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.
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 Martin Schwarzschild by David DeVorkin and Spencer Weart on 1977 December 16, Niels Bohr Library & Archives, American Institute of Physics, College Park, MD USA, www.aip.org/history-programs/niels-bohr-library/oral-histories/4870-3
For multiple citations, "AIP" is the preferred abbreviation for the location.
Life of his father, Karl Schwarzschild; father's scientific relationships in Göttingen (Felix Klein, David Hilbert); move to Potsdam, 1909; relations with Potsdam and Berlin scientists (Albert Einstein, Karl Sommerfeld); father's Jewish background concealed. M. Schwarzschild's youth in Göttingen and Berlin; early education, interest in astronomy and mathematics. Undergraduate at Göttingen Universität (Hans Kienle, Richard Courant, Neugebauer), 1930-1933; graduate work at Gottingen Observatory, 1933-1935; his reaction to Nazism. Introduction to astrophysics (Arthur Eddington), interest in stellar interiors and stellar evolution; contacts with other astronomers from Gottingen Observatory (Otto Heckmann, Kienle, Rupert Wildt); comments on general relativity; interest in pulsating stars; leaves Göttingen, 1936. Postdoctorate at Oslo (Svein Rosseland); Jan Oort, Ejnar Hertzsprung; mechanical analog computer for computations in astrophysics and celestial mechanics; comments on development of theory of stellar interiors, 1939-1950. To Harvard College Observatory (Harlow Shapley), 1938; C. Payne-Gaposchkin, Bart Bok; comparison of European and American observational style, social scene; Barbara Schwarzschild's difficulties as female astronomer; contacts with S. Chandrasekhar and other astronomers. Tour of the United States; visits Mt. Wilson Observatory (Wilhelm Baade, Rudolph Minkowski, Edwin Hubble, Milton Humason), 1940; Shapley's relationship with Mt. Wilson staff. Harvard (Fred Whipple), 1938-1941; Shapley as a leader; astronomy summer school at Harvard; work on Cepheid variables in M3 (Bok, Chandrasekhar); overall impact on Schwarzschild of Harvard period. Columbia University (Jan Schildt, I. I. Rabi), 1940-1942; difficulties there; origin and funding of Thomas Watson Astronomical Computing Center; discussion of cosmology in the late 1930s; contacts with physicists (Enrico Fermi). In U.S. Army, 1941-1945; enters as private, teaches math to recruits; refuses invitation to Los Alamos; transferred to Aberdeen Proving Ground, dissatisfaction there; to officers training school, does bombing analysis for Italian campaign. Work relating to stellar interiors and evolution, 1938-1946; nuclear energy source ideas (Hans Bethe, Fermi); Eddington, Gerard Kuiper, Chandrasekhar, G. Keller; German astronomers during World War II (Ludwig Biermann). Discussion of wife's career and her role in his career. Early ideas about red giants (Öpik, Herman Bondi, Fred Hoyle), 1946-1950. Work on acoustic wave energy transport (R. Richardson, Gold); work on chemical composition differences in stellar populations. To Princeton University (Spitzer, H. N. Russell), 1947; Project Matterhorn (start of bomb and fusion projects); relationship with Russell. Stellar evolution work in the 1950s; computer work (John Von Neumann, Richard Härm), mid-1950s; collaboration with Allan Sandage evolving a stellar model, 1952; computing towards red giants; observational cluster work, 1951; ages, metallicity, and the Big Bang; beginnings of "astrophysical" cosmology. Evolution theory after late 1950s; effect of computers on theoretical progress; relation of evolution theory to cosmology; general comments on his work in stellar evolution; interactions with Robert Dicke; views on cosmology, general relativity. Need for better solar convection work leads to use of balloons (James Van Allen); post-Sputnik funding; on cooperation with industry and engineers; Stratoscope II (Bob Danielson, Spitzer). Years advising the National Science Foundation, President's Science Advisory Committee, 1959-1976, and National Aeronautics and Space Administration (Von Neumann), to 1969; The International Astronomical Union, 1964-1970; American Asronomical Society, 1967-1973. Informal advisor to various observatories: Kitt Peak National Observatory, Mt. Wilson-Palomar Observatories, Carnegie Southern Observatory. Recent work on galactic structure. Reflects on importance of ethical standards; his feelings about religion and nature.
Last time we got through your work in the war. And normally we would somehow try to discuss in a chronological fashion your post-war career. But we often find that as people get older and get involved in more things, it becomes impossible to handle everything side by side, so we try to take one development and then go back and pick up another one. So one development is your social interactions, building up the department here at Princeton and so forth. Another development is scientific work and particularly stellar evolution. And since David DeVorkin is here with us, we'd like to go over stellar evolution first and cover the whole development of stellar evolution from the late '30's right up close to the present, particularly the really important periods of it. That will probably take us past lunchtime, and then David has an appointment over at the library; and then we can come back and pick up your career here at Princeton and some of these other things, if this seems like a reasonable thing.
From my point of view, of course, the topic of how we got into off-the-ground work was, for my life, a very important one, and that was under Spitzer, isn't that right?
Right, but I don't even want to talk about that yet. I think first we should talk about stellar evolution.
I agree, I meant only as a potential additional topic.
I haven't even talked about all the topics we want to cover. We want to cover your Princeton career, we want to cover the Stratoscope business, we want to ask about what you know about the other things going on in the astronomical community. But I don't think we'll get to all those things today. I'm sure we can get through the stellar evolution. Well, I think a reasonable starting point would be the conference that Tuve held in 1938 — or let's just say all the developments, around 1938: Weizsacker and Bethe's carbon cycle idea. And I'm curious as to where you first heard about these things, where you first encountered these ideas.
I think I saw at least Bethe's article right away the moment it appeared. I'm not quite sure how fast I saw von Weizsacker's, though at that time I think the German literature still came through.
It was still before the war.
So I probably saw it right away then. I do remember an occasion in 1938 when Bethe gave a colloquium either at the Harvard physics department or at MIT — I don't remember — where he presented exactly the nuclear sources of the sun. It was put in that form. For me it was an extremely exciting moment, because I think by that time I was aware that the moment we had had not just the processes but estimates of the rates of the nuclear processes, it was the one major missing link to really make definitive models. How much I realized that it was the opening to stellar evolution I cannot remember. But that it was a key step for new possibilities was clear to me.
And did you feel that Bethe's thing was it? You were convinced that this was in fact…?
Oh, I trusted his analysis and his numbers extremely much. It turned out that his basic analysis of the processes was just plain essentially right. That his rates were off by a factor of a hundred didn't really make much difference. I wouldn't have believed it at that time. I would have been very upset if he had estimated his rates that uncertain. But in effect it got us going to incorporate this new physics into sort of complete stellar models. And the moment you do it once, then you can do it in no time flat a second time with improved rates. We repeated, you know, the solar models over and over again for the next decade every time (we got) better data — not just on the nuclear physics but at the same time on absorption coefficients and other physical items.
Now the chronology gets a little confused here because of course the war intervenes. I'm curious as to what you may have done once Bethe's thing came out: I'm curious about what you may have done before the war and during the war to try to use these ideas.
Actually, if I remember right, I did nothing actively using that before the war. I was I think in the middle of another investigation, I believe on rotating stars, and wanted to go on to finish that. Then I moved to Columbia University, which took some new adjustments, and got acquainted with the first digital computing machines there (which was an important preparation as it turned out, even though I didn't really foresee that, for my future work). Then I was asked to help a little to teach in physics, which was I think mostly to convince me that I should not go into the army, but then Pearl Harbor came and by a previous decision that I had made with myself I did go into the army. So actually I do not think I used Bethe's work at all before the war, before I got into the army; but I was sort of mentally getting ready.
I see. You expected that would be your next thing perhaps?
In any case it was, so I presume that I did.
I see. You can't recall whether there was an anxiety to do it all that time, whether you were eager to get on with it?
No, I cannot remember.
I see, okay.
In 1942 when you had followed up some of Eddington's ideas on the photo-ionization of material in convective regions,  you used cross-sections by Bethe and Fermi and Teller, and the way that you described it in your text, in that paper, it didn't seem to be a reference to a published source. I'm wondering if you had any contact with these people, using their cross-sections?
Yes. That was a consequence of that for me marvelous circumstance, at Columbia before I got into the army, that a whole group of brilliant physicists at Columbia were working already but under self-imposed secrecy on preparation towards the atomic bomb. They used to have lunch together at Columbia and admitted the two astronomers, Schildt and Schwarzschild, or as Fermi called us, the director and directee, to have lunch with them. And since they couldn't talk about their work, astrophysics was quite often the topic. Fermi was just an overwhelmingly marvelous man; if you asked him a question, he knew the answer nearly always. And then he would apply his whole power of personality and intelligence to make sure that you understood the answer. So, in that case I asked Fermi, and he spent just an hour with me to get the basis in my head and permitted me to quote him.
That's quite interesting. Did he work out the cross-sections for you? Or did they have this material…?
I remember I asked him one day, he gave me a tentative answer and then came back a day or two later and told me that he had thought about it and that he wanted to put it a little more cautiously than he had originally said. But even that more cautious statement was sufficient for the particular purpose, in connection with Eddington's pulsation work.
Yes. I'm interested very much in talking about two things right at that point. In following what Spencer had said about your contact or interest in Bethe's work on the energy sources, did you talk about the application of these energy sources with Bethe or those at this round table, or was that too close to the classified question at that point?
I don't remember whether Bethe used to come to Columbia at that time — at least it was very rare — and I do not remember discussions with Bethe then. Indeed, I knew Bethe very little. I had a short discussion with him after his presentation, which was in '38, that is, before I moved to Columbia in 1940. And I cannot remember having any discussion with Bethe during the intermediate years. I did have some discussions with Robert Marshak, who was Bethe's student and then was at Columbia.
At this time there was, of course, primarily the CNO cycle, and then there was the proton-proton, which was suggested as secondary at that time.
They were both suggested simultaneously as important for main sequence stars. I think Bethe understood that completely. Only exactly where the switch was, whether it was above the sun or below the sun on the main sequence, was not yet clear — in part because of quite wrong cross-sections. At the time we thought the sun belongs to the upper main sequence — isn't that right? Therefore, it was relevant if you wanted to ask the question for the sun specifically; for the overall picture it was really quite irrelevant. It just happened that the sun was in the disputed area of transition, and was on the wrong side to begin with.
It just occurred to me from what you were saying, David, to wonder whether at this table at Columbia there were any talks about the possible applications of nuclear energy. Here you're talking about the sun and possible fusion and so forth.
We all understood that that was in the generally classified area. I mean I wouldn't have thought of bringing anything of that up, and the physicists themselves of course didn't.
Did you feel that it was relevant? That is, that your discussions about the energy sources in the stars were relevant to future terrestrial energy sources?
I sort of doubt that I had the foresight that fusion rather than fission might become important. I don't think that I had that penetration.
You would have thought of it more in terms of the strict uranium fission.
Yes. And even the question of peaceful uses — I really couldn't penetrate far enough to think in those terms.
This paper on cross-sections was stimulated by Eddington's discussions …
And you certainly had contact by letter with Eddington on this matter. Actually he was the one who presented your paper to the MONTHLY NOTICES or communicated it to the MONTHLY NOTICES.
After Eddington had published his paper, I felt doubtful about it — in part because I had my own picture how this observed phase shift, that he wanted to explain, was to be explained. So I had this discussion with Fermi, and then wrote Eddington a lengthy letter — why I felt that his explanation might not in fact physically work. Eddington then wrote back to me a very friendly letter suggesting that I modify the letter to him as a short paper for the MONTHLY NOTICES.
Do you have these letters, by the way?
I'm practically certain that I have. He did not tell me that by that time he already had another version coming out with the same result but not needing this physics. I was good and angry when I finally saw the MONTHLY NOTICES where my note was immediately followed by Eddington, accepting my paper but coming out with the same results by a different physics. I would have felt it would have been more honest among colleagues to tell me. It was a mini-version of exactly what he did in an extremely unfair way to Chandrasekhar in connection with the white dwarf papers, where he did not tell Chandrasekhar that he would follow, not just in the MONTHLY NOTICES but actually in the oral presentation.
Yes, Chandra told us about that. In this particular case Eddington may not indeed have started work on this until after he had seen your letter?
I do not know anything, but I would be somewhat surprised that he would have had a reason to work on the second version before he had this key information, which really came from Fermi through me.
I understand. Well, that does go with other things we've heard about Eddington. Let's see: anything else about the wartime period?
Well, there are a number of themes that we'd like to carry through. Possibly we should go back to '38 again and talk about the observational side and your contacts with Kuiper at Yerkes. You mentioned in a previous session that you had some very nice times with the Kuipers at Yerkes, and this was during the period when he was synthesizing HR diagrams of clusters, putting together Trumpler's work and Stromgren's, if I have it correct. Did you talk at all about stellar evolution with Kuiper from the observational side?
I was extremely kindly brought into the Yerkes group starting by an invitation from Kuiper to spend essentially a Christmas week at their house. I found that it was a marvelous period of Kuiper and Chandrasekhar working together. Kuiper understood the needs of the stellar interior well enough to know what were the relevant data, but I'm sure that Chandra's influence was extremely strong on that work of Kuiper's too. It was a short but extremely brilliant cooperation by an essentially pure observer and a pure theoretician. I just sort of had the benefit from both sides. Actually, I did personally become much more attached to Chandrasekhar because I was also on the theoretical side. On all my subsequent visits, which were fairly regular, I stayed with the Chandrasekhars.
Did you see specifically, or did he show you, any of the diagrams that he was working up of clusters of equal hydrogen content?
Oh, yes. That was a big topic of discussion. And Stromgren was there, not all the years but quite a number of the times that I visited there, and he of course was working very much in that same field — so that the beginning of the new interpretation of the Hertzsprung-Russell diagram was right there in Yerkes. I had extremely good luck; for me it was a very strong influence. Indeed, I remember getting a terrible scolding from Chandra that I had not already read Stromgren's papers, for example, and he gave me just homework — reading that I was to do at night to be prepared for the next morning's discussion.
Did looking at those diagrams and the way that they presented them allow you to think in terms of expansion off the main sequence?
Did you think at all about evolution at that time?
The question of evolution was clearly there. That is, that the Hertzsprung-Russell diagram did have an evolution element, but also a mass difference element, in it was clear at that time. But it was fantastically muddled up — which of these two key parameters did what in the Hertzsprung-Russell diagram. For example, Kuiper's curves were, from the modern point of view, quite wrong, but that took a long while and well until after the war to straighten out.
He never thought of the actual curves themselves, of the clusters representing evolutionary paths?
As far as I know, I don't think so. At that time I think the first step was to try to get the masses and abundances straight rather than the ages straight. That really became more than a speculation only after Baade's population separations occurred during the war.
Okay, I don't see how far I can go with that, but certainly when they were working with clusters they knew that they had a collection of stars of the same age, and so that parameter was independent; and yet they saw that their composition lines, of equal hydrogen content at least, fit so well the clusters themselves — Did they think that there was something that was correlated between age and hydrogen abundance? I mean the clues seem to be there to the populations. Or was it just too muddled up, as you mentioned?
It was; because, you see, all that work did not yet include the nuclear rates, which meant one very important condition was not yet there. All the models on which these discussions were based didn't have any helium, and much too much heavy elements, in them. Therefore the discussions were numerically completely wrong.
I see. So there was no speculative discussion that you overheard or engaged in dealing with other possible evolutionary paths on the diagram
Not that I remember.
Okay, that's the important thing.
And then there's the interesting question: at what point did you begin to think that perhaps red giants come off the main sequence?
I think it was later during the war that possibly Gamow — but I do not know exactly who had that idea first — suspected that the evolution should lead from the main sequence into the red giant region.
Would you have had any contact with Gamow?
In what way? Where would you have had contact with him during the war?
Oh, various. I just cannot remember well enough whether the beginning of that wasn't already before I went into the Army, or more likely immediately after I came out of the Army.
But even when you were in the Army you would have had contact with people like Gamow and Chandrasekhar?
Chandrasekhar yes, because we overlapped in Aberdeen for a while. Anyhow we were close friends and exchanged letters, which was not the case with Gamow. I enjoyed Gamow enormously but not in that sense.
There was a paper by Chandrasekhar and who was it in 1942…
Chandrasekhar and Schoenberg in '42 where they discussed red giants in this context as expanding.
Well, they discussed the limiting mass of the white dwarf in that particular paper.
No, excuse me. The Schoenberg-Chandrasekhar paper — at least the famous Schoenberg-Chandrasekhar paper — was showing that a limit gets reached in the main sequence phase, that something must happen after a certain percentage, roughly 15%, of the inner hydrogen is transmuted to helium. And that led to a fairly violent controversy with Gamow, who maintained that that limit occurred only because Chandrasekhar and his students had arbitrarily assumed non-degeneracy. That word "arbitrarily" or some equivalent adjective angered Chandrasekhar, who is a very sensitive person, very much. For Gamow it meant nothing. Gamow was somewhat rough — basically friendly but quite a rough person. So it is one of these examples where differences in character can come to clashes that really are scientifically quite unwarranted but are unavoidable since we are all humans. Indeed, they fought so hard for a short while that after I came back from the army, a young colleague of Gamow's, who had during the war worked with Gamow on some of these questions, Geoffrey Keller, came to Columbia and wanted to do a thesis with me on the evolution into the red giant branch, which by that time we were fairly convinced of, and I told him when two giants clashed, we dwarfs stay out for a while. (laughter)
That's marvelous. Did he come and study with you, after your statement?
Yes. Well, actually already before the war he had finished all the requirements but the thesis in the physics department, — but had gotten so much interested in astrophysics through the contact during the war with Gamow that he wanted to do his thesis with me in astrophysics, and the physics department agreed. I persuaded him to work on the sun, and he did the first thorough model, using the then slightly updated rates of the carbon cycle to produce a solar model which had helium in it for the first time. To fulfill all the conditions, one was forced to put helium into it.
Was this before or after you had done a solar model yourself using the Russell mixture?
I think it was after.
I think it was published '46, on the helium content of the sun, where you got a moderate helium content — a 25 to 50% helium content.  You were still at Columbia then.
That is right. But that paper that I wrote was a very short one and actually incidental to the rotation work I was still finishing.
Oh, is that so? So the rotation work you had just dropped, and you picked it up immediately at that point when you came back from the army?
Yes. Yes, actually I started it the last month in the army when I wasn't given any assignment. I picked it up again. And that (solar model) paper was sort of incidental, but I sent it in. Then I found that I hadn't caught up with all the literature that had come out during the war, and that some of my basic numbers — input numbers — were far outdated. I was then just married, and asked my wife whether she would agree that I would pay for the print setting of the paper because I wanted to withdraw it. She, cooperative in astronomy as always, said, "Yes." I wrote of Chandra to withdraw the paper and Chandra wrote a very short letter back and said, "It's irrelevant whether the numbers are accurate. The basic method you applied and the clear exposition make the paper worthwhile. I am printing it."
But that paper of mine was numerically so behind, and so far from what one could do at that time, that Keller picked it as his thesis and did a thorough model.
What do you think was the effect of the war in general on the progress of the field? Do you think that the war delayed the applications of this kind of knowledge or that it accelerated it because of all the nuclear work that was done?
In astronomy I think there was a definite delay, because the nuclear work needed for the nuclear bomb I think was quite unapplicable for a long time. So I think the fact that most of us lost something like three years in straight working delayed it by roughly that much.
So you didn't have too much catching up to do.
No, but some work was going on, and quite particularly, illogical as it may seem, the German astronomers were taken into the army and after one year dismissed, told to go home and do their astronomy, for reasons that I've never understood. But as a matter of fact, Biermann and all his group essentially lost only one year. Except of course at the end of the war and the first post-war years they worked under extremely hard practical conditions, so that in total they did lose more than just that one year.
So to some extent you had to catch up with them.
Yes. And in fact it was Biermann's group who had been most active — some of them.
I see. Another question that I wanted to ask you, since it came up, was about your wife. You mentioned that you had met her back at Harvard, I believe, but how you came to marry her and how your careers have meshed.
My wife majored in Radcliffe, then had one year of graduate studies, very largely under Bart Bok at Harvard; and then felt very uncertain whether she was really — made to be a researcher.
You mentioned this in the earlier interview, yes.
— and decided to go on to teaching. But we had met, and I had very early fallen very much in love with her. But with my proper German upbringing I wasn't supposed to even date a girl until I had a position, and a fellowship was not a position. But Columbia was a position. But my persuasive powers were not much better than those of my father — so it also took us quite a while to both be convinced. During my army years the two or three leaves that I did have I was always invited to stay at Barbara's parents' house, and our contacts clearly became closer and closer; and in fact we corresponded absolutely regularly, weekly, all through my army time. She was then working at the Radiation Lab at MIT on radar equipment. During the leave of absence that one gets automatically the moment one comes back from overseas we decided in fact to get engaged, and married two weeks later.
As you look back over the time since then, I wonder: how do you think the fact that you are scientists has affected your marriage?
If you combine the fact that Barbara has this scientific background with her characteristic reserve that she was not necessarily eager to have an independent career — that combination has affected my work unbelievably strongly, because it has given me a chance to have her support in so many ways. Doing a lot of things that I should do, like a fair portion of our personal correspondence, of which I do practically none; or even taking care of our financial things. Plus her participation — very active and really critical — in all our observational undertakings. That is, first, in all the trips to Mt. Wilson, she was a full member of the observational research that we did there.
In the late '40s, starting in the late '40s. You were a guest investigator for a year, or a summer.
Our first trip was still from Columbia, paid from our own pocket because Columbia didn't pay, after which we were flat broke. And from then on regularly here from Princeton; it was a key arrangement that Dr. Spitzer had made with the University, that both he and I could be absent one quarter of the time for observational work. Barbara was never paid in any way, directly, but we always did go together.
So she served as an observing collaborator, for the love of it, so to speak.
Right. And in Mt. Wilson actually, officially she couldn't exist, because female astronomers were not permitted on the mountain. But by simply not asking the question, quite openly she — was there. She was there at night and everybody could see it. To make it extreme, during the midnight lunch hour normally the 100-inch and the 60-inch used to be closed for an hour, because the night assistant and the astronomers would come together and have lunch. Barbara was clearly not permitted to participate in the midnight lunch. And I hated the idea of losing an hour, so she did run whatever instrument we were on — later on the 100-inch — without a night assistant, for an hour, and everybody knew that. It was completely against the rules.
But nobody said anything.
As long as she didn't make any mistakes; and she didn't, because she is extremely capable with instruments. She quietly worked.
Where did she stay? Where did she sleep?
That was another funny situation. There is exactly one cottage at the observatory. The Kapteyns' cottage, which was built by Hale from his own funds when Kapteyn, before World War I, regularly came to Mt. Wilson in the summer and very much wanted his wife to look after him and cook for him. Mrs. Kapteyn must have been a very marvelous and lively person. And after the first visit, when they stayed in a tent and Mrs. Kapteyn was seen to mostly spend her days sweeping the ants out of the tent, Hale built the Kapteyns' Cottage. And it was at our time the only place where women could stay. Normally it was used more for staff members to bring their families up. But they let us reserve it whenever we had an observing run. The very first time we went up, Baade (who was really my tutor in Mt. Wilson as far as observations were concerned) drove us up, drove to Kapteyns' Cottage and said to Barbara, "Mrs. Schwarzschild, this is where you stay. Dr. Schwarzschild stays in the Monastery." We were completely shocked. Barbara with her handbag stepped out; we had no idea when we would see each other again. I did not have the courage to make a comment at that time because the opportunity of observing loomed so large. I went to the Monastery with Baade. During lunch Baade and I found that there was really a very long list of tests we wanted to do that afternoon before the night. I suggested that Barbara was trained in darkroom work and since they were only tests, shouldn't we call her to speed up the procedure. That was the entrance of Barbara into the door; and from then on she just stayed there. It had the immediate funny consequence that at the end of the night I didn't know whether it was proper for me to sleep at Kapteyn's Cottage or in the dormitory. Well, I did choose Kapteyn's Cottage.
What about on Palomar? Have you both been on Palomar?
Not for observing, no. That really came into operation only just a little before the end of our Mt. Wilson visits before we switched to Stratoscope. All during the Stratoscope period, Barbara participated in absolutely every flight as a crew member. So to come back to your basic question, the fact of Barbara's training, and also her characteristics, plus the fact that, very much against our wishes it turned out we could not have children, all meant that without her I would not have achieved nearly as much.
What about in your theoretical work? Has she played any role there?
No. On the theoretical side her background, she felt, was not strong enough. I'm not sure that her assessment is quite right. But that was her choice.
Very interesting. It's quite an unusual relationship I suppose. I don't know of other cases like it.
I think the only couple where both the personal and scientific relations were quite that close and somewhat of the same character were Bart Bok and Priscilla Bok. Some differences in temperament, but very many similarities.
I see. We'll have to ask him about that.
Priscilla Bok died recently, so if you talk with him you have to keep that in mind.
Yes, it's not as easy a question. Well, I suppose we should get back to the red giants.
Yes, in a way we could lead into part of it with the work that you did with your wife in 1950, but I'd like to go back a little earlier than that. That would be on the abundance differences.
Yes, let's start with the post-war work.
Okay. You started working with Robert Richardson and a few others at Mt. Wilson, and you started generating some models for red giants very early on. You were also still finishing up the work on rotation and trying to examine the role of mixing, or how much mixing rotation would stimulate. Were any of these lines of inquiry intermixed in your mind?
The question of mixing by rotation was recognized very fast by quite a number of us, not at all just me. Because it was very clear that the evolution path in the Hertzsprung-Russell diagram would look very different for mixed or nonmixed stars; whether the helium produced in the core would stay there or would mix through makes all the difference, and that was very early recognized. I could even imagine that Gamow was the first to point that out. So the investigation of rotation, though an interesting problem in itself, had that relevance to the main stream of evolution problems. The result (not just from my work at all) was that for the majority of the stars the classical Eddington currents caused by rotation would not be sufficient to do any mixing. So we then switched to the status of forgetting mixing altogether — which now is again being to some degree questioned in consequence of the neutrino problem. But for a very long while we didn't mix except in connective zones. But that should not be attributed to me. That was a general development.
But when stratified models came out starting in '38 and continued on, you began thinking in terms of zones where the stratification with change in chemical composition was not instantaneous but was gradual?
I'm wondering if that was also the thought you had about mixing.
Well, mixing, or as a consequence of partial burning. But it was just a whole period in which we did not yet try to follow an individual star in its evolution in detail. We first felt that we ought to have some models, example models, that really gave red giants — isn't that right? After being quite capable of building main sequence models, the question of whether in fact if one permitted oneself fairly arbitrarily the consequences of the burning, one could produce red giants as observed, was the next point. There was no point trying to make very detailed evolution calculations if you didn't really know how a red giant was potentially constructed.
So there was still some question as to whether this was really what a red giant was.
There sure was. And a couple of the investigations I did then were exactly of that kind, to make any red giant that was plausible, as an evolution consequence. And there were lively attempts at other places. There were the early attempts between Gamow and Chandasekhar. Then there was a slight hiatus when we were scared of the subject, and both of those two giant characters stopped that work. But then people — for example, a student of Bengt Stromgren's, and myself with some students, and also Bondi from Hoyle's school entered that field, and there were lively but generally quite happy disputes. One of the difficulties that I was much troubled about was that the red giant models that we all managed to be able to construct did not seem to want to fulfill the mass-luminosity law of that time I remember one very lively occasion of a discussion with Bondi when I pointed out that his red giants did not follow the measured masses for their luminosity. He had actually not realized that there were mass measurements. Because he came from Hoyle's group, which of course in the English tradition came from research, not from reading up on the observational literature.
Was this face-to-face contact you had with him? Where did you have a chance to meet Bondi?
I wonder whether that was in 1950 when Barbara and I for the first time went to Europe? We stopped in England. It could have been as early as that; it could have been a little later. But I saw Bondi a good number of times after that.
By that time in terms of just making red giants, it seems like two different schools. I don't know; I'm interested in hearing what you say about it. One through the transmutation of elements in the core of the star and its expansion to the red giant stage, and the other the accretion of material onto a star by its passage through a cloud or something like that. And that seemed to be very much in the old Eddington tradition which Hoyle and Lyttleton and others might have picked up.
You're entirely right. Hoyle and Lyttleton and then their younger colleagues, Bondi and Gold together, worked on the accretion and made models that had discontinuity in composition — assuming that the accreted material was cleaner, that is, more pure hydrogen — and therefore they did try to construct models that were very similar in type to those constructed under the hypothesis that it was the burning that made the change in composition, and with the same sign. By the way, Eddington never believed in accretion. And by and by the Hoyle group themselves found that to get a sufficient accretion rate was only possible for very rare circumstances. So that reason for getting inhomogeneous stars sort of disappeared, and by the time that, for example, Bondi worked in that field, I think he was already on the side of internal evolution making the inhomogeneity, rather than accretion.
Still the idea did persist through the '40s.
Hoyle does not give up easily. (laughter)
Did you have any direct discussions with them about this, or was it all just sort of reading each other's papers? Did you have buy opportunity to meet them and talk with them about these issues?
I met Hoyle for the first time in 1950 on our first visit to Europe. I made an appointment with Hoyle; Lyttleton came along. We went into a little teahouse in Cambridge, and I remember that first meeting extremely well. By what I was taught to be polite, I asked questions about his work, which he answered with his typical determinacy and immediateness. After his answer he was quiet, so I thought of another question. I don't remember exactly the topics, but there was plenty to talk about, and he gave me an equally definite answer. Finally I got a little annoyed that he would not reciprocate isn't that right? — and ask any questions at all. So after he was through with an answer (maybe we had talked an hour and a half or so in this one-sided style), I decided not to ask any further questions and see what he would do. He did nothing. And that was the polite, but for me rather annoying, end of our first meeting. At the end of that meeting I made up my mind that our styles were too different, but most of our lives were ahead of us and we ought to have much contact. So I thought about what I should do the next time and decided that I would not ask any questions and just talk about my work, but then, obviously, give him time to talk about, his work, too. The next time we met, which was a couple of years later, I did exactly that; and ever since we have been on extremely good working terms. It was just a necessary adjustment of form that I did very consciously.
I certainly want to talk to you and ask you about your work with Hoyle.
But could we finish the Bondi question. When I pointed out to Bondi that his models did not fit the observed mass-luminosity data, in characteristic Cambridge style he said, "Then the data must be wrong." I felt that that was too one-sided an attitude. But as it turned out, he was right. The data for the red giants were essentially two stars, Capella and Zeta Aurigae; Capella was reobserved by Struve, and the mass changed in the right direction and nearly as much as what we needed. Zeta Aurigae was straightened out much later. But after that collapse of the accuracy of the mass of Capella I agreed that Bondi was more right in this particular case than I had believed; and that cleared up very fast our view of the red giants. We, then had sufficient sample models to start going systematically.
You were still working (and other people were, too, of course) with either isothermal cores or with convective cores — CNO cycling or nothing. What were your preferences at that time? What did you feel would be the end result of the work in the late '40s? … as to which of the two types of models?
That is really Main Sequence…
Red giants. Later on a number of people, and I thought you did, too, with Oke and then later on with Sandage — I may have it wrong —
No, you are right. Oke had written, as an undergraduate in Toronto already, a paper that pretty well showed that the sun was running on the proton-proton reaction, not on the carbon cycle. So for the main sequence the adjustment of the switching point came around that time. I don't think Oke repeated that work here. I think Oke and I did the first paper here, where we tried to get systematically into the red giant region from the main sequence. 
That's right, that was about 1952. You used a convective core at that point with Oke. But these are ideas, though, that had developed in the '40s. There were these options to use. You mentioned Stromgren's student. Are you referring to Anders Reiz?
He had used an isothermal core, but other people later on were using cores that had CNO in them. I was wondering, as you were building your models, had you looked at these two very different possibilities, plus the shell sources that Gamow was talking about? Had you thought in what direction you would work when you did come to the stage where you could build models and evolve them from the main sequence?
I cannot remember any details, but I think all of us felt that much of that early work was purely exploratory and therefore had very strong elements of arbitrariness in it. (We felt) that when we would really be able to follow it in detail, we would be led quite uniquely to what was the appropriate state to use. Also, by that time I think we understood that that depended quite a bit on what mass star you used. But also during that time electronic machines were not available, so it was all desk calculating. Therefore you did your utmost to minimize the numerical work, and therefore quite often made approximations which you knew were not the correct ones, but that led you into that class of equations which by transformations reduced the total number of solutions you had to compute by one full parameter — which meant an awful lot. All through those times we had these normalized solutions, (made) possible by using an approximation, just to make the numerical work at all tractable. I think one must very much realize the difference of status of computational facilities and why we, against our better knowledge, quite often chose the one or the other of these approximations, whichever we thought was nearer, knowing that we should be in between.
You make it very clear. A few more questions before we get into the beginning of the work to evolve stars. You mentioned various names of the people who were involved in the early red giant work. You didn't mention Opik. Do you feel that he played a role?
I don't remember whether in the previous session...
You mentioned Opik's pre-war work, but you didn't say anything about Opik's post-war work.
After that famous and completely ignored work of 1938 of Opik's, which is a piece of research with really phenomenal foresight, I cannot remember that he contributed after the war very substantially to that field. Opik is of very great width and has done essential research in many fields. I do not remember that he did contribute (to stellar interiors) after he was resettled in Ireland.
I see. So it was essentially the '38 paper that you were referring to.
Right. The one that we all ignored.
You knew about it but you ignored it?
In the previous sessions you indicated that this might have had something to do with Opik's personality?
Very much so.
Not only in your own case but with other people also?
What was it? Would you mind specifying?
I don't remember how much we went over it.
The only thing you mentioned was that, and that he had attacked your own work in his paper, but you didn't say anything about how he might have affected other people.
As a young man he was brilliant and extremely rough, and had made many other people just as angry as me. I cannot imagine otherwise how these papers were so ignored. I do not know exactly whether Hoyle had a fight with Opik, but Hoyle never — even to this day that I'm aware of — acknowledged that Opik's 1938 paper really superseded anything that his whole group did under the accretion idea. So I don't think it was just me. But it's very clear how it worked with me.
Would you want to get to the 1950 and '51 work on the difference in composition now?
No, let's talk first about noise arising from the solar granulation, and then we'll do that and then we'll do the models.
In 1948 you launched what appears to be an entire new subject: solar atmosphere, curve of growth and so forth. And I'm wondering why. Was this tied in? Did you feel a tie to evolution work or was there some other reason you started this?
I do not think I tied it to evolution work. I do not think that anybody then suspected that the existence of a corona or any of these thin outer layers of the sun had any effect on the main body of the sun and therefore that it was connected. I had always been interested in the convection problem, which is important.
Yes. I was thinking that it is important to understand the convection for interiors.
For the interior, and therefore I was reasonably up on the literature and the thinking on convection. The solar corona had been a fascinating, unexplained problem around, and I had many good contacts — isn't that right? — with astronomers who were interested in the sun, in the outer layers.
In that paper you thanked Wildt for discussions.
Yes, Rupert Wildt certainly was one of the persons with whom I talked, over much wider ranges than just the stellar interior. I do not know where that idea came from. But I learned soon that Biermann actually had essentially the same idea, at least as far as heating the chromosphere, during the war, actually preceding my own work. I did not know of his work when I wrote my paper, so in a sense we are independent except for a small time difference in which Biermann did actually the work preceding me. But during the war, when we could not get the material.
So you give the impression — is this correct? — that it is simply one of the things you had been thinking of. You came up with the idea: "Oh, it must be acoustic noise."
So you wrote a paper, and then that was it. It was not too much tied in with general problems of convection for you?
No, but the outer parts of the sun have been a field of interest for me, independent of the stellar interior in part and sometimes connected with the stellar interior, for a long while.
You did apply it to a certain degree in your paper in 1948 with Robert Richardson. 
That is right.
You used turbulence in the core…
In the core to make a whopping energy-carrying acoustical stream. It was essentially using the same physics that Biermann and I had suggested for making the corona, in the stellar interior. However, that paper was completely wrong, and it took one very short note of Gold to show that I had grossly violated the second law of thermodynamics.
Oh, really? No, I didn't know that.
It was a very short paper of Gold's, and at one reading it was clear that he was right and the paper with Richardson was worth nothing.
What was Richardson's contribution to this paper?
Richardson was a solar astronomer, not terribly highly thought of at Mt. Wilson, and again a very interesting and special character. He was very much interested in the popularization of astronomy, but his colleagues didn't put any worth at the time on that. Richardson wrote science fiction books under a pen name.
What was his pen name?
Oh, and this was not very highly regarded around Mt. Wilson.
Not at all. And his position really was, by salary and everything, degraded, so that he felt very uncomfortable. Also, his way of communicating was not particularly making him friends. But I think he and I started already some solar granulation work, if I remember right, or he had helped me for the first solar granulation work.
Yes, you mentioned that he had helped on the 150-foot tower telescope.
Right. And therefore we knew each other, and during that visit he just came and said, "I love computing on a desk calculator. Have you anything on the stellar interior?" That's exactly what we did together.
I'd like to dwell on that paper just very briefly, and a few other things. In there you did have your standard convection core but you had an isothermal zone around it with a radiative envelope. You were looking also for a process. You were thinking very strongly at that time, 1948, about a process for sustaining that energy output that was needed to drive the energy through that isothermal zone into the envelope, and you suggested the possibility that it might be the transmutation of a heavier element like lithium, but then discounted it. Now, what were you thinking about? Were you thinking about going actively into looking for a more energetic source than the hydrogen sources that Bethe had?
Not from that point of view. It was still in that phase when we were searching for a red giant model.
And by putting the whole burden of energy transport, over at least an intermediate zone, onto these acoustical waves, you then could introduce an isothermal zone. Because you didn't need, indeed you couldn't stand, a temperature gradient in a region in which all the energy is transported by this mechanical means. And those models gave red giants. So I was eager to find a way of driving that strong an acoustical flux. Only, from that point of view, not from trying to find different basic sources. I tried to boost the convection in the middle of the star, but it turned out — isn't that right? — that whatever the source, the flux couldn't be that large by the second law of thermodynamics. In that sense it was completely false.
During this whole phase, you said no one was actively working on evolution — everyone was working on the static models themselves — did you and did the scientific community come to the realization and accept the idea of expansion? I think this will be the last type of question on that phase.
I think the moment it was clear that you could get red giants by the discontinuities produced by the internal burning, it was accepted that therefore the evolution would lead a star from the main sequence to the red giant phase, which meant a big external expansion — core contraction, external expansion.
There was virtually no way that anyone could think this way before 1938, though.
That is correct. I have been told Eddington, for example, never believed that composition inhomogenities would in fact make a big difference and make the stellar model deviate strongly from his so-called standard model.
Ernest Rutherford in 1915 stood up at a meeting of the British Association for the Advancement of Science and asked, "Why are you astronomers all so involved with stars that must contract? Why can't you consider that they should expand?" He said it much more elegantly than that. And absolutely no one picked up on that. I'm just trying to nail it down, and I think you just did it for me.
It wasn't until you could have the inhomogenities...
Inhomogenities and the shell source possibility.
I think the next thing is the paper with your wife, "A Spectroscopic Comparison between High and Low Velocity F Dwarfs."  Of course Baade had already done his work on populations I and II, but I wondered how you got interested in this particular type of problem.
The moment we really digested Baade's work it seemed to me extremely clear that there ought to be some fundamental difference between the populations. That is, as I think I then called it, inheritable characteristics, something with which the stars were born; and it had to be something other than the mass, because we needed the mass in both populations to get the whole range of all the stars observed in each separate population. So one very natural suspicion — and not just of me; I know that Kuiper, for example, immediately jumped to the suspicion — was that the populations differed in their basic inherited chemical composition. And since the dominance of hydrogen and helium was pretty clear by that time, the main difference one would suspect was in the heavy element content. Though there was also some discussion that the helium content might differ very much. In fact, I think Kuiper worked on that but with insufficient models. So I tried for quite a while to persuade the spectroscopists at Mt. Wilson, on the upper floor of the Santa Barbara Street offices, that they should take the lower floor — the Baade and Minkowski work — seriously, that it was a terribly exciting and very fundamental question; with no success. Then Barbara and I decided that that would be a really exciting topic to try our own hand on. Even though it is not directly connected to stellar evolution, obviously if in the broadest sense you want to look at evolution for stars of all kinds — meaning all masses, all compositions, or indeed all inheritable characteristics, of which composition and mass are the most obvious ones. So we picked this group of F stars and made a relative analysis of the lines. We found, I think for the first time, at least for the first time in a conscious effort, what the abundance differences might be. It was a very marginal result, only just good enough to see the sign. The moment after that paper came, then one could go after older materials, like some of Morgan's classification work and that of Nancy Roman (a student of Morgan's at that time), and reasonably soon thereafter Spitzer, Wildt and I wrote a paper interpreting all that other material and came out with the same sign.
That was more acceptable to people because you had more data? On your first paper did you have trouble? How did people react to it?
I don't remember that we were attacked in any way. How strongly it was felt to be right, I do not know but I do not remember any troubles. I always like to cite that paper as a good example that if a theoretician thinks some observational work is very important, it is very hard for a theoretician to persuade any observer to go into it. But it works very well if he himself, does a small pilot investigation and then the real specialists get interested to show how it really should be done. We did one more short observational work on red giants, but then it became a very strong — isn't that right? — program at Caltech, largely after Jesse Greenstein arrived there.
Tell me: why did you and your wife start observational work at all? You hadn't been doing observations for some time. How did you get started on observational work?
Well, "for some time" in the sense that in the war and at Columbia before the war I hadn't, but at Harvard…
At Harvard you did observational work…
A fairly large undertaking. My training from Kienle, who even though I was inclined towards the theoretical side made me do observations, was such that thinking of wanting to do observational work at least part of the time came very naturally.
So it wasn't like a theorist going into doing observational work.
Well, officially I always have been classified as a theorist, however correct you think that was.
But in your own head that was not the way you were approaching it.
No. I mean, one point that we have not come to, and I don't know whether you want to, is the influence of Dr. Spitzer on my work.
That is the next question in fact: your collaboration and your connections with Spitzer.
In 1940 when for the first time I had to look for a job, then there was exactly one job opening that the Harvard people, meaning Shapley and Bok mainly, were aware of: that was the opening at Columbia. In some sense Leo Goldberg and I competed for that. Jobs were difficult to get.
Yes, you mentioned that in an earlier interview.
And the Columbia people did offer me the job. But at Columbia I had no astrophysical colleagues. It was a small department, and Professor Schildt was entirely on the astrometric and kinematic and dynamical theory side. I wouldn't describe myself as really lonesome, but I needed other people outside to talk astrophysics with and get stimulated. At that time up and down the East Coast there still was the wonderful habit of Neighborhood Meetings, which I think had been started as early as Pickering and Schlesinger. In any case, the young people were not regularly invited but every so often were invited to those meetings, and there were other occasions when Spitzer and I just happened to be together, and we found that we just essentially spent all the time together. I admired him enormously as a scientist and liked him as a person. Then in '47 — that is jumping to after the war — a fairly big reshuffling occurred, because during the war all sorts of positions hadn't been filled, retirements had been postponed, and things moved very fast. I got a number of offers — very flattering but also very complicated to decide. I made up my mind then that I didn't think that I wanted to be the director or even the head of a major department myself. I didn't trust that I had the judgment, and also that I had the feeling of what one did in this country, sufficiently.
At least those were the reasons I told myself. Whether there was a pinch of laziness in that decision is not for me to decide. Then I decided, "You don't want to be yourself the head, in the future. You'd better go to the place which has the best head." And then Lyman Spitzer got an offer and accepted the position here, and I was offered the second slot that was opened by way of Russell having retired and Rosseland having gone back to Norway. I wanted to be in a department led by Spitzer. I could have gone to Chicago, I could have gone to Mt. Wilson and even Lick. But Mt. Wilson was rather complicated then, and I decided that Spitzer was far the bigger asset for me. He had very much the same attitude, that though he is basically a theoretician, he too wanted to have the contact with observation. Therefore quite independent of me he had accepted the position here with the understanding that he would not ask the University for major funds to build a Princeton telescope, but he would ask — and got — from the University the assurance that the research appointments, at that time just he and me but with the possibility of a third one or so being added, could go for something like one-quarter time (that is, one semester every two years) away, if research, particularly observational research, made that advisable.
So his point of view of mixing theory and observation couldn't have jibed better with mine. His influence on my work has been quite enormous even though his field is quite different. His width is just such that his remarks have always helped me. I've never sent a paper to press without asking him to read it critically. Indeed we used to have lunches regularly — as we have still, except now the group is much bigger. For many years it was just he and me. So even though it doesn't show up in my literature, there is probably no other person after my formal education was finished who had as big an influence on my research as Spitzer. With the one possible competition of Chandrasekhar, but largely during the time that Chandrasekhar himself was in the stellar interior; when he switched into other fields, his scientific influence was rather less on my own work even though our personal contact remained extremely close.
How did it happen that you worked with Spitzer on this particular problem of the differences in chemical composition of the populations? 
Because that's a field where atomic and molecular physics play a big role, and Lyman Spitzer knows infinitely more atomic physics than I do. Rupert Wildt supplied the molecular physics.
That also, of course, got him into the interstellar grain business at that time. At least he had been very interested in the interstellar medium and this also applied, because in this paper you were taking as a cause, it seems to me, the composition differences due to differences in grain concentration in the protostellar clouds. At least you were suggesting this, you had this in your mind.
I'm sure that that came straight from Spitzer, and I doubt that this particular paper stimulated him in that direction. I'll bet you that already was in his mind before, and it was just one possible clue in this connection.
Along the same line, dealing with populations, what about your interest and Spitzer's interest in the history of the galaxy itself? Did that start appearing later on or were you already talking about this in 1950? Interpreting the populations in terms of the history of the galaxy.
I think that very much started at that time. How much later was it that we published that short letter? 
It was less than five years. I'm trying to get the exact date here.
Right. Well, we don't need to worry about that. But it most clearly shows what our level of thinking and interest in the history of the galaxy was. Of course the data were still terribly skimpy compared to today's data.
So you had from the beginning some hope that this would be applicable to these sort of questions.
Oh yes, but there the driving force was Spitzer, even though I very happily collaborated and maybe some of the ideas came from me. But the basic driving force was Spitzer.
I see. When you first became interested in populations, it was more in terms of going back to stellar evolution than going out to the galaxies.
That's right, and I think in these developments towards the galaxy, the driving force was Spitzer.
Another question about Spitzer. In the beginning of this period I wonder whether his work on the Stellerator had any connection with your work. Were there any things that came back and forth on that?
You should ask, obviously, Lyman Spitzer himself.
Obviously we will.
The way I remember it, it started when the Korean War broke out that quite a number of scientists here, mostly the physical scientists but including astrophysics and several more, felt that maybe our dissociation from national activities immediately after World War II should be corrected.
I don't want to ask right now about getting into this kind of research. I'm just curious in terms of the direct influence of Stellerator work on your astrophysics. Or perhaps we can't disentangle this from the classified work and so on? Okay, I shouldn't have interrupted you. Go on. It's all too closely interconnected.
I think part of that general consideration may have been the reason that led Lyman to think about the fusion problem.
There was a general project here at Princeton, was there not, on the fusion problem?
He started it.
Spitzer started it.
Yes. Indeed two projects were started simultaneously. Spitzer started the fusion project, and Wheeler started the theoretical work on the hydrogen bomb. It's accidental that they coincided in time.
Oh. I see. I thought it was just one project.
It was officially made one project — Matterhorn A and Matterhorn B. B was the bomb and A was the fusion project.
I see, but they began independently in a way.
They were quite different ideas and quite different aims. The two men were good friends, but very distinctly each leading one half. My involvement was really of a very secondary nature to the project. Lyman had great difficulties finding physicists, the right kind of physicists, to participate in it at the beginning, because persuading people was made very difficult by the secrecy — which none of us thought made much sense for the fusion project, but at that time one couldn't prove that it didn't have any military consequences. So, I offered that as a temporary measure I would go half-time into the project until he could find the right kind of physicist. For I think one year I worked half-time with Spitzer but entirely under his leadership.
What sort of work did you do?
Some theoretical work. One of the first plasma instabilities Spitzer suspected, and asked Martin Kruskal (who then just out of a Ph.D. with Courant had joined the project) and me to see whether we could get through the equations. Spitzer had the idea of the instability, but we then showed it in fact to follow. (At the time I think we called it the "pinch instability").
I see. Was this tied in any way to Matterhorn B? Did you do work with them or did you discuss it with them?
No. I myself had made up my mind that I would not I mean after Hiroshima I decided I would not work on any atomic weapons.
What was the initial impact of Hiroshima on you, by the way? What did you feel when you first heard about it?
Extreme distress and extreme bewilderment. I'm not very proud of it, because the impact really struck me only after the bomb was used. In fact I was once asked by Teller to join the project at Los Alamos during the war at one of the times I was at Aberdeen. I was then considered a specialist in computers by way of my Columbia experience, and they needed that kind of people. I remember clearly I asked him if we couldn't express ourselves directly, and so I asked him: "Will this work help defeat the Nazis?" And I quite admire him — he was honest with me. He thought about it and said, "No, probably it will be too late." And then I told him: "I want to defeat the Nazis. I will not come to Los Alamos." But that clearly showed that I did not consider the moral issue even though…
I'm not sure about that.
Not in the full sense in any case, even though I think I had a fair picture of the size of the weapon, even at that time.
But after Hiroshima…
After Hiroshima I was extremely shocked. I actually heard it on the airplane coming back from the Italian Front on my leave back here. And after I came here and talked with more physicists I was more and more convinced that the way the bomb was applied was … well, against my convictions. I therefore stayed out, and when the Matterhorn project here got going and as a temporary help for Lyman I joined that half-time for a year, I made the conditions that I would not even be cleared for the B part of the Matterhorn project. A point that I think Johnny Wheeler never quite understood.
Were you ever approached later to go and work with the Los Alamos people or on any of this sort of work?
This was during a very difficult time in our history with the Army-McCarthy hearings. Some of this was going on.
Did your refusal to work on Matterhorn B or any aspect of this have any repercussions?
None that I'm aware of.
So you had no interaction with that?
No. I followed it (politically) obviously very actively, but I'm not aware of any handicaps. Whether or not I realized it at the time, the fact that I voluntarily went into the Army and worked very hard to get to the front overseas, I think would have made it quite difficult to attack me as nonpatriotic in this McCarthy sense. I think that without realizing it at all, my simple actions, based — isn't that right? — on my history of having had to leave Germany because of Hitler, the actions that followed from that happened to be quite protective for me.
But did you consider it at that time, in the late '50s? Did you consider what might happen if you did refuse? Did anybody even bring it up to you casually?
I think that by that time my experience, first under Hitler and then all the moral issues that come up when you really as a conscious individual are in the Army and participate in killing people and see how it is done, in what spirit it is done, made me feel extremely strongly that I wanted to follow my conscience. And the moment you decide that, you worry very little.
And your feeling was that nuclear weapons are qualitatively different from other weapons?
Yes. I think I knew that beforehand. You know, before the secrecy started, enough had been discussed of Hahn's work that the character of it — I never understood much of the details, but it was quite clear that it was…
Of a different nature. I see.
So my Matterhorn involvement was essentially that I was convinced that Lyman's ideas were something that should be tried and investigated, and therefore I was very happy to be able to help him. But it was also quite clear that on the long run it would not be good for the astronomy department if both of us were half-time down there, you see.
Did you continue, though, after this year to have connections with Matterhorn, to discuss their scientific problems and so forth?
As long as Lyman was really very deeply involved, I think I kept quite well informed, and indeed on the theoretical side I followed it with much interest. But then as Lyman slowly retired from the project, my contacts became less and less.
Did this have any effect on your astronomical work? Was there anything coming back in that direction?
I do not think so.
It's too separated from astronomical problems?
Right. Well, maybe it gave me some training in the beginning of plasma physics, in the early simple phases of plasma physics. It helped me maybe not so much in my own work but following other parts of astrophysics, which of course is always essential. One doesn't just exist in one's own field; if one doesn't understand a much wider front, one cannot do the best work in his own field. So in that sense it probably had an indirect influence as any learning of neighboring fields has an influence, but not in any concrete sense.
Tell me then, how was it regarded around Princeton by the administration or by the astronomers, that the chairman of the astronomy department should be going off and doing something like this which was quite separate from astronomy?
I did not know the key administrators of that time very much. I was very little involved in University affairs. Indeed, Lyman protected me enormously from too many University committee involvements. And only in the most recent years really only since Vietnam and since stopping working in Washington I have said Yes to more internal committee assignments. So I cannot really judge it. But when you look at the University, to say Yes to the fusion program; to say Yes to the Stratoscope project, which started small but grew fast with appreciable risks (and the basic risks are always carried by the University); and then the Copernicus program — I think even in just Lyman's and my experience, Princeton University has been extremely courageous in letting such risky undertakings be tried. I mean that is another feature that I didn't know at the time that I said Yes to coming here, but that has made me never regret my decision to come here.
I see. Well, I think the next thing on the agenda is evolving stars through the helium flash. But perhaps we want to break for lunch now? It's 12:20. (short break) All right, so we'll avoid the crowd and we'll go on till quarter of one or something like that.
Could I ask a short question, before we get into evolution, on Russell?
Fine. That's a good idea.
Certainly Russell even in retirement was still very much around when you came to Princeton, and if we go back to that paper that you had worked on, on the Solar abundances,  it seems to have followed many of Russell's ideas very closely. Had you had contact with Russell before you came here?
Yes. Indeed, the very first contact with Russell was the third day or something like that after arriving in this country. Before going up to Harvard where my fellowship was, I stayed with an uncle of mine in New York; a brother of my father's lived in New York. But Rupert Wildt, whom I knew from Gottingen, invited me to come for a day or two here. So I met Russell when I was here. Indeed, Russell kept me for something like five hours and spoke English with such speed that I practically understood nothing at all, which didn't disturb him at all.
Did this kind of relationship or attitude on his part persist even after you came here as a faculty member?
He loved to talk. And even when the Russells were quite old people and Barbara and I invited them to tea to our house, we found that to make them happy Barbara listened to her and I listened to him so that they both could talk during the whole visit. That's no exaggeration. They were friendly and interesting, except in their old age the stories were always the same.
I'd be interested to limit our discussion now to just his possible influence on some of your ideas and some of the directions that you may have had for research. That would be one thing. And then the other would be, I would be very interested to know about some of the IAU meetings when he was still president of Commission 35 and acknowledged your aid and assistance in the preparation of the reports. This was about 1950. I would like to know what your contact was, because he was still very much in stellar structure. His heart was in it at that time.
Yes. In a general way I would say that his influence on my work was very little except that, particularly in some of our earliest contacts — for example, when he visited Harvard and came down to where my desk was and asked me to tell him (of my work) — he gave me an enormous boost by just being enthusiastic.
What did he do on these visits? Did he walk around and talk to different people and ask them what they were doing?
Quite a lot, quite a lot.
I see. That was basically what he was supposed to do?
I do not know, but he certainly did it. Obviously, I wasn't watching what he was doing, except I remember the first occasion that he came to me and let me tell him (and not too long he understood very fast and asked very fast questions). It ended not with any suggestion but just an enormous moral boost, because when he was enthusiastic you couldn't help but catch it. And even later, here in Princeton, that still occurred on occasion. His enthusiasm really helped you. But as to concrete suggestions, or even concrete detailed discussions, I cannot remember, because he was not that deep in the stellar interior.
During the period when people had finally realized that giants evolved from main sequence stars but hadn't started evolving then yet, just when the realization was there, had you ever caught him recollecting or reminiscing about all the other possible revolutionary roles that had been considered for giants?
No, I do not remember any discussions of that. We all were aware of his very early speculations, but since they had basically turned out not that way — isn't that right? — that was not a natural topic for us to bring up; and he did not bring it up that I remember. His active following of that field sort of stopped around the time of that paper you referred to by Schoenberg and Chandrasekhar. I remember that because much later he asked me once: "Is the problem that the Schoenberg-Chandrasekhar limit presents overcome now?"
Has it been overcome? Does one understand how the evolution can go beyond that limit? And when I said, "Yes, we know how it works both for small mass stars and big mass stars," he was just pleased, but he did not ask exactly how it worked. By that time he was quite old. I didn't remember at all that at the time of his presidency of the Interior Commission of the IAU, I was about and helped him. But he and I always had terribly friendly relations, and therefore it's not at all unnatural that he would have come to me and that I would have discussed with him what I thought was exciting in the field.
I brought it up because he acknowledged you directly and thanked you for discussions, and he was also not able to show up at the meeting itself. Milne at that time, I believe, took over as acting president. Not to get into Russell's history, but I was just wondering: was it possible that you had taken over most of the duties of the president of the Commission as far as correspondence?
No, quite definitely not.
I don't even remember that whole story. That he and I would discuss these matters would be entirely natural, but I definitely didn't take over.
Okay, fine. Along the same vein, while we're talking about the IAU, there was something that I had noticed in the TRANSACTIONS which was of interest and I'd like to ask you about it. It goes beyond these kinds of questions. When Chandrasekhar took over the presidency of Commission 35, he changed the method of reporting in the transactions completely from one where it was a report of the president to one where it was a collection of reports by each of the members, who would respond to a question posed by the president. The question he posed to each member was: "What did you do in the last few years?" and "What have been the advances in the last few years?" which was the standard reason for the IAU TRANSACTIONS statement. You were in continual correspondence and discussion with Chandrasekhar on various things I imagine, and I wonder if he ever talked with you, because you were one of the people that he would have written to and you indeed responded in that volume.
I wonder whether the initiative to that change came from Chandra just for that commission or whether that was a change that was suggested by the executive committee. I do not remember whether in the other commissions a similar change at the same time occurred.
Well, I can't say but it sounded like it was his own idea, because he stated the rationale right there.
I see, very good. That sounds entirely possible, and the likelihood that we discussed it prior to his making the decision is very high, because that type of thing, quite particularly, he would discuss with me. But my memory is far too short to remember that particular detail.
It would be interesting to retrieve that from your correspondence, if we could.
Yes, except that much of these things were done on visits, so I'm not that sure whether that would be in the correspondence. There had been fairly depressing precedents before. I remember the very first IAU meeting in 1938 where I was present, Eddington gave a report that presented nothing but his own work on the stellar interior; and when it came to the movement to accept the report, for five minutes there wasn't a second, so it couldn't be moved. I didn't know what was going on and looked to Rosseland and whispered, "Don't you want to second?" and he shook his head with absolute determination. Well, finally somebody seconded.
What happened during that five minutes, just silence?
I'm interested by some of those reports of Eddington. This is one where he not only presents his own work but he comes down hard against other people's work, as I recall.
So there was some precedent on which to base Chandra's decision.
Could there be a sense of Chandra's feelings about this? Because in the same report, in this spirit, he published a letter that Milne had written to him, obviously just before Milne died, and in his tribute to Milne he expressed the tribute in the following way: that it "removed the last of the triumvirate of Eddington, Jeans and Milne which had dominated theoretical astronomy from 1910 to 1935." But he used the word "removed," and I thought that was a rather interesting and strong word to use. Again, this is Chandrasekhar's words and I'd be interested.
Chandrasekhar was quite strongly personally attached to Milne.
Yes, I know.
And therefore if he used the word "removed" I think as to Milne, it was with regret.
Okay. That's very good.
Well, perhaps we can go on now. Does that finish your question?
Oh, sure. I was thinking of the time.
Oh, we have time. We can start, at least, on evolving stars.
Wait, there is one thing that does tie into this and evolving stars. When the IAU Commission next met and was headed by Cowling, in '55 now, the format went back to the original, that Cowling just gave the entire review. He started talking about the various aspects of the problems of stellar constitution that one must identify, and one of those was numerical procedures and numerical machines — i.e., electronic computers. And the only paper he could identify that had anything to do with electronic computers and stars was one by one of your students, Ostervock, in 1953. And even later in the discussions when other names came up — Henyey and Biermann and Kippenhahn — still, your papers in the '40s were not mentioned. I'm concerned why; if there's any reason that you can think of (other than that they didn't pertain directly to stellar interiors, but stellar atmospheres).
The early work was with the machines at Columbia; the machines were not powerful enough to handle nonlinear equations like stellar structure equations.
You were doing just straight linear differentials.
That is, perturbation equations, just by the method of linear perturbations, the equations are linear both for pulsations and for rotation. The machines were very powerful but not for constructing stellar models. We did not do any stellar model work on those machines. So Cowling was entirely right. Cowling was a skeptic about the usefulness of the machines and felt that some of us, including me (he was always friendly to me but also critical, in a positive sense) were overstating the effectiveness of the computing machines. (He was) therefore somewhat interested to point out as of that time that they had in fact contributed very little to stellar model construction and stellar evolution. And objectively he was right. In my mind however, … the question was much more in which direction the future was lying. Therefore my feeling was very strongly that we had to get into the computing game, to push the computers and push the universities to have computers, so that in the future their effectiveness would be large. I very consciously spent many useless nights on the first electronic computer here, von Neumann's machine, with very little scientific results, but just so that we slowly built up a group who saw the potential and who eventually could persuade the university to build a computing center.
I was going to ask when you first began to become involved in computers here? Was that the MANIAC?
The MANIAC was von Neumann's machine, which really had von Neumann's brilliant logic embodied for the first time to the full extent. But it was built by theoretical engineers, extremely good except with no practical experience how to arrange it so that it was serviceable. The MANIAC suffered from enormous difficulties. Whenever something broke down, how to get at that part? There were occasions when they had to cut a hundred wires to get to the part, and then resolder the hundred wires and cross their fingers that they had resoldered each correctly.
Gamow in those early papers, in the mid '40s, had suggested that one must use them — he identified the MANIAC directly. That's why I was so interested in it. One must use those things to get at stellar evolution computations, to follow these things step by step. Was there a general undercurrent that this must be, even from that time?
Oh, I think it's very typical for the characteristics of Gamow, and in a sense of me, to be optimistic in this direction and consider that that was the future. And I made a quite major investment of effort in that direction. My colleague of over 25 years, Mr. Harm, kept pounding the desk computer and most of our publications of those years came from Mr. Harm's work, but I got the background of the electronic computer and then was prepared when the good IBM computers appeared.
I see. Who did the technical work on the computers? Did you do all that yourself? Did you have people to work with you?
Von Neumann was very interested to have a problem which was nonlinear and sufficiently complicated to really need the whole power of his machine, but where lots of hand computations for checks were available; and therefore the stellar evolution work, which I think von Neumann also considered interesting in itself, though not all that deeply — he thought that that was an excellent one. So actually next to the official major program, the meterological dynamics for which the machine officially was funded, stellar evolution got the biggest share of time.
Did you talk about these problems quite a lot with von Neumann then?
Not a lot. Von Neumann, however, assigned one of his best coders — but that was a very high level type of coding at that time because you really had to understand everything — Mrs. Selberg, the wife of the mathematician, to that project. She did all the coding, which was still in machine language at that time. I understood what she did, but it would have taken me much, much longer.
But you could follow when there was a bug what was going on and so forth?
Oh, yes. If there were bugs, we worked together.
What about subsequently? Did you keep in touch with your own programming, or have other people mostly do the programming?
No, subsequently Mr. Harm really took over the working of the machines, but we found sort of the interface that worked the best. He did not know enough mathematics to go from differential equations to difference equations. So I did the handling of the physics and the formulation down to the algebraic equations — difference equations and all of that. And also laying out the flow diagram; we had a special form, more or less standard but with little private touches of fast-drawing flow diagrams, with all the equations put in by me. Then he took over, and I hadn't to worry at all to actually run the machine. He knew the systems cards that you have to put in front and rear…
Job control cards, yes.
And all of that. I mean, so to say the basic mathematics down to the actual equations, and how you make a condition statement so that the machine did what you wanted, I did; but he did the rest. He just retired, and now for the first time I have to do the rest myself.
What about when a problem would arise, when a model was blowing up or not going the right way?
Which was, of course, the normal state.
I always like to say that 90% of the time we are stuck. In the 10% of the time, the machine runs so fast that we get through a whole evolution phase. So the debugging was the normal operation. Richard Harm did all the debugging that can be done with the help of the machine. When the error was deeper, we would study it together; if we couldn't find it directly, we had a system of putting it back into the machine but letting the machine print out very many more intermediate things. It's quite a normal technique, except we had developed it so that all the print statements were already in our code, and we just had to sort of turn indicators so that it printed more and more and more.
So the hard part of the debugging we did together, because you needed both his skill of understanding how it was in detail and my understanding of what hints you could get from the physical circumstances.
There must have been times when it was a problem to tell whether it was a bug in the program or in the physics of it.
Oh. indeed yes.
And then you had to work together on it.
Well, sometimes there were just plain errors in the equations that I had written down, that he couldn't have possibly found. That was one. And then there were a couple of marvelous occasions when everything went wrong, and we did all the normal tests, also the numerical instability tests, and it still went on, and finally we found that it was a new, true physical instability. That's how we found the helium shell flash.
Right, I want to get back to that. What was Harm's background and how did you start this collaboration with him?
He was a high school teacher in Estonia, but high school teacher meant sort of half an American college. It goes two years further than the American high school.
Right. Socially it's a fairly high position.
Yes. He and his family fled Estonia when the Russians came in, got to Germany, ended up with Biermann, and worked for Biermann to the end of the war and a year or two after the war. Then his quota number came up for immigration to this country, which seemed economically the only permanent chance. He got, first, no position at all. Biermann had written me about him, but I did not do large enough computations at that time to warrant a full-time person. But then Harm worked for a year at Yale on celestial mechanics computations, and then our computational needs had grown so large (and Lyman had found the funds) that we asked him and he stayed until this summer when his retirement came up. That was another one of these features of working under Lyman, that he did make it possible and steadily through contracts helped get the funds. I have been enormously spoiled, having had Mr. Harm, without whose care I never could have done as much stellar interiors as in fact I did.
What did he teach in Estonia? Was he a high school teacher of general subjects in science?
Oh, mathematics. So he had no training at all in astronomy?
Not in astronomy, no. He had taken more courses, while he was in Gottingen, in mathematics. If he had stayed there longer, whether he might have become an independent researcher one never knows.
One more question before we break for lunch: what are your general feelings about the impact of computers on astronomy? I mean their impact on our knowledge has been very clear, but on the way astronomers approach problems, or just the social patterns that take place in astronomy? Have computers had an impact?
I don't really see the computers as all that different from new really major tools on the observational side.
A new telescope, for example.
Well, a new telescope of a new nature. Or, for example, photoelectric photometry. It started earlier, but became powerful really only after World War II — isn't that right? During the war the photomultipliers were invented, and suddenly became for astronomy from a highly specialized tool, a generally available tool, and made an enormous revolution in observational astronomy. The computers, of course, have now widened their role enormously. They started as a theoretician's tool. Now they are running telescopes. In fact, they take the data out of the telescopes and, if you program them, do at least the first steps of analysis. But it isn't all that different from any type of mechanization.
You don't feel that it changes the way an astronomer regards himself or his subject.
No, but every tool always brings with it a balance; you can fall in love with a tool and forget about astronomy, and overdo the use of the tool and not do the maximum science, or you can also be overcautious and be on the conservative side and start using the new tool too late. So with a powerful tool like a computer, that danger always exists and you always see the whole spectrum. But that's not peculiar to computers.
No, it's not.
We're going again then just after lunch. You were saying that really your first work on evolution was with Oke.
The 1952 paper. 
'52. And that was still not step-by-step evolution, but really for me the convincing result that red giants were the right configuration toward which evolution would go.
So it was somehow a combination or a transition between the two ways of approaching it.
Right. Oke was quite young then and this was not his thesis. He was very clever and very well trained (from Toronto) in this work. Then, I think it was actually the next year, a real big step happened when Alan Sandage as a student under Baade had succeeded in getting a Hertzsprung-Russell diagram of a globular cluster pushed so faint that he had reached, in a convincing way, the main sequence. And that means that one really had in one cluster with plenty of stars the connection in the Hertzsprung-Russell diagram between the main sequence, which we theoreticians were reasonably sure we understood basically, with ample red giants — indeed a clean pattern of red giants in the Hertzsprung-Russell diagram. I think it was really that observational step, together with these previous exercises on red giant model-making, that then made it quite clear that we should now systematically start with a main sequence star, go through the main sequence phase, and presumably follow the star into the red giant region.
This presented a challenge in a way: can you match this diagram, this cluster diagram?
Very much so, very clear.
Not only for yourself but for others?
Yes. I don't know how many others, because, you see, the young groups were not yet there. I mean the Ibens and Demarques and all of the nuclear physics group other than Willy Fowler himself really didn't exist in the field yet. So there were not all that many involved. Sandage arrived here on a post-doctoral fellowship with the aim that he and I should get the whole evolution of globular clusters star.
That was his idea when came?
That was definitely his idea and of course I joined him gaily, probably being a little more cautious how much we could achieve in a year. What we did do — and Mr. Harm was always a key element in all this work, just to get through the computations was to get a globular cluster type star to move into the red giant region. We got around the first corner, the knee, as it is now called, but we could not make the star go up to become a real bright red giant as observed. 
And this was with a desk calculator?
At that time I'm quite sure it was.
You would perhaps be trying it also on the MANIAC or whatever the next thing was.
I don't quite remember how the timing was. The MANIAC came later.
I see. So the fact computers were becoming available was only ancillary to this development. The problem was being pushed independent of the existence of computers.
Other than desk computers. And therefore you still made simplifying approximations, both to your nuclear physics and to your atomic physics, to get it all through.
As far as what you used in the structure, in your IAU report of 1952 you had still been working either between isothermal cores or convective cores. But with Sandage you definitely decided upon a nonequilibrium convective … well, an exhausted core, that was helium I imagine at that time.
And you at that time know of the possible existence of the transmutation of helium?
Oh yes, oh yes. But we had no reason to believe that we could get the core hot enough if we couldn't get up in luminosity. I think we understood that. There was clearly something wrong after we got around the first knee, because our theoretical stars grew and grew in size but didn't want to go up in luminous.
Just went horizontal in the diagram.
Got redder and redder as it expanded.
And so it wasn't as if you saw the triple-alpha process and said, "Aha, there's our answer." You had to be able to get the temperatures before that process could work.
Right. Because we now didn't want to experiment with arbitrarily picked models. We were now switched to true evolution computations — isn't that right? — where every model followed from the preceding ones. I think really between the Oke paper and the Sandage paper, as far as our work here was concerned, was the switch from sample models to evolutionary sequences that one could defend every step in. Getting around the first corner, the knee, gave us — and Sandage and I realized that together — the possibility of making an age determination for the stars of the cluster, of course.
Because you have the turnoff point.
The turnoff point. That was the paper in which the first age was given. We were terrifically optimistic. If I remember right, we gave three billion years, and in the original manuscript we said that we estimated the accuracy was 30%. When the manuscript then came to Lyman Spitzer, he read it and said: "30%, Martin? Doesn't the absorption coefficient enter into it? Then he listed a whole series of uncertainties, with this marvelous physical understanding. He persuaded us to say that we thought it was right within a factor of two. Well, it turned out to be wrong by a factor of four. But I don't think that over-optimism damaged things significantly.
It was really just a matter of not having the physical constants, in a sense.
Right. The atomic physics that determines the absorption coefficient is a maze of things, and that was I think by far the biggest part.
And that just caused a difference of the scale — the time scale of stretching or contraction, so to speak, rather than a serious problem?
That's right, because basically the luminosity times the absorption coefficient is what the model gave. You measure the luminosity; if the absorption coefficient is wrong, you place the knee wrong and you get the wrong time. But clearly, beyond that first knee there was one very major thing missing. That got settled only… I don't know exactly how soon Hoyle came here. He spent two springs.
Your paper was '55.  I'm not sure just when he came as a visitor. It was in the mid '50s anyway.
He was two springs here and not overlapping with Sandage.
There's one other name I would like to know about in the interim, before Hoyle, a name that comes up a number of times, and that's your work with Irvin Rabinowitz in '53. I found two papers not listed in your bibliography, and this was one of them.  This was on homogeneous models and partially degenerated exhausted cores.
Goodness, I would be interested. That's not in this sheet?
No, it's not.
Where should it be, at the bottom of the page? (pause, looking over bibliography)
Okay, I was just interested in that paper. Was this another one along the same lines?
I think that belonged still to the pre-Sandage phase — that is, to the phase where we investigated individual models, rather than sequences of evolution phases of one and the same star.
So you didn't completely drop this, or was this something that was a vestige from an earlier interest that just didn't get completed in time?
Timewise it is after Sandage?
In 1953 it came out and Sandage was 1952.
It might have been just a question of the publication being delayed.
Okay. I have it here, Hoyle was a visiting professor here in 1952 and '53.
Right. I think both springs, not whole years. He had become really very actively interested in stellar evolution at that time. He, of course, understood much more of the nuclear physics than I. I don't remember the details, and those two springs were very hard to remember in detail, because Hoyle came in at least twice a week, if not more often, with new ideas — what physics was, you know, forgotten and what we should put differently and what-not. He had not had the experience, but we (Mr. Harm and I) could react fast to his proposals and check which might make sense and which didn't make sense. Mr. Harm and I worked extremely hard. Most of the ideas came from Hoyle. Some of the technical ones I think we supplied. It was an extremely lively time, and we tried very many things. The key result of that time, which I should have known before from lower main sequence experience, was that one has to include a more careful surface condition, which turns out for these red giants or sub-giants that Sandage and I had built to force the envelope from being radiative to being convective. The moment you put that in, you find that the stellar envelope stops expanding and forces the luminosity to go up.
I see. It was essentially the change to a convective envelope.
As a consequence of formulating the boundary conditions of the photosphere more physically.
I see. So how did it work? First you formulate the conditions and you put it in the computer and you find from your models that it becomes convective, and then immediately starts to expand.
Right. Now, I don't know exactly how we found it. Another way is, you've got to prove that the previous models were wrong. Our old idea always was: officially the models go to zero temperature. Go a little in, find where the temperature is equal to the effective temperature as governed by the radius and luminosity, then look at the density there and see whether the density is high enough to give you an optical depth of one or two-thirds or something like that. And it turned out that the densities were hopelessly too low. The stars were just completely transparent to much too high temperatures, so they were inconsistent.
I see. So that demonstrated that your model for the surface was wrong. You had the wrong physics for the surface.
That's right. Now, in the upper main sequence, nearly to the sun — not quite to the sun — you can get by with what was called the Eddington Approximation and let everything go to zero at the surface. But the moment you go to cooler stars, both on the main sequence or toward the red giants, you can't get by. You have to take a bit better formulation of the boundary conditions and then it leads you directly to the convective envelopes. Then that gave the second kink, and then we got up the giant branch. The models got very complicated then. And that was still all desk calculator work. But by that time Harm and I had enough experience so that by very old tricks, of which the first developments were in Emden's book, but (also in) Chandrasekhar's book, the so-called homology invariants and variables — by using them strongly, we could construct the cores of the stars, including the burning region, the inner region, separately from the envelope, and then attach the envelope. Fitting procedures were known before, but you had to do both simultaneously and they interacted very much. In these extreme cases the cores hardly care what the envelope does, so you can essentially do the core by itself.
So you just hold the envelope and evolve the core for a while?
Well, you build a new core and then find the envelope that belongs to it.
I see, instead of doing them both simultaneously.
Well, and letting them interact. You do the core.
I see. And this, of course, speeds it up quite a lot.
During all these phases, though, you knew you had a shell source?
And so that was basically your support for the envelope.
That's one way of putting it.
Okay. What is your way?
Oh, no. It's quite complicated. This structure, the red giant structure, you can describe in very many physical ways, and there is still no really neat and simple way that I can say as to why the envelope expands when the core contracts. We know the core has to contract to reach the higher temperatures, to get to the next fuel; but why that causes an envelope expansion, well… Expansion is hard to describe. If you want to describe it in roughly the way I like it most, the core has a very high gravity and therefore pressure and density drop at the edge of the core very steeply. The envelope therefore starts with a very low density. But the mass that you have to put in the envelope is given. And you just find that you need a very large volume, because you are forced to low density to start with. Only a very expanded envelope can contain the mass and have that low a density at the bottom. Even though that is plausible, still there are so many interlacing factors, none of which I have mentioned in this argument — isn't that right?, — that it would convince nobody.
I've always never quite understood red giants, and I've always assumed that somebody would have within his head a total physical picture of the thing, so he would know if you push it here with your thumb, you would sort of have a physical picture of how it comes out there. But what you are telling me is even you can't do that, until you've put it into the model and put it through the calculator.
No. One can make these arguments where you pick on certain factors, physical factors, and ignore others. And if you pick the right ones, you come out with the right answer. But nobody would believe it if the complete technical solution wouldn't prove your answer.
And what about yourself? How much do you rely on intuition when you've been doing some of these red giant models?
Well, after you have worked a lot with them, you do have just a body of experience. As you say, when you squeeze them here, how they are likely to react. But that is an intuition not based on basic physics but rather on experience with oodles of models. In that sense it is a little unsatisfactory. But also, the interior equations are of the fourth order even in the simplest case, and are ferociously nonlinear. There just aren't that many problems of this complexity — isn't that right? — that have been extensively investigated, and for a good scientific reason.
To get back to your work with Hoyle and Harm, at this time did you feel there were other groups that were also on this track, that you were in a race to reach the red giant region? Were there other groups?
I don't think I was particularly worried. The basic idea of how the giants came about — namely by inhomogenities and shell burning on the inside — that had been understood by a number of investigations. And the young groups didn't exist yet, and therefore there were very few working in a major computational way. Chandrasekhar was pretty well out of the field by then.
But there was Wrubel and there was Henyey and some other people…
Wrubel came from Chandrasekhar's school, spent one post-doctoral year here, and worked actually on magnetic fields in stars, not stellar evolution. And then when he went to Indiana he switched to stellar pulsations, and otherwise very much into atmospheres. Henyey was not yet in the stellar interior at all.
That was too early.
Right. So there were very few. A German group kept working, but they did not go into this massive calculation, which really was only possible because of Mr. Harm.
What about the groups that continued? Hoyle was involved, and others in Britain. Were there any computational groups? I have a few names: Qvist and a few others were doing numerical solutions of interiors.
I don't remember that we were particularly worried that we were in a race at that time. That does not say that other people did not have good ideas. But I can't remember anybody doing this systematic work — which soon thereafter was picked up.
At what point did you become aware of competition for this type of work?
Well, the leaders of the younger groups — Icko Iben, for example, came and visited me to show me what work he was doing, and asking me how to proceed. He was then a young man out of a physics department.
It wasn't until the late '50s, was it?
Right. And Pierre Demarque dropped in a fair number of times — you know, just for a day's session. My main role with them was not to teach them very much. They, like most young people, had too big plans, and my role was mostly to urge them: pick this section, finish it and publish it — instead of, you know, like Sandage in one year wanting to finish it all, and it took practically two decades to do the same. These groups — I don't remember exactly when they started appearing. Also, the German group — Biermann's general group, Kippenhahn being the most clever one. And there was one very tight race later on, but that is a little bit later.
Okay, well, let's go back then to the story. We left you with Hoyle and Schwarzschild. I guess we hadn't quite got to the helium flash.
Right. Well, our computational methods at that time were not sufficient to actually compute through the helium flash. But I think Hoyle and I were quite convinced that that was the next thing that was going to happen. We actually skipped it in that paper, made a guess what the configuration should be, and computed a few models that did turn out to simulate the horizontal branch, which was the next phase in the globular cluster diagram. But that is where we left it.
You mentioned earlier this morning that you hadn't anticipated the helium flash. How did you find it?
But he had anticipated the helium flash.
There's a helium core flash and a helium shell flash. With Hoyle we are talking about the helium core flash. That was anticipated on the basis of purely thinking analysis, published by Mestel for quite a different configuration. But it was quite clear that it applied to any configuration that was electron-degenerate and hot enough to start a new nuclear reaction. So Hoyle and I did not invent the likelihood of the helium core flash occurring. I think we sort of saw it coming, but to actually compute through what is really a thermal instability — the capability, just wasn't there. We jumped it, made a guess how it should end (which turned out reasonably good), and then got just a glimpse of the horizontal branch, really only a glimpse. That was the end of that phase. Hoyle left with the determination that he wanted to continue that particular way of investigating stellar evolution. I was uncertain that I wanted to do that. And if I remember right, for a year or so, I didn't. Hoyle did go — indeed that had been arranged — to Caltech, and there got to code an interior code himself. I'm not that sure whether it was the second year that he came from here or whether that was one year later. He fell into the habit of going to Caltech quite regularly for many years thereafter. He worked at Caltech more or less himself, but at home, in Cambridge, with a young applied mathematician. Basically I think that was sort of a mistaken choice of Hoyle's. The painstaking part of that kind of computations isn't where Hoyle's strength lies, and he did not follow it very long after that and very far. After a small interlude, if I remember right, Mr. Harm and I did decide that we wanted to go back to that topic and in fact compute through the helium flash, particularly to see how violent or how quiet it turned out to be. It was is an instability; it was hard to predict how it would work.
I see, with the hope that perhaps you wouldn't have to do too many intermediate models to get through it?
Yes. Or in any case, to do whatever was necessary to get an impression of what really happened. We had terrible difficulties, and that was still in the hand (calculator) phase. That got through and we published the first paper of what happens in the flash. Numerically it turned out that we exaggerated the violence of the flash appreciably, but still, even with that unintended exaggeration, it turned out not to have dynamical consequences. There had been some hope, and quite a few people indeed published notes, that that probably would lead to supernova and had made estimates. I obviously hoped that it would be a supernova, but it turned out not to. We tried very hard.
Because then you would have the explanation for supernovas.
Yes. It was the first real instability. And even though it started clearly as what we now call a thermal instability, it was not clear that it would not go into a dynamical phase. But it turned out not. There's quite a number of papers where we — several groups — did the helium flash a good number of times. Actually, about that time, exactly that time when we worked on the helium flash, the MANIAC work started, and it was the helium flash that we did both ways.
I see. You checked one against the other, so to speak.
Yes. Maybe that was in the meantime, when I had decided maybe I would not follow it, and Hoyle was to follow it, that we did here the work on massive stars, the beginning of the evolution of massive stars, and got into other kinds of problems. In the massive stars we never followed beyond the main sequence phases, but at least we found the phenomenon of semi-convection in that work. I think that was sort of a detour, but a beginning of the equivalent work for more massive stars, which we never followed.
Had you started on that simply because you had the apparatus available, so to speak — your theoretical apparatus?
Yes, and we were curious how things did look different, depending on mass.
And you found the mass limit, in effect how large a star could be.
That, however, I believe is not considered correct any more. The idea was not mine. There is a much earlier paper by Ledoux. We repeated that and found another limit. But in the present picture, the mass limit is set by a phenomenon that Larson pointed out much, much later — a few years ago — in the formation of stars. But what we did find, I think, in that work that has lasted is the phenomenon of semi-convection.
During this period then, up to when you start finding that you can get at least to the helium flash and get through it one way or another to the horizontal branch, a number of different people — B.J. Taylor in England and a few others, including you — started noticing that the low-mass stars have evolutionary tracks up and, to the right whereas high-mass stars have evolutionary tracks which are horizontal; and you started realizing they all start converging about the same part of the main sequence. Also, simultaneous with that you were still doing some cluster work on M-15 and M-3.
You mean observational work.
Observational work with different people at Mt. Wilson, trying to identify the two branches and what they may or may not connect with. At one point or another you were very interested in seeing if the horizontal branch connected with the sub-dwarf branch, and you concluded finally these connections and cross-overs did not exist. But at that time were you getting worried about the fact that you were having this tremendous blending effect, that evolved stars of all different masses of the main sequence were ending up approximately at the same part of the diagram?
Let me first react to that one paper I did with Howard Johnson.  I think that preceded the work that Sandage and I did.
Oh, yes, this was '51. This was very early.
I'm somewhat proud that he and I, I think for the first time, combined photoelectric and photographic photometry. That was done, under some pressure from Baade, who just was a marvelous teacher and was convinced that the two techniques had to get married to get highest efficiency. But as to extrapolating the lines, that is all mine and not Howard Johnson's fault, and it's ferociously wrong — isn't that right? The tie-up of the horizontal branch, which I think I made go into the ordinary main sequence and the other one go into the sub-dwarf sequence. Well, nothing of that makes any sense. I just didn't understand stellar evolution sufficiently to do that. So on the purely observing part I'm quite proud, but extrapolated lines show only how little I understood about stellar evolution at that time.
My only thought there was: what were you thinking in terms of by wondering whether they crossed or not?
Well, when it comes to what later on Sandage calls the "funnelling" for various …
Yes, that's the word I wanted.
That was a point that we all fast realized, but from the stellar evolution theory point of view it was no problem. Because you just follow a star and if it ends, at the same place as another star in the Hertzsprung-Russell diagram, well, the masses are quite different. Physically they are quite different. It just happens that the two things that you plot on the Hertzsprung-Russell diagram don't show the difference. From the observational side it is a much more serious question, because if you only observe the two things that you plot in the Hertzsprung-Russell diagram and know nothing about masses or anything else, to disentangle that region is a hard one. But if you stick to individual clusters — isn't that right? — you don't have that trouble. Therefore, I love to stick, and essentially I stuck, just to the oldest clusters, globular clusters, practically altogether. Except for these occasional excursions in the early phases into the intermediate-mass and truly massive stars.
Were there any particular points at which your photometric work or any work on clusters, or any observational work, had a strong impact on your theoretical work?
I would say on two occasions, yes. One extremely early far before any of this work. That was work I watched when I was a student in Gottingen. Heckmann was doing a very careful color-magnitude photometry for the Pleiades and Praesepe, and the sharpness of the diagrams seemed entirely striking. It made it very clear that the family of stars within such a cluster had some critical parameter in common. And that permitted us later to jump (to the conclusion) that it must be the age; because the moment we understood the main sequence, we knew it couldn't be the mass. So I think that very early, very fine photometric work laid the foundation for that understanding, the moment we understood the main sequence. And then I think the thesis of Sandage, tying the complicated pattern of red giants and supergiants and horizontal branch in globular clusters down to the main sequence that we theoretically understood by that time — that was a key step. In principle, logically it shouldn't have been necessary, but that's not the way the human mind works.
Did you make the connection in terms of dynamical considerations for the relative ages of clusters at first? Could you see that the youngest clusters, with the gas and dust and where they were in the galaxy, had one kind of H-R diagram, and then the metallicity changed and the kinematics changed and everything seemed to change in terms of what you would have expected would be an age difference…
So you saw all that.
That picture was already very plausible immediately following our understanding of Baade's populations. I mean that population I was basically young, and population II, the globular clusters and similar things, were old. That I think we got straightened out right away. You see, that made this question of the metal abundance differences so terribly important, because we knew that the high velocity stars were the old ones. If they had less metal, this question — was the metal formed in the Big Bang or was it formed later — was already clear then. Indeed, the moment we had convinced ourselves by the observations with Barbara and then the analysis with Spitzer and Wildt, which way it went, we considered that a fairly strong argument. It fitted the Big Bang theory well. It was not conclusive, but it fitted it extremely nicely.
Had you gone into that work because of its cosmological implications?
At the moment the stellar population difference was there, you knew — at least I felt strongly — that it must have an inherited cause. And the moment you say "inherited cause," that means you are in cosmology.
Right. So one reason for doing this work with Spitzer was in fact to get a handle on cosmology?
Well, at that time I think at Mt. Wilson you wouldn't have called it cosmology. The classical cosmology was Hubble's cosmology. This was the beginning of astrophysical cosmology. I'm not sure we would have called it at that time cosmology. Now you do.
But in your mind this was one of the reasons you were interested in studying this particular problem?
Yes. And I couldn't understand why the spectroscopists in Mt. Wilson wouldn't go after these new phenomena that Baade had put forward.
We'll get back to that, but I wanted to ask again about the observational questions: up in the '50s, or even in the '60s or '70s for that matter, have there been points at which the observations provided an important check on the theoretical observations? Have you felt constrained to follow the observations?
Oh, I would say, for example, that the one year work of Sandage and mine made the stars go to the right in the Hertzsprung-Russell diagram instead of after a while turning upwards is a good example. The way Lyman Spitzer kiddingly says: "You miss every corner, and it takes you five years to overcome every corner." He would put it much more politely, but, that's a good description. Without observations we would never have caught our errors.
But many of these things don't require current observations. This is simply fitting a pattern you could have already had there in '45, so to speak.
Yes, stellar evolution was well behind the observational data. I think that's simply because the nuclear physics was missing. The observations went ahead; theoretically we were basically stuck.
So there was never any point at which you said, "Now we need observations for this type of cluster or with this much more detail?"
That has occurred. For example, the question of the narrowness of the branches in globular clusters, whether they have intrinsic widths or not, has played a role. And there are more delicate points in which new observations quite often were indicated. Through people like Sandage, or Sidney van den Bergh in Toronto, and quite a number of other people — also spectroscopists like Searle and Oke (who became an observational spectroscopist at Mt. Wilson) — those questions were picked up and many contributions have been made. Indeed right now the spectroscopists have presented the theoreticians with problems that we don't understand.
In terms of cluster diagrams.
Yes. One of the most interesting unsolved problems now is that spectroscopists see in many cases results of the burning inside the star without us theoreticians being able to bring these ashes to the surface.
David is leaving, but we will continue. Now, on your work with Harm on the helium flash, you eventually got to the point in '62 where you were doing 70 successive models, took 500 hours on an IBM 650, and so forth.  This was getting quite successful now, models as close as two seconds apart. Did you begin to have a feeling that you were in a different kind of undertaking when you were working on such a fine scale, and with such a massive amount of computing time?
It didn't really change my work all that much. You see, the moment we were on the IBM machines, then Richard Harm took over the handling of the computing. It did change by and by in one respect — namely, that in the earlier hand-computing phase, a much smaller fraction of my time was needed, because in the computing, after we once settled what we would try Richard Harm could get along for a week at a time (even though probably I dropped in more often than that). For example, for me to go away for a half a year was very little of a problem in those years. We would correspond maybe once or twice during half a year, but the pace was slow. As the machines became more efficient the computing phase became very much faster, and therefore Richard Harm needed much more frequent contact with me. At the end it was fairly true that when I came to the observatory in the morning I spent one hour with Richard Harm practically every possible day. And that included purely the computational problems that arose; that did not include my thinking of the physical sides and preparing new equations describing new physics and putting it in the flow diagram. So the time required from me grew, just by way of the machine's speed increasing and Richard Harm not being able to take over the physical part.
I see. But in terms of the way the science was going, it was similar to the hand-calculated problems?
Now, there's another problem, and that was that as you went more and more into computers, you needed support for this sort of work. There's two things you mentioned, for example, in this particular paper. You cite the "generous attitude of the computer section of the Princeton Plasma Physics labs that decisively furthered this work." What sort of relationship did you have with them?
Just on our asking they permitted us to use their IBM machines — that was one of the earliest sets here in Princeton — at night.
Oh, without charge?
That was very generous, very substantial.
Yes, but nobody else would have used them. I mean they would just have been dead.
I see. There were not that many people around who…
No. There were many years when the question of usefulness of computers was still —. Indeed, the question of the University having a computing center was an enormous battle.
You had to really struggle to get it?
Yes, because key people like Wigner, for example, were entirely against it. If it hadn't been for von Neumann's standing (though of course he didn't belong to the University), and Tukey and Oscar Morgenstern, the economist, and me, Princeton would have been much slower getting into it. And I think that is proper — isn't that right? — if there isn't at least a core of active research professors, the university shouldn't act. But of course the total use wasn't that large. I mean the musicians and sociologists and all had not yet — the way you can write a senior thesis on the computer is a very convenient way hadn't been invented yet, and indeed initially wasn't possible. We didn't pay explicitly for computing for a terribly long time. Even in the first years of the computing center, that was taken as an overhead of the University as a whole. Explicit payment for computing came quite slow. But the moment, of course, it came, then our requests for grants had to jump.
I see. Now, even in this earlier one you mentioned a contract with the U.S. Air Force Office of Scientific Research, and I wondered how this contact came about.
Well, before the NSF became really the channel, initially the Office of Naval Research was the main channel for funding into science from the government, but then the Air Force followed actively. The contacts with the Air Force, if I remember right, were done by Lyman (Spitzer), not by me.
You didn't have too much dealings raising this sort of funds?
Well, I took part in writing the proposals, but Lyman was marvelous in knowing how to handle it. He made the work extremely light for me.
Did you have much contact with the Air Force people? Did you talk with them about what you were doing?
I think there was one contact man I could not now remember his name — with whom I had regular discussions, and really got on very nice and friendly terms. There was no faking. I mean we openly proposed this research. We did point out in which way it pushed applied mathematical techniques; but we said we wanted it for the astronomy and the understanding was always that.
That was all right by them.
Well, I think in those years both the Navy and the Air Force were very anxious to somehow have contacts with scientists, that either might have applied fallout — isn't that right? — or that in emergencies could be called upon. After all, the experience of World War II was still close.
So it was to keep the lines of communication open.
I think that was the essential idea, but I am really not one to be the best judge.
(To get back to) stellar evolution: when one got through the helium flash, at least for some models, what would you say came next? Where did things go from there? This is the '60s now we're into.
We here — essentially the two of us and always here and there a student — after the helium core flash rushed through the next phase, which in globular clusters is the horizontal branch: we didn't pay much attention (which I considered afterwards was a mistake) and went through that rather easy (I mean computationally easy) phase quite fast, and then went into the phase when the star again becomes a red giant. It has now a helium-burning shell and further out a hydrogen-burning shell. By way of paying fairly little attention to the horizontal branch, we missed quite a number of things which then Iben and Demarque picked up. They did some extremely fine work, of which we later on did a little but very much as the third or fourth in the game. But by not recognizing those opportunities, we did get fast into the second red giant phase and encountered, with no foresight, the shell instabilities which ever since have played a major role with all sorts of consequences. Relatively recently Iben, for example, has done the same work for somewhat more massive stars and shown at least one case how some some burning products in medium mass stars can appear at the surface. That recognition of the helium shell flash phenomenon, this regular thermal cycling inside: that was the one case, without us realizing it, that in fact was a race. Without us knowing it, after we clearly had convinced ourselves of the flash and computed through a couple of flashes and could say roughly what the consequences were, we wrote a manuscript and sent a few reprints around, including one to Kippenhahn. Kippenhahn wrote a letter back in his typical friendly and lively manner, and described the impact of that reprint, because they had followed a medium-mass star through into that same phase and had just gotten into that state that Harm and I had been in a half a year or so earlier, where things just misbehaved. The first thing you assume is you have numerical troubles — one of these Courant-Fredericks numerical instabilities or other types.
Because the numbers in your last significant figure are adding up or some silly thing like that.
Right. The models didn't want to converge, or if they converged, they skipped around from model to model, and it just didn't make any physical sense. We pulled all the tricks to discover whether it was numerical trouble, whether we had made a coding error — that's always the worry — and finally we just went to finer and finer time steps and also put more and more shells into our models, so that we didn't miss anything —
Very fine resolution
Very fine resolution, both in time and in space, and convinced ourselves that it was real. With very small time steps, then everything went smooth but violent. And the moment we had that, it was very easy by rough analytical simulation to show that that instability in fact should occur under the circumstances that the star was then in. They were just about convinced that their troubles were not numerical, but they weren't quite sure; and they, as Kippenhahn said, had just mentioned at luncheon maybe there was really something physical going on. So in a sense they are co-discoverers of that instability, because two months later and they would have had it.
I see. I'm interested in your Darwin lecture talking about this.  You say, "There was much confusion despite warnings by a non-numerical theoretician." This was about the instability. Who was this non-numerical theoretician?
Again, I see. And this was a warning in print, not that he had said something to you personally?
Not all so personally. He had written a couple of key articles in handbooks about all types of instabilities. And this class of instability that he called secular, and of which the thermal ones are a sub-class: he had emphasized in long expositions and equations that no general statements could be made. He had in private conversations (we knew each other very well) told me that he didn't feel at all at ease that the stars were necessarily thermally stable, except in this one case that Mestel had pointed out, the degenerate core; we all understood why it should be thermally unstable. I must admit that I just took the attitude that Paul Ledoux, being by heart a mathematician, is impressed that we have no clean proof of the stability. Which obviously was true. But I thought the physics makes it very unlikely that thermal instabilities occur in a star in non-degenerate regions. Well, it turns out that my guess was wrong, and Paul Ledoux's warning — though without specific cases, a general warning — was right.
Tell me: all through this period (and this would even go back to the late '30s), there was a rise of the numerical methods as against the people who were simply doing equations and so forth. Was there any hint of tension or banter between the people as to which way really gave you true results?
I don't think there were very strong feelings that one was the right way or the other was the right way — at least not in my generation and younger. But individuals had strong preferences what to do. For example, Mestel by preference did very little numerical work and much more basic analytical considerations. Chandrasekhar, really, in spite of his enormous ability on the straight analytical side, put large-scale computing as an accepted method. And that was a very important point, because a generation before him — I mean the English school: Eddington and Milne — as I have I think described were actively against computing as not a proper scientific tool. But I think from Chandrasekhar on down… Oh, we got kidded every so often that we did too much computing and too little thinking about what the computer told us, naturally, and often that was true. About the horizontal branch results that Mr. Harm and I had, for example, we clearly did too little thinking, and Iben and Faulkner and Demarque got very fine results out of it. On the other hand, that permitted us to get ahead and find the helium shell flashes, which have a lot of physical consequences, some of which we understood at the time and some of which came later.
Would you say that one couldn't have gotten through the helium flash, even the helium core flash, without big computers? That this was a task beyond the desk calculator?
In any secure way I think that is true.
I'm a little puzzled. It seems almost like too good a coincidence to be true that the substantial computers came along just at the time that you were at the point where you needed them within a few years.
In a certain way, if the computers wouldn't have come at that time, we would have done other things.
You wouldn't have tried to go through that. You would have done something else.
That's right. And then it wouldn't have looked so much as a coincidence.
You would have gotten to that point and stopped and then gone on to something else.
For example, continued with the more massive stars — isn't that right? — or varied our compositions more and investigated something else. So in a certain way you may call it a coincidence; but if it hadn't occurred, we wouldn't have been dead.
You wouldn't have just stuck there trying to get through the helium flash.
I understand. Okay, well, one thing we've skipped a little bit, and that is your book on the structure and evolution of stars.  How did you come to write that?
I don't remember whether the first idea of writing that book came from myself or from people asking whether I wouldn't write a book. I got quite interested by that idea. But it was also clear that stellar evolution was in an extremely rapidly developing state, and therefore I was quite ill at ease whether that was the right time to write. Lyman Spitzer was inclined to favor my doing it, but he is also terribly cautious not to persuade staff members to do things against their own initiative. So he was encouraging but not driving. I remember very definitely one discussion which essentially I think clinched it in my mind. I just asked Struve on one occasion when we met (I don't know, probably at Mt. Wilson or Yerkes) what he thought about writing a book. He put it extremely clearly: "Well, there are two kinds of books. One you write when a field has closed affairs, and is likely to stay at a certain plateau for a while — then you write a definitive-book." For example, Eddington's book was essentially definitive until nuclear physics came in, with one exception and that was Chandrasekhar's development of the white dwarfs. Therefore Chandrasekhar's book has a special characteristic because that very important personal piece is contained in it. "Or," said Struve, "you write a book in the middle of a development. Then you will know the book will be outdated very fast. But if it is written in the right way, it can stimulate and speed up the development in a very active fashion. But then the book has to be written knowing that it's not a definitive book. It shouldn't be a terribly big volume. It should be essentially addressed so that those in the field and those who want to get into the field have a basis, an easy access to, the field." That seemed to me a very clear view, and I was very attracted to write that kind of a book — not a big volume and very much describing the state and the background rather than being complete.
I see. So your intended audience was, for example, a graduate student who wanted to get into the field?
Yes, or a post-doc. Graduate student is probably right. As it turned out, the development of the general knowledge became fast so that it was used by the beginner graduate student or even an advanced undergraduate course, after a few years. I must say that I was terribly grateful for Struve's putting it so clearly. In a certain way, of course, it prejudiced my decision, that he indicated that a book written at a state when you give up writing a definitive book could be the most helpful for the development of the field. That gave a particular challenge to me. Then when I had decided to write it, Lyman Spitzer gave one other terribly important point of judgment. He said, "If you want to write that book in a field that is so fast-moving, you have to write it in a short time; and the only way you can do that is to spend one of your half years not going to Mt. Wilson but to go somewhere where you are relatively undisturbed and concentrate entirely on writing the book." As a matter of fact, Barbara and I went to Paris to the Institut d'Astrophysique. It's a little insulting, but I do not meant to be insulting, to say that we chose it in part because there was a good library and I don't understand French, which reduced the interruptions to a minimum. And 80% of the book I wrote there; 10% I had written before as a trial. The last 10% I finished after I came back.
I see. How in fact has the book sold? What has the selling curve been, so to speak?
It initially sold far more than I ever had expected. The Princeton University Press took it after quite some haggling. I wanted it to come out as cheap as possible, never mind how it looked, because it was supposed to be a working book, not a tome to look nice for a long time. And therefore I wanted the price as low as possible. Actually for the first printing I agreed not to get royalties, so as to put more pressure on them, and they found a very cheap printer in Austria. I think they got one copy from him and then photocopied or some such way. I forget now; I think they printed 1,500 copies and bound 500, because I told them, "I'm not sure you will sell more." But it went very fast to the 5,000 mark. Then I succeeded to persuade them to sell it to Dover; that was something like five years later, and there were already appreciable portions of the book outdated, and I felt uncomfortable because their price started going up. They sold it to Dover.  I have no knowledge whatsoever how many copies Dover sold, even though I really want once to contact them and find out —
You don't know how it's done since.
I have no idea.
Do you have any idea whether people are still using it in courses?
I think at some places it is still used, to read the first three chapters and then jump into a more modern book. But I think even that must be petering out. I would not use my book.
But you feel it's fulfilled what you wanted it to do?
Yes. I'm very happy that I had the courage to write it. Because I think quite particularly it made it easy for physicists to enter that field. At that point I may be overproud, but I think it really helped to get physicists interested in astrophysics, at least by way of one particular branch.
Had that been a particular interest of yours?
It's always been an interest of mine to have physicists or applied mathematicians or geophysicists interactive with astronomy, but that is a fairly trivial recognition of a necessity for astrophysics, and not limited to myself. After all, my father certainly was an early example, even though I realize that now better than I, did even ten years ago. And Hale tried to get a spectroscopist already at the Yerkes observatory.
Well, okay, I don't want to go into detail on the work in the later '60s and '70s on evolution, because that's a little too recent. But I'd be interested if you could give me a summary of what you think have been the most interesting or significant developments since the helium shell instability.
You mean work of myself or in general?
In the field in general, and your own work of course, but the field in general.
Well, in the field in general, one enormously important development occurred immediately after the paper that Hoyle and I wrote. That was by Hayashi, realizing that on the right-hand edge of the Hertzsprung-Russell diagram there was a limit (now called the Hayashi Limit) to the right of which no hydrostatic configurations can occur that fulfill a couple of other conditions which are entirely reasonable. This suddenly threw a common light on what we had encountered on the main sequence, that we couldn't get along without convective envelopes, and what Hoyle and I had encountered, that we had to introduce convective envelopes for the red giants. Actually in his basic first paper Hayashi used the very equations that Hoyle and I had used, but pointing out that they were sort of independent of our problem. That was one of the early very fundamental steps: you had a main sequence and you had the Hayashi Limit and most everything happened in between, except the white dwarfs. Then occurred a very important period, to a large extent connected with the horizontal branch in globular clusters, but also connected with the wonderful development of finite amplitude pulsating stars. One of the most interesting results is the question of the helium content, particularly of very old stars — again, clearly a cosmologically decisive question. The horizontal branch investigations were started by John Faulkner at Santa Cruz, but Iben and Demarque's groups participated very heavily. The pulsation work was started by Robert Christy at Caltech, and much of it was carried on by (A.N.) Cox of Los Alamos and (John P.) Cox of Boulder, Colorado. I think that is not yet quite a closed issue; particularly the pulsation arguments are very difficult and not closed.
You haven't gone into those things yourself, these pulsating stars?
I did very very early work. My initial topic was pulsating stars but not in this connection, not at all. I have done work, and that is the very latest work — there is still one manuscript, of which the material is there (pointing to desk) that I have to write — in the pulsations that do occur at the very end of the stars that Harm and I have been following so many years. Long-period variables, which presumably end up in a dynamical expulsion of a planetary nebula. But we found that to be an extremely hard problem, as some others who have done shorter attempts had found. We still are all convinced that planetary nebulae are ejected in that phase, but truthfully we have not succeeded in proving it. Harm and I in our last work found one more thermal instability that occurs during pulsation that complicates matters enormously and I think is the cause of some of the difficulties of the previous investigations, but we have not overcome it. We have recognized it but not overcome it. Perhaps the most active fields that we have practically not participated at all in but which I consider very major are the following two: One, that the spectroscopists find more and more cases, as already mentioned, where material burning products show up at the surface. Iben found one particular circumstance where a mild form of mixing from the interior to the exterior can occur. However, that does not cover, by a large margin, all the cases where such mixing is indicated by the observations, so there is a strong suggestion that something is missing. Much theoretical interior work is being done to find the mistake, and the observations are steadily accumulating. So that I think is a very active and terribly important field, because once we understand it, it ought to help us to check whether our interior burning cores are really right.
Right. You mentioned over lunch that you yourself aren't going to go into this because of Harm's retirement.
You're going to other areas.
There is a second area.  I don't know whether you want that first.
No, I wanted to continue with the evolution. In the last ten years or so, do you feel there's been a lot of changes or surprises in areas such as energy transport by convection vs. transport by radiation? Those sort of problems?
In the last ten years, no. But in the very massive stars there are still very grave problems. If one does not feel quite sure about neutrino theory, then one has choices that play a big role in how much of the energy produced by nuclear processes in massive stars is carried away by neutrinos and how much by photons. And that makes a big difference in the evolution of massive stars. In a general way the models for the cores of the massive stars are fairly well settled — at least until the very end phases. But what the envelope does depends enormously on details of the interior, and there have been some disputes, in the sense of differences of opinions. So for the most massive range of stars, advanced phases corresponding to helium ignition turned on by a flash, helium burning, and higher burning phases, much more work still needs to be done.
Extremely complex cases, several shells and so forth.
Right. Complex cases, but there are also some of these fundamental questions — what is the right physics. Also, the observational work is much harder because the total number of massive stars is much smaller, so to get clean statistics, one just never has anything approaching like a single globular cluster diagram.
There's been a lot of work on double stars and the interactions between them. Do you feel that this has had much contact with regular stellar evolution work?
Has this fed into regular stellar evolution work? Or is it just a separate activity?
The way I would describe it is, there is what you might now want to call "classical" stellar evolution, which means a single star that has no companion, no rotation, no magnetic field. And that is really all that we have been talking about and certainly all that I have been involved in. We know that very many stars are double stars. We know that very many stars rotate. We don't know too much how many stars may have strong magnetic fields.
Some of them do at any rate.
Some do. And therefore these three (shall we call) classes of stars that are not simple classical bodies — there, there is an unbelievable lot to do. On the double stars, very major headway has been made. The very first paper was written by Don Morton in his thesis with me, but that was a single paper and not driving the problem very far. But the recognition of mass exchange from one component to the other was a result of that paper. But the variety of circumstances under which this occurs and the speeds with which it occurs and the consequences it has, have only been investigated in the last 15 years. That has been a major field, and it still has not gone very much beyond the phase of one mass-exchange — that is, from one component to the other. It's far from clear how two stars in a close binary will fare when both want to get into the extended evolution phases. So there is, I think, a very big field open and a potentially very interesting one. With regard to rotating stars and the techniques, applied mathematical techniques including with computers (but it takes more than a computer, it takes really good applied techniques to handle these two-dimensional problems), much progress has been made. There are still some basic problems left, essentially connected with the Eddington circulation that is a consequence of rotation. When a star rotates really fast, then you cannot ignore that circulation because it redistributes the angular momentum that makes the rotation. In rotating stars, big headway has been made. But, for example, we can't even answer for the sun whether the center does or doesn't rotate ten times faster than the outer layers.
On the basis of theory.
Right. But very big steps have been made in those connections. As to magnetic stars, the progress has been extremely slow. As far as I am aware, there is not a single model in existence now which has a strong enough internal magnetic field so that it counts, that it produces an electromagnetic force at least somewhat comparable with the gravitational forces, and that is self-consistent and stable. As far as I know, there is not a single model. But quite a group, particularly in England, are working on that.
Some of the things you've mentioned remind me of another question I had, and that is: have Ray Davis's results on his failure to detect solar neutrinos at any point affected your thinking? Have they led you to wonder whether there may be something seriously wrong with the stellar interior theory?
Indeed, that was the second point that I referred to, because that is a phenomenon of a single star not rotating fast and presumably not having all that strong a magnetic field in the interior. So in that sense it belongs to classical stellar structure theory. I consider that Davis's experiment is an extremely crucial one, and in some way more direct than these complications of seeing the burning products on the surface without us theoreticians being able to transport them there. The most recent result of Davis seems to be that he sees neutrinos; he thinks now definitely that he sees not just "around zero" but a distinct excess over background, but three times less than the standard models give. I feel that this is a very important clue. For all I know, it may be the same type of mistake as the one that doesn't permit us to mix the ashes out.
It may be the same error somewhere in the theory.
That's right. Not necessarily but I wouldn't be surprised if that turned out to be.
By which you mean it's not a terribly essential one. It wouldn't affect the overall structure too seriously?
I would be a little surprised if the overall picture of stellar evolution could be that basically wrong and fit so many observational points. I don't think we have that much freedom left. Whenever we got discrepant results I think we found an honest physical mistake. And therefore even though this neutrino discrepancy is very frightening, I would be somewhat surprised if the total picture would have to be changed. But I think something is missing, and therefore I terribly much hope that funds can be assigned for Davis or his younger collaborators to do the next experiment which more directly measures the neutrinos that come from the basic process, not the neutrinos that come from a long side chain where you always may worry about very minor details.
Do you think there's a possibility, but perhaps not too strong a possibility, that the mistake is in nuclear physics rather than in the stellar theory?
I would love to put it in another way. I think if a discrepancy of that kind occurs, where the error can be in roughly three different dissociated fields, it is better that everybody in one of the fields assumes it's in his field and works very hard to see what could be wrong in his field and search for new ideas. Otherwise it never gets resolved. So instead of blaming the nuclear physicist, I'd much rather find a little error in model making — a little error, that can be crucial on certain points but not overthrow the whole field. Because I think the whole field is anchored by now in too many points to be all that wrong.
Okay. Now some general questions about the connection of evolution work and cosmology. We've touched on a couple of points, but I'm curious in particular about the problem of age. I wonder what has been the evolution of your views with time on the chances that stellar evolution theory can provide a good check on the age of the universe.
Well, after the first wave of very strong optimism about how accurately you could determine ages of clusters, old clusters being the key ones, naturally I have become more cautious. I would still say that if the ages of the oldest clusters turn out to agree with the classical cosmological time scale as measured by the Hubble constant — agree, or differ in the right sense, namely that the clusters are younger than the Hubble constant by a reasonable margin — I think that strengthens the picture enormously. Even if neither the Hubble constant nor the ages of the oldest globular clusters can be determined with an accuracy better than thirty per cent. Just the fact that they are very much of the same order of magnitude I think is a major contribution to cosmology.
And this happened already back at this fairly early point.
Yes. Though there is this uncomfortable history, that I think one should admit to, that we always have agreed, even when we both were terribly wrong. In that sense the suspicion of some human, wishful influencing I think is justified.
Well, were, you ever aware of any influence? Did you ever calculate something and find "My gosh, this is so far off from the Hubble constant, it must be wrong"?
No, that concrete example has not occurred. But if Sandy and I had found a really whoppingly discrepant value from the then believed value of the Hubble constant…
By 20 billion years, for example.
For example. I think we would have at least expressed ourselves very much more cautiously and given it automatically very much less weight. There was one point when Sandage (but that was a very short job) found a new old cluster. I mean he newly discovered an old cluster. With his very first observations he fitted an age and got twice the Hubble constant. Nobody believed it. It turned out when he did good photometry and a couple of other people did good photometry, it all completely changed.
I see. So there has been this interaction.
You can't help it. I mean if you have wide enough views — isn't that right? — that you don't just take your own line, that interaction is there; and normally the instinctive wish is to want to make things agree.
What about abundances? What has been the evolution of your views on the chances that evolution theory could provide a reliable initial helium abundance, for example?
May I turn the two parameters around and first speak about the heavy elements? Because historically that was the first. The moment the population problem was posed by Baade, it seemed to me very clear that we had a chance of a major contribution to cosmology; and even though our own earlier work was still very weak, at least it gave the right sign for a Big Bang cosmology. More recent work has pushed that very much further. Now, that is mostly spectroscopic work, though stellar evolution is always needed to classify the stars by age; otherwise the abundances alone don't tell you so much. So on that point there was, between the age determinations and the spectroscopic abundance determinations, I think a clear feed into cosmology. The helium problem I consider a terribly important one. I wish I was more convinced that from the stellar interior — like the work on the horizontal branch or on the pulsations — I wish I could be more convinced that it is already tight. Always the results seem to agree with what the Big Bang says, but I personally have a hard time to be yet quite as convinced as some of the younger authors are as to how safe the results are.
Do you feel that the results have at least gotten better with time, have gotten safer?
Were there any particular milestones in this?
On the horizontal branch I think one should credit, probably, Faulkner with the original idea and even with maybe the most critical work.
This is in the late '60s and early '70s?
Roughly. The pulsations started quite long ago but have been an extremely tough, uphill battle. In part computational limitations play a role, because there you are essentially computing sequences of dynamical stellar models to get through one period, and that's really a tough job. My last experience doing pulsation work, not for that purpose but in quite violent dynamical phases, has made me particularly worried about how far we are. I'm very optimistic that that field can and will be pushed, as numerical techniques are just continuing to improve by leaps and bounds, and of course the machines are getting faster and bigger; that combination really pushes the pulsation field ahead. So I doubt that ten years from now we will still be in the same dither regarding the helium content.
Has there ever been any pressure from Sandage or other people interested in cosmology that you might know for you to go into a particular area, to look for particular things that were interesting to them?
Oh, yes. Sandage I'm not quite sure … But for example, we had relatively recently a very clean-cut exchange agreement. Merle Walker from the Lick Observatory had observed some clusters in the Magellanic Clouds, which he suspected might be fitted to evolution curves of stars with somewhat peculiar abundances. So I promised him that we would compute those, and we did. Mr. Harm and one graduate student in fact did a very thorough paper on that, and that was published. But the agreement was that Walker, who had just told me that he had marvelous observational material that was already ten years old (but he didn't think anybody was interested in it) connected with the nucleus of the Andromeda Nebula, that he would really use it and publish it. He did that too. So we may call it a horse deal, but I think it was a good horse deal furthering science in two areas.
Your interest in cosmology: how strong a motive has it been in your work in stellar evolution? Why have you been so interested in stellar evolution? Why has this been a major field of yours? What's the motive for studying this particular subject?
I think it would be false to maintain that I had all that broad a foresight of its potential early in the game. I think that basically I should be modest and say that the first theoretical book that I was advised to read was Eddington's book, which fascinated me just no end. That was the book on stellar interiors, and I chose the field of the stellar interior, and I've described to you already what I think as an afterthought that that was a terrible mistake because nobody could guarantee when the nuclear physics would come. But then the nuclear physics came and a whole wide field…
So you automatically went into evolution; once you were in interiors, you went into evolution.
Right. So I don't think I can claim any foresight at all. Except that with some pride I feel that I have picked worthwhile projects and not let myself get entangled in too small details.
What have been your interactions with Dicke then, on questions of cosmology and stellar interiors and the solar interior?
Always very lively exchanges. We have, for example, computed through the evolution of the sun under Dicke's assumptions and have shown that a parameter that occurs in Dicke's theory has to have a value smaller than a certain number. Otherwise the sun couldn't, after four and a half billion years, be as it is observed. And that value actually presses that parameter fairly close to the limiting value that makes Dicke's theory go into Einstein's theory, but not identical. I might say interestingly different but not ferociously different. Since then I think over tests, dynamical tests within the planetary system, have pressed that parameter even closer to Einstein's value. And, of course, the question of the solar oblateness and the question of fast rotation in the interior are questions he and I have discussed to a very large extent.
I see, just in the matter of lively discussions. Have you taken any particular view on his theory?
No. Indeed my urging to Dicke has always been, "Either repeat the experiment in a good climate, not in Princeton, or write your experiment up in detail against all the opposition that it has and go to something else." I feel that Dicke has probably wasted his magnificent brain-power and time more than is, scientifically effective for too long after having done the experiment.
After having started it. Well, we could go on, but it's ten to four and we promised to stop.
MONTHLY NOTES OF THE ROYAL ASTRONOMICAL SOCIETY 102 (1942): 152-153.
ASTROPHYSICAL JOURNAL 104 (1946): 203-207.
ASTRONOMICAL JOURNAL 116 (1952): 317-30.
ASTROPHYSICAL JOURNAL 108 (1948): 373-87.
ASTROPHYSICAL JOURNAL 112 (1950):248-65.
(with L. Spitzer, Jr., and R. Wildt) ASTROPHYSICAL JOURNAL 114 (1951): 398-406.
(with Spitzer) OBSERVATORY 73 (1952): 77-79.
ASTROPHYSICAL JOURNAL 104 (1946): 203-207.
ASTROPHYSICAL JOURNAL 116 (1952): 317-30.
ASTROPHYSICAL JOURNAL 116 (1952): 463-76.
ASTROPHYSICAL JOURNAL 121 (1955), SUPPLEMENT 2.
(with Rabinowitch and Harm), ASTROPHYSICAL JOURNAL 118 (1953), p. 326. The other is "Inherited and Acquired Characteristics of Stars," ASTRONOMICAL JOURNAL 57 (1952), p. 5.
ASTROPHYSICAL JOURNAL 113 (1951): 630-36.
See ASTROPHYSICAL JOURNAL 136 (1962): 158-66.
"Stellar Evolution in Globular Clusters," QUARTERLY JOURNAL OF THE ROYAL ASTRONOMICAL SOCIETYi 11 (1970): 12-22.
STRUCTURE AND EVOLUTION OF THE STARS (Princeton, N.J.: Princeton University Press, 1958).
A reprint edition was issued, still available in 1978 — S.W.