well, you know, these people are talking about what a mystery irreversible thermodynamics is, and how difficult it is to calculate transport processes, but here we have these line broadening techniques, and correlation function techniques, and we should calculate these things.” He hadn’t invented the Kubo Formula yet. But it was clear, I don’t know whether I was saying this to Kubo or to other people, it seemed clear to me that there was a lot of nonsense being spoken about the necessity of going through a master equation and the H theorem hypothesis, and assuming irreversibility, and all that stuff, when all you really needed to do was to calculate the correlation functions directly. And Kubo was intrigued by that, and he certainly was intrigued by my work on spin waves, and so they started trying already in 1952 to get me to go to Japan for a Fulbright. The first Fulbrights Japan had were in 1952. I don’t know why they picked me particularly. I was very young. Perhaps they were – well, Kubo had noticed my work. I said, “No, I can’t possibly go, and besides I can’t go for a year. We’ve just bought a house, and it requires a lot of work on it, and Susan was very young” and so on. And so they said, “Well, maybe that’s all adjustable,” and before I knew it, I had committed myself to going for half a year in 1953.
Did you go with the family or without?
We all went. And we also — that’s the reason I was at the big international meeting at Kyoto. Not that I was the kind of person they would have invited normally. But it was clear that they were aiming to get the elite in theoretical physics together, and I wasn’t yet the elite of theoretical physics, but I was there, and met all those people for the first time, because of course Bell Labs people did not do much traveling or socializing with the high muckamucks in physics, except in solid state physics particularly.
Well, just a few.
You don’t want all that. Well, let’s see, some important ones — Froelich was there, of course.
Why of course? He wasn’t that well known, was he?
Oh yes. At that point he was very well known. In fact, much more well-known then than before. 1950 had been the year that he and Bardeen had developed the phonon theories. And 1950, Froelich had been around Bell Labs, I think he wrote a book on the theory of dielectrics — and he and I talked a lot, and he was very influential on my — well, he had the static version of my correlation function idea, correlation function also determined the susceptibility, and he used that in a practical way to calculate the Debye formula which was very interesting to me. And Froelich was just in the process of introducing solid state physics as a field theory problem. He was almost the first to say phonons are a field, not lattice vibrations, which you happen to quantize, but a field, which you then interact with the field of electrons, and do quantum problem. And his theory of superconductivity was essentially a metallic version of the polaron problem. Now, he was I think at the top of his reputation at that point. But he was in the process of climbing back down from the Froelich-Bardeen theory, as John Bardeen wasn’t. John was a very stubborn person, and he said, “It will explain all that stuff,” and Froelich had realized, no, it wouldn’t. And so he gave actually a sliding charge density explanation at that meeting for superconductivity. Onsager gave this totally incomprehensible talk about the de Haas-van Alphen effect. And that of course was the source of all of Fermiology. But nobody who was there seemed to understand what he was talking about. I understood a little more, because Shockley had been interested, for purposes not of the de Haas-van Alphen effect but cyclotron resonance, had been interested in orbits around Fermi surfaces, but Shockley had been restricted in his thinking to simply ellipsoids, whereas Onsager was thinking in general geometry, and you realize exciting things happen with general geometry. I didn’t really comprehend what he was up to. I didn’t actually talk in the main program, but there were special seminars. One of them was on line broadening, and I gave a fairly long talk.
Do you know when that was?
September ‘53. The Proceedings exist and are fairly well known. I have a copy. I gave lectures in English.
Were they translated or did they understand English?
They were supposed to understand it, but what they did was transcribe my notes including errors into a little book which then later was compiled, called THE LITTLE RED BOOK, and I do have a copy of it somewhere on the shelf...I finally took one of the three copies of it, they let me have it because I lost mine. And everyone —
— that’s priceless —
— in solid state physics, in Japan, basically, was in those lectures.
Let’s see, maybe we should — it would be a good idea to comment a little bit on what was happening in Japan in solid state physics by this time. Do you think this is a good kind of way of getting to that? You can’t tell from here.
No, you certainly can’t tell. This is my own stuff.
Sure. Just do it from memory.
Well, Japan had a long history in magnetism.
Right, going all the way to —
— going way way back, practical magnetism, and it also had a few already established theorists, particularly Nagamiya, fairly senior established theorist who was respectable in the theory of magnetism. Van Vleck had some friends who I guess had done experiment and theory confirming his early work. And there were some technical ferromagnetics types. The — I think the president of Tokyo University, one of them, no, it wasn’t the president of Tokyo University, but it was very senior, maybe it was Kaya, this man that Kubo came and talked to me with, was a senior experimental magnetics person. There were various people who were Kubo’s students and Kitani’s students who were at various levels, and I never really distinguished between the junior graduate students and people who were more or less post- doc level. Essentially some or all of them seem to have gotten to Tokyo to listen to these lectures, and also to lectures on line broadening problems. Actually I didn’t give lectures on line broadening problems. I gave a seminar, in which they had to report on various things. And Moriya; and there was a guy, Hasegawa Ohno, I wrote a paper with him, it’s missing — I don’t know whether Kondo was in those lectures or not. Maybe he was not. Ohno Pagila and Oguchi — whom I now have some problems with, disagreements on certain aspects of physics, but he was pretty good. I think those are the main people. Kondo must have been... (off tape) There wasn’t, aside from Kubo, and Nagamiya, much of a solid state establishment, as yet. There was a —
All magnetism types?
Yes, they tended to be —
— to be, yes, that was the main thing.
And that was because of the tradition in magnetism.
The tradition in magnetism. There were some field theory people too, already there, it had a tradition, but actually, was not doing great work at that point, I guess later on... Anyhow, the seminar I think was quite influential and did tend to keep contact with a lot of people. There was a spectroscopist whose name I’ve forgotten now also, who later on worked with Charlie Townes and came up and stayed with us with his daughter. She was five at that time, and later on at seven or eight, when he came to visit us. Charlie Townes came out to see us. I made contacts with a lot of physicists. Kamagai was one. He was a solid stater then, and had a big group in spectroscopy, but then he later became a nuclear physicist.
It’s not the same? Who built the accelerator?
Yes, later on he built an accelerator. Now, he had a magnetic resonance group, pretty big and very active magnetic resonance group. I have a lot of pictures of — because of course there always were pictures that they gave you. I have a lot of pictures of this guy and his enormous group. Also of Joyce and Susan and myself looking at chrysanthemums. Beautiful chrysanthemums. They really were. Hundreds of chrysanthemums each with a paper tucked under it to keep its petals in place. Oh, it was a marvelous time to be in Japan. And all these things were still in place. You got your strawberries six in a beautiful wooden box, sturdy wooden box, and all kinds of fruit came in little woven baskets. And handicrafts all over the place, marvelous. We also had to live cold, dirty, unsanitary. The stove kept things going. You had to stick your behind out of the back door in order to light the fire and then put it under the bathtub, you were lucky to have a bathtub. We lived in a place that was two miles from the nearest other Gaijin and where there were no trees over ten feet tall. Well, anyhow, it was a very interesting experience.
So what came out of the interactions?
Oh, a lot of things came out of the interactions. Kubo greeted me when I came with Kubo-Tomita theory , which was the next step from my exchange narrowing theory, a next step, and then Kubo was working all through that period, and we were talking together, and wrote some joint papers together. He was working all this time on the general idea of these correlation function methods, which eventually evolved into the Green’s function methods. One of the things he showed me and we worked together a little bit on during the time I was there was this periodic boundary condition for the Green’s function. And he is of course credited with that. So I think it’s fair to say that the stimulus for the Japanese developments in Green’s functions, and the direction of doing correlation function approaches, came from that interaction, from Schwinger by — via me — to Kubo to Matsubara, to a great extent.
Well, Matsubara was a field theorist. He seems to have worked later on with Kubo, after I left.
I see. You didn’t meet him.
I didn’t really meet him. I didn’t have any direct influence on Matsubara, but Matsubara was two or three years later. How that then got to the Russians, I don’t know, but they always referred to it as Kubo’s boundary condition and Matsubara’s functions, so — a lot of other people invented them independently, of course, Montroll seemed to have, and Martin-Kadanoff, but the one that got to the Russians and got used went via Kubo and Matsubara. Well, I don’t know. It’s not too important, except that there was this period when Kubo and I were working together and thinking about this fact, that we really, that this approach, at least to linear response functions, had something which made it just look silly to fool around with Boltzmann equations and distribution functions and all kinds of things that went into it, that the chemical physicists were doing. Now, shortly thereafter, essentially independently, Kubo and Lax I guess in that order, came up with very straightforward ways of expressing conductivity and various other transport properties, in terms of correlation functions. Clearly, Kubo was thinking in these terms at that time, and busy inventing the boundary condition as part of it. And this was the first step, and I guess the next step he published, about the time that I was leaving. I wrote a long paper for the J. PHYS SOC. Japan, about the whole question of motional narrowing. I formulated ways to calculate relaxation times, given the spectrum of the motions in the material, as well as the line shape, in stochastic theory? Some of these things I did in stochastic theory in that and in the pressure broadening problem came in useful later on, just as methodology were re-invented and made formal mathematics out of. But really just using the most practical devices. I guess that’s probably enough about the Japanese experience.
OK. But then you came back to Bell.
And I gather you repeated some of the lectures, is that right?
Yes, I did some of the lectures. Bell was interested in getting into magnetism.
They already were into it.
Well, we were into it, and there was Bozorth’s group, and Williams and Galt and in particular had been doing this work on the Domain motion in ferrites, and there was beginning to be an interest in ferromagnetic resonance in ferrites. A group had moved into – there was a group working with Pierce on the traveling wave tube, generally electron tube development and research. I guess that people more or less felt that the electron tube had had its day, and solid state physics was a good direction to go, and a gang of them came up from Holmedale, came up to Murray Hill, and were interested in magnetism, particularly ferrites, including Al Clogston, Suhl and Walker, who all were ex-tube people. I guess I don’t know whether Harry Suhl was down there. He must have been. Harry had already done some work, had some interest in trying to do magnetic resonance work in semiconductors, not particularly successfully — the Berkeley group got there first. And Shockley always had it in for Harry because of that, and I guess Harry thereupon sensibly moved to Holmedale and didn’t come back until Shockley was gone.
When was Shockley gone?
I think he was gone by the time I got back, and then came back for a while and left again.
I have this inclination from somewhere —
– I don’t know. Oh, before Charlie left, he had been named — an independent department had been made, independent department of magnetism, with Charlie as department head, because he squabbled with Shockley so much, and I think basically Charlie, although he was negative about Shockley, and I think whispered in Jim Fisk’s ear about Shockley, I don’t think he left because of Shockley. I think he left because of opportunities that were bigger. Bardeen, as I was saying at lunch, left about the same time, partly because he couldn’t get along with Shockley, and when he left, gave this rather bitter speech about “transistoritis.”
Oh yes. Oh, he did actually give a speech about this disease, the disease transistoritis.
I never heard that.
Oh yes. But also he got two professorships, each full time, and his salary was, each of the professorships had a salary no different from what he had at Bell, so he was gaining quite a bit in salary. Bell was very conservative about salaries in those days.
I had no idea.
They didn’t respond, actually they still don’t respond to outside pressure. They will respond over the course of the years, they’ll tell you, but they don’t respond to this year’s offer. Hal Lewis had been there working with Bernd Matthias [or not with Matthias], working kind of in superconductors.
Oh yes. Superconductivity, we had, at the same time that we got the Bitter Magnet we got liquid helium. This liquid helium brought back Matthias, I think they made a virtue of necessity. If they had to work with helium someone had to run it and Bernd Matthias was the only good liquid helium physicist we had on the premises, already trained, so he then hired Corenzwit. He is the kind of person that only Bernd could find. Incredibly good, incredibly smart, and no formal education, and incredibly faithful, went on providing us with liquid helium for decades.
Is that 1951?
1952, or ‘53, he came back. So we had liquid helium on the premises, and Matthias, as a condition to coming back and running the liquid helium for them, was given his head to do superconductivity. And Hal Lewis we hired because of McCarthyism. He was fired by the University of California, spent a year at the Institute, and decided he might like solid state physics, and came and joined us at Bell. Bell had no prejudices against hiring him. You know, Hal would make a funny story about all the funny questions they asked him, but when they got all through, Bell made the sensible response of hiring him.
I didn’t realize Bell was so liberal.
It wasn’t liberal. It was just practical.
Practical, he was a good man.
Well, there were other people who were practical. He never lost his clearance during this whole period. Ended up being a terrible hawk. But he just didn’t want to sign any Loyalty Oath. So — and there was a Bell Labs loyalty oath. Some of us signed it, most of them signed it. I didn’t. And nothing happened. It wasn’t a loyalty oath, it was a questionnaire. I didn’t sign it. Nothing happened to us. Bell wasn’t liberal. Don’t pretend it was liberal. But it was sensible.
It just, for some reason, had sensible practical policies, in-so-far as they could have them. Bernd came back. A lot of things. One was, George Feher had been hired, with a salary of slightly more than my salary when I left. I had finally gotten a couple of raises, reasonable raises. And a lot of people were leaving because of more freedom in academia, certainly freedom to travel a lot more in academia.
This is a time to continue on tape the conversation we were having in the car, on Bell, the beginnings of Bell as the major force that it came to be in solid state physics, although that’s not fair either, because it was important before.
It was highly important before, but I think everyone who had some — certainly, before the war, were aware that it was a unique institution before the war, and was doing something that no other industry was likely to do. But before the war certainly it had had some very peculiar policies. It had its superiors and you had to wear a white shirt and a tie and a hat. And all that kind of stuff. There was a man named Gorton who was executive assistant for Harvey Fletcher, who came around when Bernd was hired and kind of objected to his not wearing socks. People who were hired right after the war I think still lived in this state of some surprise that the place was as liberal-minded and easygoing as it was. Once I got past that first invited paper, nobody really ever told me what to do. I still felt a sense of guilt about ferroelectrics, but not that much.
Did that have to do with Shockley?
That had almost entirely to do with Shockley.
Right, and he —
— there were people who were — some of the things they let people get away with were incredible, in the way of not working at all. Other people more or less did what they were told. The whole chemistry department was in the position of doing what it was told.
Was Morgan still on top of it?
Morgan, no, Morgan was not in the —
In the early days he was co-head of the solid state —
— yes, he was co-head of the solid state. Then the solid state split and Shockley took the transistor people, and Morgan took the solid state in general, and at the same time there had been what was called the physical electronics group, which was the electron tube group under Addison White. Just previous to that, that had been Fisk’s and then Wooldridge. Fisk, Wooldridge and then Addison White. And that was quite independent.
By the way, Wooldridge, Fisk.
Yes, or, the other way around. And Conyers Herring was in that group. That had a slightly different sociology.
Herring was in that group?
Yes, Herring was in that group.
In Addison White’s group?
Yes. Yes. Yes.
Again, I must have known that at some time, but —
— Addison White’s group, and Kittel was in the Morgan group. Shockley and Bardeen were in the Shockley group, when it split. Later on, of course, Suhl and Walker and Clogston came up and they were in a separate group. This was this little magnetism group. So solid state began to fragment. And let’s see, — Harold Lewis when he came was in our group. Peter Wolff when he came was in the physical electronics group. That physical electronics group was in the process of being dismantled, because it had a very good set of people in it, like for instance Jalisca Molnar, Homer Hagstrom, Addison himself, John Hornbeck, who were extremely powerful people, but they obviously were not the wave of the future. They were not a tube group, they were the physics of tubes. The tube group was Kompfner and Pierce and company. So solid state was growing and as I said, no one was really complaining about being forced to do anything they didn’t want to do, except me very briefly. But it began to be clear that, well, we were having a very easy time replacing the experimentalists we lost, we weren’t having an easy time replacing the theorists we lost.
Namely, Bardeen, Wannier , Kittel.
Kittel and shortly, Hal Lewis. Wannier hadn’t gone yet, actually, either. Lewis and Wannnier were still around in 1955-56. So two things happened. One was, there was the Shulman revolt, which was a list, it took the place or form of a list. Shulman would take your name and your salary.
This is Bob Shulman?
Bob Shulman. And if you gave him your salary, he would show you the rest of the list. So having started with three or four, soon everyone knew everyone else’s salary, and what made people angry was not so much that they were poorly paid but that everyone else was poorly paid too, even people they really admired like Conyers Herring and Bernd Matthias were equally poorly paid. So it was realized that starting salaries had risen so fast that everyone was at exactly the same level, and all this secrecy in management was about the fact that no one was paid very well. And I guess it was Baker at this point, who was already a fairly high level executive director perhaps —
— this was when?
— broke the ice by simply initially, the first year must have been ‘54, or ‘55, jumped the salaries of about six or seven people by a factor of 50 percent. Herring was one of them, Shannon was one of them. It was also true of the mathematicians at the same time. Herring, Shannon, and I don’t remember who the others were.
This happened just about when?
‘55. And he broke the salary lock —
— secrecy —
— secrecy, and after they got the 50 percent increases, the salary secrecy went back on again. Of course no one was going to admit that. Then the next year they hit I think Matthias and myself and some other people, so the salary thing was solved. This made us all feel our oats, and I think there was a great deal more of sense of freedom beginning to be present as well. The second thing was, the theorists, stimulated by Hal Lewis, who was ever a trouble maker, decided that they didn’t like this idea of all the theorists, each theorist being a house theorist for one particular group, and being responsible for all of the theory in that group, and being judged primarily on his abilities as a consultant and not as a theorist, and they were of course probably wrong to do this because they probably would be better paid and better treated if they did that, but it gave us more self-respect to say, “We want to be a group.” That was stimulated also by the fact that we had begun, Quin (J.M. Luttinger) and Walter Kohn to begin with —
Did that start at the same time?
Oh, that must have started just about the year I came back from Japan.
Both of them.
Both of them. Oh, they started at the same time. They came in every summer, and we also had people like David Pines visiting regularly, and later on many more consultants, but these people were traveling from Copenhagen to Brussels to London to Oxford to Leiden etc. and coming back, and then going again and coming back. They seemed to be free as birds and they had their own contracts and they could hire their own post-docs. And they had people they could tell what to do. We of course had nobody except the experimentalists, who were busy telling us what to do. So we were all very unhappy, and Hal Lewis was busy telling us that we should be unhappy, so management asked, what our problem was, and we said, “This is it,” and designed the department, which had post-docs, which had a very big budget for visitors, and a rotating department head who was to be a theorist.
Who was the first one?
The first one was Conyers. And after considerable discussions, which were very exciting and rather surprising in outcome, they gave us everything we wanted. In addition of course at that time Conyers and I at least had had massive increases in salary, at least what looked massive at the time. We’d laugh now but at the time they looked massive. At least we were matching our visitors. You know, we knew what Walter and Quin were being paid, and it looked ridiculous to us. I suppose when you think in terms of benefits and traveling salary and all that stuff, it wasn’t so huge. But we thought it was very unfair for them to get $50 a day when we were getting $15 or something like that. It sounds silly now.
So Bell at that period started to be very exciting.
Yes, very exciting.
Especially summers. You had more visitors.
Yes. And we had our own program, stimulated by Peter Wolff, who had come from the particle physics, high energy and nuclear physics area, and they weren’t separate then. We also began to realize, we had to get more people, and no one was being trained in solid state physics.
(crosstalk)... the fifties — into the issue of applying the techniques of high energy particle physics to solid state?
Yes. Although Peter himself didn’t do that as his thesis, he certainly was conscious of the big developments happening in nuclear physics for instance. We all were. Conyers of course is always conscious of everything, nobody knows what he doesn’t know.
But that was something that Bell was beginning to understand.
Yes. Earlier than that I would say — well, through Conyers I learned about — through Gregory I learned about the Landau-Ginsburg kind of — or at least Landau, phase transition theory. I invented my own Landau- Ginsburg theory.
Well, that was my ferroelectric theory. Not Landau-Ginsburg theory as in superconductivity, because I knew nothing about superconductivity. Hal I guess knew surprisingly little about superconductivity, because he did not make much of a fuss about Landau-Ginsburg theory, just at a time when it was an important thing to know about. Hal seemed very bright to us, but I must say he didn’t achieve much then or later, except for the military.
So now we have this new group.
Yes. So, Peter — well, Conyers really had us in touch, I guess it was through Conyers I first heard the term elementary excitations, and began to understand what that was about. That would have been ‘51 or so, very early. ‘51. And we had talks about field theory. Through Gregory, I got in touch with statistical mechanics. Through Charlie, Harry was really interested in — no, Charlie and Conyers both were interested in collective phenomena. Conyers of course was talking with Pines a great deal in this era about correlation energies and collective phenomena, and at this period, Charlie, about ‘53, Charlie and Conyers had written a theory of spin waves in metals.
You were commenting, do you want me to backtrack or do you remember your —
I think I can remember, yes.
OK, you were talking about the new group.
We had these various interests, and also a need for new people, and a feeling that there weren’t enough people being trained in appropriate parts of mathematical physics. There were actually at the time more, but we weren’t getting them. The world as a whole needed or wanted condensed matter physicists, and they were getting better jobs at universities, and of course theoretical people do tend to like to go to universities.
Were you suffering a lack of students?
Well, we had of course summer students. I told you about, I guess I did mention both Talman and Frank Stern. I had, let’s see, one other, I was trying to remember who the other student was in that period. And Conyers I guess, even before this period began, had had some summer students from Princeton. We had a Joe, what did we call it, Joe Glitsick — I forget what his — no, he was a post-doc. He was a post-doc. Gosh, his name has disappeared. He came later. Anyhow, we began to feel we ought to look around among particle theorists, and nuclear theorists, and also we had to know what was going on in these fields, so we began to make much more contact with the outside world than we had, and in addition, we expanded the summer student program, so we just had a big group around all the time, and during that period — oh, Adler was another student I had during the period.
These were regular students or just summer?
No, these were summer students.
Summer students from other institutions?
From other institutions, mostly Princeton. Stern, all of them were, in this case. And you know, although they were very able, I now realize how incredibly able they were then....
OK, we’re back.
Anyhow, we had a lot of people come in, including of course Walter Kohn and Kerson Huang and Keith Brueckner. We had J.G. Taylor later on, had a lot of interaction with him. We had John Ward, several times. John Ward actually was a regular visitor at Bell Labs. Part of the time he worked in the tube group of Pierce’s. He was, like a lot of people, more people than I had realized until recently, he was obviously manic-depressive, and in his depressive periods, he wanted out of particle physics, and one thing he could do was play with electron tubes, except that in his depressive periods, he was also a danger in the tube laboratory. He broke things. But anyhow, he came around, and then sometimes did theoretical. Gregory Wentzel was around, Brout was a very regular visitor. John Blatt was here in ‘56 and in ‘58. He was doing a nuclear physics problem on the computer. Made a mess of it. Made a mess of the computer. He did the problem just fine. Because he insisted on programming it in machine language. Nozieres, at least two summers, worked a lot with Conyers. Schrieffer was there one summer with Conyers. Of course, I had Pierre Morel later. And later on we had a lot of post-docs and so on. Once we got the group going, then we had the possibility of post-docs, and our first post-doc I think was Phillips and our second one Hopfield, who did reasonably well. And then the third one was this Joe Blitz who’s disappeared. By that time we also had other people who were very good as post-docs.
Whose idea was it to have such a big travel budget?
All of ours. Peter was pushing for it.
Yes. I was pushing that. I was happy with it and Conyers certainly was pushing for it. We’d been so successful with Quin and Walter.
I see, they were the original ones.
They were the original ones.
And they were the model for —
— they were the model for it, but also, there was the idea that we would bring people from different areas in, and familiarize them with what we were doing, and have interactions with them. I guess the need for people was somewhat later than just the idea of, let’s get a lot of people around. I mean, the need for hiring people, that was later. That was in the late fifties, when we began to be desperate. We hadn’t been able to get any of the people that we were seriously trying to hire.
Just one question before we get to the people. At the time that Quin and Kohn started, I guess they were not that well known yet.
You were sort of taking a chance. Who was the person that recognized them?
I don’t know. Probably Conyers or Charlie — I was not senior enough. Charlie would have known Quin. Who else would have known?
I don’t think, Quin said, when he first came to Bell, that’s where he met Walter and the magic happened.
Well, you know, these things were done at that time by department heads, some experimentalist basically taking advice, and Addison was best at taking advice from Conyers, so I would have guessed that Conyers was responsible for Walter Kohn.
Because they only did that really terrific work after at Bell.
That’s right. But I don’t know, Charlie, yes, that’s probable.
Again, before Quin really did anything.
That’s right. I knew Quin actually because I’d been interested in local field problems, and I wasn’t consulted, but certainly Quin had been around and given talks. And I knew the work, Gregory knew the work. We were both interested in this Luttinger-Tisza stuff. That was part of our magnetism and ferroelectrics game. So that might well have been part of the interaction. I don’t know.
He may have introduced it in a lecture. He was lecturing to experimentalists at that time, wasn’t he?
Yes. Oh, yes. They may have been brought in also to do that.
To teach solid state theory to experimentalists. Bell was always very good, from the early days, in educating and in self-education.
That’s right, they started all these lectures. Then they just gave that up and these other people just came. Did I say Pines? Abrahams?
Yes. You didn’t say Abrahams, you said Pines.
Yes, year ‘57, Pines and Abrahams were both there and working together with Feher. So, it suddenly became a very exciting place. Oh, Klauder was — well, he wasn’t really a post-doc, he had had a Bell Labs fellowship at Princeton (with Wightman), and with the Bell Labs fellowship came the obligation to bring him back somewhere in the organization, and since he was a theorist the only possible place in the organization he could come back to was us. So we got Klauder. And he actually was helpful. We hired Lax during this period. He was a visitor one year and then we hired him permanently. Then we made what was probably a mistake. Westinghouse and GE, well, first Westinghouse and then GE were beginning to collapse as fundamental research labs, and it was probably a mistake to take the best people, what appeared to be the best people from those labs, hire them, because they turned out, although they were the best people from Westinghouse, they weren’t all that great. So we got three who never turned out to be really superb. Two from Westinghouse, one from GE. They weren’t as good as they’d appeared to be. So we filled up considerably more. But our best success was with post-docs. We had one marvelous post-doc after another. And then gradually began to be able to hire them in.
So the institution had really shifted, what, in the direction of having more theorists?
Yes. Well, yes, we added three theorists, just by bringing in people who had been more or less electron tube theorists, although Harry Suhl had done experiments in solid state. So Walker and Suhl were additions. And Clogston, who had been partly experimental but became totally theory.
Someone in the administration was definitely working with you, somebody high up.
Oh yes, Baker and White, both. Well —
At that time wasn’t —
— (crosstalk) Millman was not positive.
He was not.
Well, he just seemed to follow orders. That period Millman was director, White was executive director and Baker was a vice president.
He was a vice president, I see, and Fisk was president.
Fisk was president, and Fisk I think was already becoming detached. So it was strictly Baker, as far as we could tell. Baker just — he said very vague things, and White said, his way of dealing with it when Baker was vague was to realize what he had not said he couldn’t do, and do it. So Ad was responsible for giving us things but Baker did it.
I didn’t realize that Ad White had been such a powerful person in the administration.
Oh yes. Oh yes. Very effective, in that sense. He was a person who got along with Baker just fine, and Baker was in as vice president, later the president, for a long time.
So we were just completing the organization. You were talking about the comments, this great change that occurred at Bell in the mid- fifties.
Yes. Well, in the early fifties, I guess, I had been offered — it was kind of a standing joke — an assistant professorship at Yale, they had gone to everybody, and the likelihood of promotion to tenure was zero, and the salary was not very good, and there was no solid state physics at present, and I turned it down. But with the proviso in my own head, that if anyone gave me a really good job, I was about to go. The only other offer I had was from Illinois, and that didn’t suit me because I wasn’t going to be where my parents were. And that was actually a very nice offer. If it had been somewhere else I guess I would have gone. Later I was almost sorry I didn’t take the offer to Stanford.
Was this all around ‘55?
Late fifties, early sixties. It would have been very nice, it would have been good for Stanford, he said immodestly. They needed me. Things very exciting, what was going on, at that time.
What was exciting? This was a really exciting time. This was a period around ‘55. We were talking about the details.
All kinds of things were happening.
Yes, this is where solid state theory goes crazy, in a sense, because of the application, serious application of new techniques.
Yes. And at first, I wasn’t really so much involved in that. At first, I was mostly, I was doing a number of things, but mostly I was busy being a house theorist for George Feher. I was busy with what George Feher was doing. He was doing the first very good resonance measurements on semiconductors. I guess the thing had been started by Bob Fletcher, and George picked it up and it was really very exciting magnetic resonance work, and three men, particularly Walter, were busy doing the physics of the actual wave functions that he was measuring, the numbers he was measuring. I was kind of the house theorist, whose responsibility is to do a back of the envelope and see what order of magnitude things were, the hyperfine structures and everything else. To me it wasn’t that interesting to do the detailed calculations of wave functions. Actually, I think it was a very exciting detailed calculation. Earlier we had cycloton resonance work on semiconductors and Walter and Quin did this beautiful job on exactly how the orbits worked too, and again that was spectroscopy. It was really very like what had happened back at Harvard, I found myself drawn to the non-equilibrium and breadth problems, kind of the underlying interaction problems between [??]. At this same time, I lectured with and worked with this little magnetism group that was growing. So I was thinking about ferrites with them. Thinking about magnetic properties, the magnetic exchange interactions, and we began — well, we were focusing down on exactly how the line width problems worked, and how do you figure the transfer of energy between spin waves and the rest of it, and Harry and Larry began to develop these very intricate ideas about nonlinear effects. So there’s one paper where Harry and I began what later became Harry’s specialty, the nonlinear theory of spin wave excitation by high powering ferromagnets. Kind of one of the first nonlinear instability problems, thought it was nowhere near as classic as Rayleigh instabilities and so on, but it was kind of the first solid state instability problem. We rather liked it because it had its quantum aspect. But really, quantum levels interacting with a macroscopic branch of modes, rather than the whole thing being on the classic macroscopic level, so it was an amusing thing. But in principle perhaps not as exciting as other things we were doing. So there’s a number of magnetic things. But meanwhile, I was thinking with George about what was going on in his work. Also, I guess I was doing house theory with Bob Shulman. Some stuff there on line shapes. Then I began to really focus down on the physics of what was going on in Feher’s samples. Feher had some of the first really clean doped semiconductors. He and I were both very interested in these problems. Mel and later on Halperin were interested in spectroscopy, the question of impurity bands and distributions of energy levels interacting and so on. I was listening to some extent to Quin and Walter who at the same time that they were doing this stuff were doing quantum theory of, transport and at the same time, I guess, also there were some quantum transport ideas going around from Hal Lewis, and from Gregory. Now, we all were aware of the Pearson-Bardeen doctrine, who said the two words, degenerate semiconductor, and the other two words, impurity band. We said the words, we didn’t really know what they meant, just knew that if you had a random substrate of energy levels, that somehow the electrons would communicate. Quantum transport theory — that’s one of the things, that’s THE thing I think above all that’s missing in your chapter. It began, of course, with Einstein and Onsager and there were, during the war there was this multiple scattering business developed, that Mel Lax had a lot to do with and Les Foldy developing a formalism, kind of a randomized generalized version of Bloch’s theory, just telling you how waves move through a medium, whether it’s regular or random; generalized medium, which was really a very important development.
It came out of the war?
Well, Foldy was interested in bubbles, in sonar and bubbles, and I think Mel was doing radar in the presence of junk. (Hoddeson inaudible) They both were — you know, wartime work led to an interest in this general problem of waves going through media, in a very generalized context, where sometimes the medium may have big, strong scatterers. So everyone kind of knew that they did these things with perturbation theory, and they looked for a smooth medium, that’s dielectric constant, and for a regular medium, that’s Bloch theory. People hadn’t really thought, how do you make an irregular medium. And Lax and Foley had developed a formalism. Second source was —
It wasn’t really semi-classical. It was basically saying — well, it’s kind of the zero order thing was to replace the potential by the scattering matrix and use perturbation theory. Use the optical theorem. Van Hove had a diagram theory that was actually very insightful, it’s just saying what people already must have known, in a way, like ferroelectric theory. It was very important to say what people must have known, in a sensible way. It was really a perturbation formulation, the most naive thing in local scattering theory, but it was done formally and very clear. How one might take any problem and do this to it, with systematic perturbation theory, classifying orders in N (the number of scatterers). It’s a little bit like what Brueckner did in cluster theory, classify orders of N and then you’ve got a perturbation theory that’s appropriate to a large system. In fact, what Brueckner did is based on scattering theory too. There’s a multiple scattering assumption underlying. So multiple scattering theory was important in underpinning Brueckner, but also Quin and Walter were interested in, and Van Hove, were interested in these kinds of waves. Resistance in metals, transport in metals waves, scattering as they go through a medium, are all logically equivalent. The problem they got stuck on later on was, or got involved in later on, was the anomalous Hall Effect problem. So we were all kind of talking to each other and having — added to our arsenal, multiple scattering ideas. And we had begun to think, there is such a thing as a common transport theory. There is such a problem as how do electrons actually travel through metals. And in this sense, we recognized it as that. But we at least had multiple scattering theory hanging around in our minds.
May I just interject something? I forgot to ask Quin this. Are you sure that — he was not only coming summers, but he was coming once every two weeks or so?
Yes, I think he gave some lectures once every two weeks, as well as coming summers.
I think that was one of the things which had, maybe that’s the way we got in contact. He may have been coming from — maybe it was Michigan or Penn. Maybe he was —
(crosstalk) No, he was at Penn. He was commuting.
Maybe he was commuting —
— he was living in New York and Penn.
Yes, I see. And giving lectures, and (crosstalk) —
— and then eventually he — well, he wasn’t at Penn very long. Just a few years. He moved to New York.
Well, he also was at Michigan for a while, wasn’t he?
That was earlier.
Earlier. Alright. I think I first made contact with him when he was at Michigan.
I see, and then he went to Penn.
Then he went to Penn, and that must have been when he was lecturing. Walter was of course at Carnegie Tech, and would come lecture.
Right. OK, I just wanted to check that out because I don’t think actually — OK, continue. I’m sorry.
That was one thing. The other thing was, we were all interested in — Conyers, I guess, as usual, brought it up in the literature — his journal club was incredibly valuable to us all the time. Once every two weeks. Motts’ ideas on the Mott transition, and well, the general idea of the metal-insulator transition, and I don’t know why we were familiar with it, you won’t find it in modern textbooks on solid state physics, but maybe it’s in Seitz, I don’t remember, I doubt it. But somehow we knew that there was this question of when screening starts and you have a first order transition to being metallic, and maybe just because we’d read Motts’ papers. So there was a question of why doesn’t the impurity band conduct? That was Gregory’s question. Gregory wrote a little paper, which disappeared in everybody’s mind but my own, on what he called “wiggly bands”. He pointed out that diamonds are incredibly impure, and yet they have no conductivity at all. There is no natural diamond with impurity of less than 1 percent. For a semi- conductor, it would be a metal. Why isn’t a diamond conducting? He was interested I guess and worked on it, diamonds as photo conducting elements, if you want to make particle detectors out of them it’s very impractical. Ken McKay was another of those powerful people in that group that I mentioned in management. He worked our diamond. So the question was, why doesn’t the impurity band conduct? And Gregory and I began to think about that, you know. If it were regular, it would conduct like a metal. And its conductivity would be fairly high. And at the same time, I was looking at Feher’s experiments and thinking about all the various phenomena that were going on there, in terms of relaxation line breadths. Not much of the work got published then. Most of that work got published as part of Feher’s great massive paper on “passage effects,” and then Meyer Weger and Feher did a second paper which was even longer on passage effects. But I was most interested in all of this Feher stuff, and advising him. We talked about various aspects of it. Finally I came to the conclusion that if conventional transport theory and multiple scattering theory and so on were true, two things would be true. One was that George Feher’s samples would be metallic, and the other is that he would have very rapid spin diffusion. Another aspect of quantum transport was Bloembergen’s theory of spin diffusion, another theory, kind of a model of calculation. Rough estimate, how spin diffusion takes place in any large system. And we found indeed that you could demonstrate the existence of diffusion in all kinds of systems. Anyhow —
Hold this one, and that one looks superb, and this must be a backup.
OK, fine...Well, all of this is a long preamble to the fact that although I wasn’t publishing anything else, I was thinking hard about George Feher’s experiments, and was gradually evolving the idea of localization, which was my contribution to quantum transport, which — as Larry Walker called it, is quantum “cisport”. Transport in the opposite direction. That was what I was busy doing during all the time that everyone else was having such a wonderful time doing correlation effects and collective excitations and so on. And I guess I was learning a lot of diagram theory and field theory for my own purposes, but they were a different set of purposes. I developed this sort of locator, instead of a propagator formalism, developed this locator formalism and perturbation theory for transport from the other point of view, starting out from the idea that maybe the transport wasn’t happening, and seeing where the perturbation theory diverged, and it was the aegis of my localization paper. There’s been a lot, there’s a lot of history in my localization paper. You don’t really need much more than that. That was what I was doing during this exciting period.
You say there’s a lot of history. You’re referring to other interviews?
Yes, other interviews and all my Nobel stuff, which is excessive. So really we can kind of pass that over, except it does seem to me that independently localization, the whole field of quantum transport, all these developments are something that was building in this period, and took enormous strides, especially the multiple scattering theory, the business of calculating energy levels in random systems, densities of energy levels in random systems. That’s a very important field.
Where can I get a list of all these important, unclassified —
I don’t know.
It may not exist.
I’m not sure that that really exists. Well, Motts’ books, to some extent. Of course, my own metal insulator transitions. It’s — Mott, as we know, is a very weak reed and he has his own point of view towards almost anything. It really wasn’t to burgeon until much later on, you know, starting with the 1978 paper, when things really began to happen and we really began to understand quantum transport, and from the localization point of view, and even from the interactions point of view. But the beginnings are way back there when we were worrying about impurity bands. And for a long time, we didn’t understand anything, even when it was localized. I mean, it stopped for ten years. There was also the Mott transition and the Mott phenomenon. That didn’t stop. That went off in different directions. At the same time that we were thinking about, that I was thinking about localization, the reason why I did not in my paper really, focus on any particular physical outcome. My own instincts said, there are so many materials that are insulators that ought to be conductors, and localization has something to do with it. But at the same time, there is the Mott phenomenon, the question of, when does conduction start in the presence of interactions. And there’s a long long complicated history of that. The idea was Peierls. Mott picked it up quickly. Landau repeated the same idea, under different circumstances. Tyablikov had similar ideas in Russia, and finally Mott formalized it in the ‘49 and ‘56 papers, and I’m sure that’s covered in some other part of the, Well…, maybe it’s not covered Motts’ two papers in ‘49 and ‘56 are the key things. The same kind of jump — well, let me finish off the semiconductor part. We didn’t really know whether Feher’s materials were not conducting because, when they were not conducting, he did have a metal insulator dividing line, and he could get it up to about a factor of 2 level of the correct critical concentration, but not better. So he didn’t have precision measurements that he could see, in his resonance work and in everything else, where a substance for example was conducting and when it was insulating. Where the spins were moving around and where they weren’t, how the spectrum looked, whether it looked metallic or it looked like a superposition of different lines. And homogeneous versus inhomogeneous broadening was another aspect. We didn’t know at that time whether this was a matter of the Mott transition, basically the phosphorus energy levels in the silicon being, not conducting because the electron can’t jump from one phosphorus to the next, because of the electron that’s already there, or because of the difference in energy levels, due to the random distribution. And we now know the answer to both, that two the things are happening at the same time.
That’s the result in the ‘78 paper?
That’s the result of work in the seventies and eighties. So this is something that’s happening 20 years ahead of its time, and the only reason it’s happening is because Feher’s measurements were 20 years ahead of their time. They were very very beautiful and very precise measurements, that really haven’t been equaled, or hadn’t been until a relatively few years ago. But this got us all interested in the Mott transition. Nobody kind of wrote a paper about it except maybe Mott. Lax and I certainly discussed it. Maybe even Lax wrote it up. Abraham and David and I were working together that summer. We were working together that summer, and thinking about these problems. And I’m not sure that anything but a first draft exists from that. But one [there’s a Pines discussion ?] way or another, we were all aware that there was a problem of the Mott transition possibly taking place and the Anderson transition taking place and we didn’t know which was happening, except that we didn’t necessarily believe in the Anderson transition. We did believe in the Mott transition. So, but we also knew that the silicon energy levels were magnetic, and that indicated that there was some repulsion between, there certainly was interelectronic repulsion, and so the Mott explanation was surely there. I think there was a certain amount of work, kinetic work by Mel Lax that was related to our considerations, but I’m not sure that anyone ever wrote even the dilemma down clearly at that time. The other thing that happened was, I wrote this peculiar paper about absences and diffusion, and that it stuck in my mind about the Mott transition and the Mott phenomenon.
Did Mott work a lot on this himself?
Oh yes. The other thing which happened at a similar time, maybe earlier, was again, in my consulting guise, I worked a lot with Frank Morin who was interested in magnetic oxides, and one thing he saw and published at that time was these gigantic metal insulator transitions that took place in the magnetic oxides, as a function of temperature in most cases. And again, just as a consultant, I said, “Frank, those are Mott transitions.” And it was only much much later that they were indicated to be Mott transitions. One of them, one or two of them. Some of them were in fact not Mott transitions, but they all were — this was again the beginning of a field that was going to be something that happened, really happened much later, the question of metal insulator transitions in oxides. Jumping over superconductivity, what I then want to say is that a consequence of thinking about the Mott transition was that the thing that I’d been puzzling about for eight years, two years later, suddenly became clear, which was how super-exchange really worked. Seitz had asked me to write up exchange integrals as a review paper for the Seitz-Turnbull series, maybe already in ‘53, ‘54, and I said, “Sure, fine,” accepted the responsibility, and the more I tried to write it, the less I liked it. It was like your history. There’s a time when you can write it, and there’s a time when you can’t, and this was a time when I couldn’t. I realized, after a while, — I didn’t realize until it happened to me — that the reason was that I hadn’t really understood what was going on. What?
How terrible. How awful when that happens.
Yes, it really is. But you can’t write, you know, you know you can’t, if you don’t have any systematic feeling about what’s going on.
It’s happened to me in some other history.
But those fortunately I could postpone.
Yes. And so, in the summer of ‘58, I kind of first began to realize that antiferromagnetic exchange is an automatic response to being in a Mott magnetic insulator, and that super-exchange and antiferromagnetic exchange phenomena and the Mott insulator are the two sides of the same coin, and did the super-exchange paper, with some additions by, fortunately having been out at Berkeley and listened to Leslie Orgel talk about crystal field theory, which I had always thought was kind of uninteresting and just a detail until that time, and then I realized, far from it, it was a very nice way of understanding the relationships among the different oxides and normalizing why they behaved the way they did. So putting some crystal field together, some ligand field type crystal field theory together with the Mott phenomenon, I was able to do super-exchange for the magnetic insulators, and to my mind, that’s a very important paper too. Of course it’s important these days, because it’s behind cuptrates but it’s also important — well, it’s important because it’s one of the early — well, I guess I was taking the spirit of the times, which was quasi-particles and renormalization, and saying, how can I think of these spins and electrons, in the states in the magnetic insulator as re-normalized quasi particles? Instead of thinking wave function A talks to wave function B talks to wave function C, and then they super-exchange, think wave function A re-normalizes and appropriates a little of wave function B. Wave function C also incorporates a little of wave function B, and then they react. And so it was an attempt to do for insulators what Landau had done for, at least for Mott insulators, what Landau had done for metals, to make a formal re-normalization that incorporated the, all the — well, all orders of the perturbative effects of the interactions, in the definition of the entities you’re talking about, and then ask, what kinds of interactions are left over that I can’t incorporate? And if you use the Mott insulator idea, and that concept, you arrive at super-exchange, and that, believe it or not, is also the right way though people don’t believe it is also the right way to do high-TC superconductors to start from the re-normalized particles and ask what interactions are left over. So that, along with the funny old problem of the antiferromagnetic ground state, that old problem of, how can you deal with insulators in terms of re-normalized quasi-particles is modern as hell, though it began then.
What are we up to now?
That was published in ‘59, and written in ‘58.
Somehow both of us have...the documents.
That, let’s see — oh, no.
Have you gone through my list, because you’re free associating.
As I said, all of this is mostly kind of house theorist stuff, in ‘55, ‘56. Oops. I’ve got, oh here. This is this business of, we’re in ‘57, again the house theorist stuff.
We have some earlier cyclotron resonance also.
Yes, that was kind of house theorist stuff. That also was unsuccessful. And also led to ructions with Charlie Kittel which are better left unsaid. Then I jumped from an absence of diffusion, which is ‘58, to super-exchange interactions, which was ‘59, over three papers in superconductivity, and one paper in ferroelectricity, the reason being that the fact that I didn’t go further with that had to do with two things, one was the Mott transition, and that we didn’t really know that localization was causing it, and the other was that I got interested in superconductivity. But it is a natural to jump to really two new concepts in the magnetic transition metals and to the antiferromagnetic contribution to polarization of free electrons in the local moment paper, all of which came from this general development of thinking about the magnetic state as a distinct state in which the Coulomb repulsive interactions are the vital and important part, and in which you’re trying to do the best you can to diagonalize that away, that, plus the kinetic energy, and keep the rest of the interactions properly. So all those are kind of this program. I say, in the magnetic metals, there is this gigantic Coulomb repulsion, which is the Mott term; there’s kinetic energy, in an insulator you can get rid of that entirely because it’s an insulator. In the metal you get rid of it by forming a resonance. And then finally you do the rest of the interactions. So that’s kind of, it’s that concept, and the second concept is the concept of the magnetic state, which is kind of a state which is dominated by Coulomb repulsion interactions. It always annoyed me a little bit that Mott didn’t really accept that for quite a while. That the magnetic criterion for the Mott state, which is I think very important, but — that’s that idea, by that time, we’re jumping past 1960, and you probably want to go back to superconductivity.
Superconductivity is in a sense treated in your paper, which is — Anderson — yes, yes.
I don’t know if you want to go beyond it. Maybe so. It is very superficial in there.
I guess the thing I mostly want to say, if I want to say anything, is that it followed, what I did to begin with followed from the antiferromagnetic ground state.
That’s certainly not in there.
That’s not in there.
And that basically the insight I had was that the antiferromagnetic ground state business, that the Goldstone-boson business, was the key to this problem of gauge invariance.
That’s what got me interested in it. I was puzzled. Wentzel? I guess —
You had not been working on superconductivity at all.
No, I hadn’t been. I’d been keeping interested. I became very interested — well, let’s see, I guess I did one, I didn’t do one thing, a group of us did one thing. In the course of talking about the BCS theory, Larry Walker and Harry Suhl and myself, and I think it was mostly Harry, developed this pseudo spin way of thinking about the BCS theory, which I guess, I did later publish in my papers, and it’s best publicized really by the fact that Josephson used it in his Nobel lecture. It was kind of a — well, it was just an algebraic way of thinking about it, and then —
— which was when?
Oh, we went and listened to Pines. I guess none of the three heroes actually talked to Princeton or Bell Labs about it. They were too busy. David came to Princeton and gave a colloquium about the BCS theory, after, and we all — I, you know, the minute I heard the theory, I knew it was right. Fortunately, I think I, simply through being uncommitted, you know, looking for an outside view with no prejudice, you could see that prior to this there hadn’t been anything and now there was something.
There’s a lot of negative response.
There was a lot of negative response.
Yes, that’s right. That’s right.
But you were not negative.
Not at all negative. No way.
Did you have a feeling that gee, I wish I’d worked on that?
Oh, yes. Yes. Well, I’d heard the Cooper paper, and I’d been very dubious about the Cooper paper. I was negative, as a matter of fact. And then I thought, this must be it. And I was just kicking myself for not having picked up Leon and shaken him and said, “Tell me more, tell me more.” And I’d been thinking, you know, there has to be another parameter. There has to be some way you can put this within the context of a new field theory. And none of us had seen how to do it. And Larry then, and Harry, I think we all drove down together, and I think he worked it out in his head as we were driving, or we worked it out through talking back and forth.
Suhl. Then it made us see that this was really basically a magnetic-like problem. There was magnetization in the X direction, and you had this with, attractive interactions that produced a...spin in the X direction — self-consistent field in the X direction. And there were a few little things that I had begun realizing, like that the problem of projection was something about averaging over phase. (?) ... (off tape) All right. Well, you know, that’s just, as Landau used to say, I’m writing my autobiography, forget that. Anyhow, I had been listening to Wentzel complaining about the BCS theory, listening to people at meetings complaining about gauge invariance. It was a favorite topic of Froelich’s, I don’t remember why Froelich was around at that time but somehow he was, and I realized that this gauge invariance problem we were running into was probably just exactly this old problem of, how can you have an order parameter which isn’t a constant, an order parameter which is not obvious and not initially a physical variable, not manifestly a physical variable? And in particular, not a constant on the motion, and that you’re going to run into trouble with these rules and all kinds of other things, unless you take the collective excitations into account, and it’s not as obvious how it goes in this case. After going to Bernd Matthias and being totally squashed by saying all that nonsense basically, I spent a long time in the library learning about gauge invariance. My first manuscript, incidentally — I knew, so little about gauge invariane that I spelled it “guage” throughout the first manuscript. But I went to Buckingham, who had done some papers, all the people who had been involved in superconductors, and sorted it out more or less in my mind according to these principles. Then I realized that in fact there weren’t any collective excitations here at all anyhow because they all got mixed up with the plasmon excitations and got wiped out. And I wrote it out in that way, using ideas from Feynman and from various other things, in what is really pretty much a cobbled together mishmash of a paper, and in the meantime, as that was being published, I began to think how really to do this thing, with these spin variables that Harry and I had learned, or so I gave credit for them in the paper to Harry, as a kind of general algebraic way of handling the thing. And then got a way of doing it which I called the generalized equation of motion method, and that seems to have solved the problem all right, at least it seemed to get the right answers for excitations and response functions and so on. And I guess nobody after that, after similar papers which didn’t have the proper treatment of plasmons by Rickayzen and Nambu, that people didn’t worry about gauge invariance. The generalized equation or motion method was my attempt to do something that I didn’t realize had already been done, which was the Green’s function theory that the Russians were using. It’s not as good.
Which is not as good?
My theory is not as good. It’s solving ad hoc problems as you get there. It was based on some Bohm-Pines work in the fifties, where they too didn’t get really to the proper Green’s function approach.
How come you didn’t know? Lack of communication there?
The cold war problem?
Partly the cold war problem. Partly — but we did have Montroll and Ward and at the same time Kadanoff came, Martin and Schwinger had done a Green’s function theory, of the superconductor problem. I just had a thing against formalism. I never was intentionally a formalist. And I really believe that, you know, just doing equations of motion of the operators was just as interesting as doing equations where you are introducing some fictitious source functions and so on. And I didn’t realize that introducing the fictitious source function was a useful thing to do, and perhaps I should have. But it wasn’t until much later that I forced myself to learn the Green function techniques. I’d gotten along without them until then, and I didn’t like them, and didn’t like formalism, if I could get away without it, so I was trying to find a cheap way to do it. I think I eventually produced a generalized equation of motion method that was more or less equivalent to the Green’s function method, but if you’ve got the Green’s function method, you might as well use it. However, the Green function method can very often conceal mistakes, physical mistakes which you’re making, as they have done again and again and again, in some of the magnetic problems. So I’m not overly thrilled by Green’s function method; it’s just that for this problem they happen to be an absolutely clean way of doing it. There wasn’t actually anything I wanted to know that I couldn’t find out the way I was doing it, so that it is fair enough to say that I didn’t really need it for what I was doing. Later on, I could have used the Greens function formalism, and eventually I forced myself to learn it. Well, that was what happened on that facet of superconductivity. I really just got interested in superconductivity. George Feher got quite annoyed at this, because I’d been such a useful house theorist, and I think, he didn’t use it as an excuse, he kind of joked with me a bit about it, and he decided to leave for La Jolla anyhow. But in fact, I was known as the house theorist for Bernd Matthias, which was the last thing in the world he or I wanted. From then on, I think the thing you already have has most of my superconductivity work in it. That’s right. Dirty superconductors. That was actually useful. That’s ‘59. It came out in ‘60. There are three things. As I said, we were all very busy. It was all done in the summer of ‘58, mostly in the two months when I was with Charlie Kittel, the super-exchange paper, I completed the random phase approximation paper, that spring, but did the last details in California, and the whole super-exchange paper was done in California. The theory of dirty superconductors just came out of conversation, a kind of a walking to lunch conversation with Jim Phillips. Jim Phillips happened to be out at Berkeley that summer too. This is useful in superconductivity, it’s been a useful idea, and introduces the kind of a concept of dirty versus clean limits and so on. But it’s also useful because it has a method in it, namely the “representation,” which is this representation in terms of scattered wave functions instead of in terms of plane waves, which is a key to a lot of things, actually. It’s a key to a lot of the theory of metals as well, theory of impure metals. And it’s the kind of thing I use all the time, in our theory of metals. I picked it up later, and it’s the underlying basis of modern Abrahams-Anderson type of localization theory. Then there was helium 3, does that fit?
Did I talk about that at all?
Helium 3. People have always explained that the history of helium 3 has to be the history of the nuclear program. In fact the history of cryogenics has to do with the history of nuclear power. Just the amount of helium 3 available at any given time controls the temperatures and the techniques that people have been using in cryogenics, and the fact that the hydrogen bomb program started whenever it did start, ‘54, means that by about 1960, we were worried about He-3 things. It’s kind of a Marxist interpretation of scientific progress.
Well, isn’t it true that helium 3 did become available?
Yes, it became available
Because of the bomb?
Because of the hydrogen bomb. And then we were able to do further things with helium 3 by 1970 because there was a lot more available and so on.
But it’s true, isn’t it?
Yes, that’s true.
The Marxist program.
The Marxist program. But in fact, you know, the Landau-Fermi liquid helium theory, you know, Landau didn’t know that there was going to be helium 3 available or pay any attention to that. Nonetheless, I got interested —
It did come out about that time.
Yes, it did come out about that time. I got this crazy idea that — well, it wasn’t a crazy idea, it came from a lot of other people. It turns out that there’s a sentence in Thoules’s book about anisotropic. BCS states. There was a funny, long preprint by John Fisher and there may have been several other suggestions. Actually, let me tell the story. We had Keith Brueckner as one of our many visitors at the laboratories. And I interacted with him about two things. One was, I said to Keith, “You know, it seems to me that nucleons have attractive forces between them, that there ought to be some indication of BCS in nuclear matter. Tell me about it.” He said, “Oh, no, no. Nonsense.” And about the same time, probably the same instant, David Pines was in Copenhagen asking his friend Bohr, “BCS must have some application to nucleii. Tell me about it.” And Bohr said “Parity.” And it’s been around since Mayer. And so Bohr, Mottelson and Pines is the paper that introduced BCS into it, and I’ve always thought that it was quite appropriate that Mottelson and Bohr had the Nobel Prize and Brueckner doesn’t. But Brueckner did know about helium 3, and I had — well, while he was in the process of visiting us all, during this two weeks, two or three weeks he spent at Bell Labs, I did tell him about my own [?] isotropic states that I’d put in my notebook, and he said, “I wonder if those would work with helium 3?” I said, “I’ll bet they would,” and we talked about it a bit. The next thing I knew, he sent me a preprint, so that what is it, generalized BCS states in the proposed low temperature phase of liquid helium 3, written by Brueckner and Soda. Now, I had already assigned Morel to think about my anisotropic states and do better with them, try to figure out how they, what they actually did, whether they were superconducting in a field, and Pierre liked that problem, and I wouldn’t ever have objected as strongly to Brueckner if it had been me, but if it’s a student — so I said, “Anderson and Morel, and we’ll tell you how it goes”... So I marked up the paper, and said, “This is now a paper by Brueckner, Soda, Anderson and Morel.” It came out considerably before actually Emery and Sessler did something like the same idea. But neither Bruecker nor Soda nor Anderson nor Morel had any idea what the meaning of having a BCS state would mean, would be. And in particular, it’s very important from the fundamental point of view, well, how fundamental, because helium 3 is the first instance that really destroyed the Landau criterion. That’s what’s responsible for superconductivity, superfluidity. Because you know the Landau criterion. If there are no excitations with zero velocity or low velocity. In helium 3, if you got two nodes, and you always have at least two nodes in the wave function, you always have zero velocity excitations. So the question really is, does it have a superfluid response to external fields? Not, is it, does it have zero velocity excitations. And it’s the fact of having a superfluid response has to do with having an order parameter, having a complex order parameter, not with what the excitation spectrum is. So basically we were — well, what I had asked Pierre to do was to convince me, and he did, that these really were superfluid, and give some idea of — for one thing, we calculated the superfluid density, and showed that it was anisotropic. We thought about the question of why orbital anglular momentum, and we came out with, there were three possible answers and none of them is unequivocally right. It’s very interesting. Actually, it turned out to be one of the ultimate interesting problems, is what is the orbital angular momentum of helium 3? But basically, we made the correct physics. We made a numerical error, in the sense that we didn’t take all three components of the spin into account in our states. We were actually right in two states. And we saw this fact that there could be, for any given angular momentum, a whole spectrum of states, a whole spectrum of different kinds of phases, which is not at all obvious, and doesn’t really occur in other collective systems. And also that the angular momentum would be a quasi-classical tainted(?) in fact for you would have states that are all (?)= 1 or all (?)= 2 or all (?)= 0 or whatever it is, not states which are some mixture of either the different substates or the different angular momenta. So it was kind of — I had a feeling that that was the point at which one really began to understand what the nature of a complex order parameter was, and what caused superfluidity. So it wasn’t an academic exercise, even though it was an academic exercise, because it was a matter of defining in principle what superfluidity is and what causes it, and what Landau had not understood and certainly Bardeen never understood, and I think by the time we got through with it, we did understand. The other thing is, by a miracle, it happened eventually. It happened in one of the states that we had proposed, even though we’d proposed it for the wrong reasons. So in 1960, it had really happened the summer of ‘59, when Brueckner and I talked together, and then David Pines had left Pierre Morel because he had been sent off to Illinois by Princeton physics department, and couldn’t, and Pierre had to stay because he was the science attache of the French consulate. Pierre was a wonder. He had this beautiful wife that he had apparently won in spite of competition from Philippe and she was extremely high social class and very beautiful in addition, and they were quite fancy in the embassy society. David took advantage of this and went to several parties at the French embassy that I didn’t, and it was fun having Pierre. He later ran the space program in Paris. He was up against too much competition among the French theoretical physicists. He was very good, but he wasn’t a de Dominicis. And he just wasn’t willing to be kind of third or fourth fiddle. So he probably was wise to go into the space program.
Just getting organized here. (inaudible question.)
I do want to get to, but maybe you could do it on the phone with me. If we can get to it today? But I don’t want to cut you short either.
I’m kind of getting to the point where the rest of it — you know, there’s a lot of history, lots of papers. The question is, do you want go past your 1960 cutoff, or do you want to?
It’s a way to sort of — the trouble is, it will be quite a while before someone comes back and gets the rest, and it would be a shame to skimp on that just simply because of my cutoff. We’ll have to skimp anyway because we’re running out of time. Is that... do you think?
Yes, OK. Where’s the off? Here.
This is Lillian Hoddeson. A decision was made off tape to discontinue the interview, and to continue it at a later date, and instead turn during the moments left to Anderson’s specific suggestions for improving the present version of Chapter 8 and that discussion is on a separate tape.