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 Jeremiah P. Ostriker by Christopher Smeenk on 2002 May 14, Niels Bohr Library & Archives, American Institute of Physics, College Park, MD USA, www.aip.org/history-programs/niels-bohr-library/oral-histories/34487
For multiple citations, "AIP" is the preferred abbreviation for the location.
In this interview J. P. Ostriker discusses topics such as: his educational background; Harvard University and the University of Chicago; working with Subrahmanyan Chandrasekhar; W. W. Morgan; University of Cambridge; Princeton University; Lyman Spitzer; Donald Lynden-Bell; early computer programming; astrophysics; stellar structure and stellar dynamics; Fred Hoyle; white dwarf stars; Peter Bodenheimer; James Gunn; pulsars; Martin Schwarzschild; cosmic rays; P. J. E. Peebles; Roger Blandford; cosmic background radiation; Geoffrey and E. Margaret Burbidge; Vera Rubin; steady state theory; Ralph Alpher; Fritz Zwicky; interstellar medium; Beatrice Tinsley; stellar evolution; cosmology; Ed Witten; Marc Davis; dark matter; Martin Rees; star formation; Sloan Digital Sky Survey; National Academy of Science.
Editorial Note: Bracketed comments in the text were added by the editor.
It’s May 14th, 2002. This is Chris Smeenk interviewing Jeremiah Ostriker at his office in Cambridge. I’d like to start with a background question about your undergraduate education. You majored in physics at Harvard.
I started in chemistry, and then changed to a physics and chemistry major.
And what caused you to switch to physics rather than just directly chemistry?
It’s actually more complicated, because I even took an astronomy course. But it was so terrible that I had to petition to get out of a one-year course after half a year.
So you left the course after half a year?
Yeah. I remember it vividly.
So, despite that you developed an interest in astronomy later?
Yeah. I was interested in astronomy, but that course was so terrible that it discouraged me. And then after college, I thought about it some more and decided to go back into it.
When you went to the University of Chicago…
Let me answer your first question. I was interested in chemistry because I had liked high school chemistry, basically. But I liked physics a bit better, and it was a joint major at the end. And I remember, when I graduated I went to ask my favorite chemistry teacher — whose name I won’t mention just now because he’s alive and around — whether I should go and get a Ph.D. in chemistry. And he said, “No, I wouldn’t advise it.” I was crushed.
Did he give you further reasons?
So I asked, “Why?” He said, “Oh, I think you’d be bored.”
Wow. And you went to work in Washington D.C. for the government?
And that was solid state physics? Is that right?
Yeah. It was experimental science. It was a — spent a year to not get drafted, and see what I wanted to do. And then, I read a little bit and… it was actually Fortune magazine had a series on scientists of the times, and they had included one on Chandrasekhar. And he looked so wonderful. Then I read his papers — some of his papers — and it sounded like, “Hey, I want to go there.” So that’s when I decided to go back in astronomy, and applied to the University of Chicago.
Now they, at that time, had a separate department in astronomy?
Yes. It was in Williamsville, Wisconsin. But I ended up taking courses both in the physics department and there. So I would go back and forth with them.
Okay. And did you start working with Chandra [Chandrasekhar] immediately?
No, and it was very strange because after my first year, I approached him and he was very formidable. I said I wondered whether he could give me a project to work on, because he was around, he lived there. He had been an editor of the Ap. J [Astrophysical Journal]. The Ap. J. office was in [inaudible name]. And he said, no, he didn’t take astronomy students: only physics students. I said, oh, that was disappointing, because I had come there to work with him. And I said, “But I noticed that you’ve taken so-and-so and so-and-so in the past.” He said, “Yes, I used to, but I don’t any longer.” And then I managed to prevail on him to give me some little test problems, to work on this and that. And so finally, he did take me, as a thesis advisor. But it took quite a bit of persuasion because he hadn’t had any [astronomy doctoral students] in a decade or more.
Was it because his work was more theoretical than most of what the astronomers were doing?
That may have been a part, but he was also feuding with senior members of the astronomy department.
Oh, so it was sort of departmental politics?
Yeah. I don’t know what the mixture of motivations was, but those were two of the motivations. I assume that those were.
And did you have a stronger background in physics than most of the other graduate students at the time?
When you did prevail upon Chandrasekhar to take you as a student, what was it like working with him?
He was very warm and friendly with his students. He spent a lot of time with people. Formal. It was quite enjoyable. He was really a model advisor, I thought, in terms of his interest and enthusiasm in details. You know, he did things like — I think I went over drafts of his things as well as the other way around. Those were sort of test problems he would give me — so and so — to find any mistakes in.
What sort of work was he handing you to look over?
It’s hard to go back to that point, but I think he wasn’t yet in his relativity stage. So it was more hydrodynamics at that time.
What did you end up choosing as a thesis topic, and how did you decide on that at the time?
I’m not even sure I now know. But I know that I did 1-D polytropes — that was certainly a part of it — and cylindrical polytropes and ring-like solutions. The ring-like ones had had a lot of applications. You know, if I looked and decided, because what I did was pure cylinders — isothermal cylinder I found an exact solution for — and other polytropes. And then [I] showed how you could wrap them around with a perturbation calculation and make them toroid. And then, that subsequently developed into a big field of tori because you find them in quasars and other things. So I think that was the thesis. I was going to do stability, but never got around to it.
Yeah. I think some of your very earliest papers were on this and this probably [led to] the idea of thesis research. So how did working with Chandra influence your scientific style, do you think, in this very early work?
The idea of mathematical rigor and taking well-defined problems where you could actually understand the solution — which is the way he went about things — appealed to me a lot.
Is that what appealed to you so much about his work? You were reading his work before choosing graduate school.
Right, exactly. It was a combination of him influencing me. But also, I wanted to because of that. I had read things by a couple of people who were here [at Cambridge], just by chance — Eddington and Jeans — while I was young, who were similar in spirit. And in fact Chandra said, “Go ahead. Study here.” So my returning here was just sort of cyclic, I think. The fact that you could, with relatively simple and soluble mathematical models, really learn something about complex physical systems appealed to me. Then and now, it always appeals to me more than reverse engineering back from the observations to try to deduce things, which is another way that theoretical astrophysicists work. I have always preferred the ab initio calculation, where you just put in good physics and math and just march forwards.
Who else did you work with closely while you were in Chicago? Are there other graduate students or faculty?
Other faculty — I had a disastrous encounter with Joe Chamberlain. I believe he’s still alive; so I won’t go into detail. But that didn’t work out at all well. My first paper, or one of my first papers, was on Venus, and I had concluded that Venus had a very hot atmosphere — the gradient transfer of Venus. And he didn’t like that result at all.
So you were a graduate student and you had just published this paper and he —?
Well, I was working with him and I studied the atmosphere of Venus. I was a first-year graduate student and I came to this conclusion which he didn’t like at all. And I won’t give the rest of the details, but I stopped working with him at that point.
Okay. Just because the interaction was so negative?
I mean he was mad at me. After I reached that conclusion, he didn’t want me to work with him anymore.
Oh wow! So this was your first year. This was actually before you were working with Chandra?
Right. W. W. Morgan was very influential. I didn’t write any papers with him or anything, but —
What was his work on?
He was a very early stellar astronomer. So the MK [Morgan-Keenan] classification of stars was the fundamental system for stellar classification for many years. And he was a very dramatic and impressive personality. But I didn’t write any papers with him. He was purely observational, but he had a very penetrating intellect and a very strong personality. He died last summer. Let’s see; those were the two most influential faculty members.
How large was the department at that time and how many colleagues, how many other graduate students [were there]?
There were probably about six or seven faculty members, and maybe twice that number of graduate students. It wasn’t huge, but kind of a standard size for one of the larger astronomy departments, actually. I don’t think any of the students I was that influenced by. I became friendly with Yousaf Sabuti, who is now in Iran and is still a scientist; and Jean Laqueue, who was a French Canadian. And Morris Aizenman. I wasn’t particularly working with him, but I knew him. He’s now at NSF. He was in my class. He’s an NSF astronomer. And I’m trying to remember… who else from that period? Those are some of the names I remember.
And when did you marry? Is your wife’s name Alicia?
Yeah. That was before graduate school. It was my senior year in college.
And she’s a poet. Is that right?
And was she also pursuing an academic career, or was she publishing?
So was there — we call it now “the two body problem”? I don’t know what they called it then.
[laughs] Oh, it was intense. We looked around for graduate schools together. We had gone to high school together and we were both in the Boston area. She went to Brandeis and I went to Harvard. And then, we got married in our senior year and moved to Washington. Then, we looked around for graduate schools and we picked Wisconsin and Chicago, and lived in Madison first. She got her Ph.D. in Madison, and I got it in Chicago. But they are close enough together, so you can work it out.
Right. And has that been a continuing —?
Yeah. Then after that, she got a fellowship and I got a fellowship which we took to Cambridge.
Okay. So you came to Cambridge immediately after, in ‘64 or ‘65.
And so she got a fellowship here as well?
No, she got one from some woman’s organization. I’ll remember in a minute. American Association of University Women, I think, or something like that. She got a fellowship she could take anywhere, and I got an NSF which I could take anywhere. So we both came here. Then we had one child in Wisconsin and one child born here.
And the one that was born here is now professor of [astronomy] at Maryland.
Wow. Okay. Great.
Then we had to look for jobs. Back in the United States, postdocs had one year. Now they have three. So we looked all over, and the best combination we could find was the Princeton area, where I went to Princeton and she went to Rutgers.
And was that position originally a postdoctoral position?
I’m trying to remember. I remember I had a choice of the Institute [for Advanced Study] or the university. I chose the university. And I think I was a postdoc for a year.
I didn’t remember from what years.
[Looks at biographical notes] Doesn’t actually say, but I believe I was a postdoc; research associate and lecturer. So I was a research associate. And “lecturer” means that I occasionally lectured in a course – ‘65 or ‘66. And then Lyman Spitzer asked if I wanted to be considered for a faculty position. So then, I was in a faculty position for — I guess it looks like — only two years — assistant professor. And then I was promoted to associate professor, and was an associate professor for only, I guess only three years. I was [then] promoted to a full professor. That was very rapid at that time.
Yeah. That does seem very rapid.
Compared to now. So one year postdoc, two years assistant professor, three years associate professor, and then professor in ‘71. So I was full professor about six years after my Ph.D. Some people are ending their second postdoc now.
Yeah. Let me go back to the research you were doing. What was the main draw of coming to Cambridge? Who were you looking forward to collaborating with, or were there particular people that you wanted to work with?
I don’t think I knew that much about it. And I really just asked Chandra about good places to go.
Were there other locations that were competing?
Yeah, I looked at all the usual places, but this was the one that looked best. And it had a record of being the best place in theoretical astrophysics, historically. I think the competition has always been Princeton and Cambridge — the best two in theoretical astrophysics — with, maybe, after that Chicago; and then Cornell, Berkeley, Harvard — something like that, probably. So really, it was the logical place to go. He had ties with it, he recommended it. But I don’t think there is anyone in particular who appealed to me.
After you arrived, who did you work with most closely and what —?
Donald Lynden-Bell. I was his postdoc. And that worked very well.
How would you characterize your working relationship with him?
When I came here I was working on a project. I was still working on my thesis project, like everybody. And then, it was sort of lively teatime conversations, and together we wrote a paper — which is mainly his work in this case — on a variational principle to find equilibrium. It’s a much-cited paper, but I would give him most of the credit on that. Then, I tried to apply it computationally: to find the equilibrium of rotating white dwarfs, because I am interested in rotating stars. And I spent most of my year trying to do programming for this problem, and so I had been [inaudible phrase] computational.
At that point, it must have been working with punch cards?
It was punch cards. But it was before punch cards. It was paper tape.
Paper tape. So how —?
Or using the insert machine and coding in autocode. These are all things that you — It’s basically machine language. And paper tape was unbelievably cumbersome, because if you —
It was one long sheet of tape or —?
One long row of paper tape. Just to give you an idea. The first thing, when you went into the laboratory, which was at the DAMTP on Silver Street, was a fire truck. Let me explain to you [why] the fire truck [was there]. The computers were vacuum tube computers, and there was paper all over. And it was a wooden building — the insides were wooden, and it was brick on the outside. And they had fires all the time. And you would have paper, about this size was the keyboard [into] which you’d punch the little holes that held the tape. And then, you’d get a roll of paper tape, maybe about like this size, a big roll of scotch tape. And you’d run that in the machine, and then it would run your program. If inside you made some mistake, there were no screens, to output. I can’t remember how things were output. Oh, any output you would get was on the teletype.
So you would have to click-click-click-click. It would give you some output at the teletype. And then, if you decide, “Oh I made a mistake. I want to change something. A 2 to a 3,” at, say, someplace in the code, you would have to do the whole thing over. And when you got to that 2, you would stop the machine. So you copied the whole tape. When you got to that 2, you would change it to a 3 and then you’d copy the whole rest of it.
Wow. So there was no way of splicing the tape or anything?
No. Now, then you would have to run it through a comparator to make sure that it hadn’t made any errors, because when we type things, we make errors. Well, very often the comparators ran fast, they would tear it.
So then you would have to start —?
Right from scratch.
Wow. So was this one of the —?
It was so unbelievable! You could literally spend a day changing one number. So it did teach programming skills. You didn’t want to make mistakes.
And the only way you could see your code was to have it printed out in teletype which, of course, took a very long time. I do a little programming now; but when I see the young people doing the coding, it is just so much easier, when you see it all on the screen, you can go to whatever line, and change it, run it, etcetera, etcetera.
Was it unusual to be doing these sorts of numeral calculations?
Were you one of the earliest to —?
Yeah. Yeah, I was one of the earlier people to do it. And every place had its own language. There were two languages in Cambridge. There was no FORTRAN or anything.
So this was all before that.
Had you done any programming at Chicago, or was this —?
Very little. I just picked it up.
And was this something that, just given the nature of the problem, you knew you had to do numerical simulation?
You had to do it that way. Yeah.
So what was the problem again? You were looking at rapidly rotating stars?
White dwarfs, because they have an easy equation of state.
And so that was the main work. I didn’t work with any of the other people here, but I was enormously impressed by them. I shared a room with Ken Freeman, who went back to work in Australia. Martin Rees and Steve Hawking were graduate students. Richard Ellis, I think, was just finishing up. Faulkner, Eggleton, Malcolm Longair. Peter Goldreich had just preceded me here; so I had his bicycle. We overlapped a little bit. But, I was so impressed with all the people. I thought, “Gee whiz! They seem to be much smarter in England than they are in the United States.” It looks like it was a local delta function, because it’s not every year that Cambridge has, as finishing graduate students, Longair, Rees, Hawking, Ellis, etcetera.
Did you interact much with Dennis Sciama?
No, I didn’t. He was a very famous person, but I just didn’t happen to because I wasn’t his research associate. Doug Gough was a Ph.D. student too at the same time. He’s department chair now. So I interacted mainly with Donald and with the other students. I can’t remember others, but they all seemed so smart to me. I thought, “Well, they are a bit brighter than the students I run into in the United States.”
At this time, other than your rooming...
Hoyle wasn’t there. I remember, I used to joke that Ken Freeman and I each had half the Plumian chair, because we shared his office.
So has was on sabbatical then?
No, he was feuding with the department, and so he never came in.
Okay. So you got to share part of his office. That’s interesting. Do you have any anecdotes about interacting with the group when you arrived, or anything that — were you having informal gatherings regularly? You mentioned just talking over tea.
Oh yeah. Yeah. It was mainly that. One thing which I remember was Doug. I don’t know whether it was he and Donald, or whoever, had this idea that convection would expel [inaudible word]. So they were doing experiments on this basis in, what was bachelors’ lab, and with boiling liquids. And they would drop Alka-Seltzer. With liquids, they would drop Alka-Seltzer tablets in it to make it so it would convect, and then to see if it really sped up or slowed down the container. And I thought this was quite bizarre.
And I assume it didn’t, or —?
Well, I think they got — it was positive results at the time.
Okay. So they were excited about the —
Yes. That’s just one anecdote I remember at the time. What else do I remember at the time? I really just remember lots of [inaudible phrase] conversations with people. But I don’t remember too many that were specific, in that year.
Do you remember what you thought, other than your own work, were the exciting areas in astrophysics at the time?
I think I was mainly interested in star stuff, because that was a period where there was a lot of [inaudible word]. I don’t know when B2FH [Burbidge, Burbidge, Fowler, and Hoyle, Reviews of Modern Physics 29 (1957): 547.] was; probably pretty close to the same time.
Yeah… It was before the background radiation.
Right. But trying to understand what made stars shine was the big science of the time. And Chandrasekhar at Princeton had worked with Hoyle on the red giants at about that time. So the two subjects which interested me were stellar structure — since I was doing work on degenerate tsars and white dwarfs — and stellar dynamics.
Could you just describe briefly what the main lines of thought were in your work on rotating stars? I’m not all that familiar with it, unfortunately.
Okay. I still think I understand some of these things better than what’s been done. People had the idea that if — and there are still papers written on this, which struck me as really strange — if you took a star and you made it rotate faster and faster, it would spin off matter at the equator. If you hold things together with mechanical forces, that’s true if you spin a rod fast enough it will break and the things will fall off. And that’s the minimum. But that is not true of fluid systems. What happens is, before they get anywhere near that point, they distort and their moment of inertia increases. And then, for a given angular momentum, they rotate more slowly rather than less slowly. So, if I give a fluid star more and more angular momentum, first of all it stays circular and rotates faster and faster. But then it begins to get distorted, it rotates slower and slower. So the outer planets, which have more angular momentum, go around the Sun slower than the inner ones. And so, there was a fundamental misunderstanding of the effect of rotation on stars which, I would say, of the community of people working on this, about a third still don’t understand.
It’s amazing to me. I just saw one wrong paper on the web from some General Relativity group doing this. They simply don’t understand it because what they do is tend to work in a rotating coordinate system, which has frictional forces and it’s all right if you’re in equilibrium. But it doesn’t represent — at some point, it’s true you can’t make a model which is uniformly rotating. But that doesn’t mean that you can’t put more angular momentum into the system. It’s just that the systems are not uniformly rotated. The solar system does not become Keplerian.
It’s not that they spin matter off the equator. They just become r(-1/2). Well, so I wanted to show that with demonstration models because I had said that in lectures and things. The other thing that excited me was that the Chandrasekhar Mass Limit struck me as wrong for the following reason. If you have a self-gravitating object and you have just the pressure of gravity, then if you increase the mass beyond a certain point, it acts like a polytrope of index three halves, where both gravitational and thermal energy goes 1/r. So it just contracts and contracts and can’t find the stable equilibrium, and even with General Relativity would just collapse. With General Relativity, it collapses at a finite size because the degeneracy pressure can’t hold it up anymore. But both those energies are going as 1/r — one always beating the other. But if you have rotation, then for fixed angular momentum the kinetic energy goes as — let’s see, E = Iω2; j= Iω; so E= j2/I, which is j2/ mr2.
Where I is the —?
Moment of inertia.
And therefore, the kinetic energy for fixed angular momentum goes 1/r2. So if I have a star of any given angular momentum, there is always an equilibrium which you can come to, whatever the pressure is. So there is no change of the Chandrasekhar Limit for a rotating white dwarf, even if any momentum is ε. The lowest contract to a point where the kinetic energy plus the thermal energy will balance, but the kinetic energy may be dominated.
So basically, you thought that this is a way to avoid gravitational collapse if you have enough angular momentum?
Right. Well, any amount of angular momentum.
Right, there will be some limit at which —
Right. Now, it could be that if the angular moment is too small, that you only reach that when you are in a general relativistic domain, in which case it would have already collapsed. But in Newtonian sense — and the change of cycle limit was Newtonian, it wasn’t general relativistic, initially — there is no change of cycle limit at any angular momentum. So I was trying to demonstrate that. Then the work I did on polytropes was to show that in general, there is no mass shedding and, once you specify the star at a certain angular momentum, you can always find an equilibrium. So I wasn’t being anti-Chandra, because I did sequences of polytropes, which are like the Maclaurin and Jacobi sequences that he had worked on, but for compressible objects. And I used this code and other things to do that.
Okay. So that’s where you started doing the numerical simulations.
Right, because I had to do that numerically. So the idea was to show that rotation qualitatively changed things, and that the simple-minded pictures that you had of uniform angular velocity and mass shedding at the equator were qualitatively wrong. So that was my first several years of work.
Right. And that was part of the received view in the community, that you could have mass shedding? You said that it still is.
Yeah. It still is, which I can’t understand. It still is. Now, the bizarre thing is that it’s not quite wrong, as it turns out, but for reasons which no one would have guessed then. And that is that when you become rotationally distorted, then if you look at the stability of the objects, to non-axi-symmetric models, for example —
So the axi-symmetric are just rotationally symmetric?
Right. So we have been talking only about axi-symmetric objects. But if you have them with enough angular momentum, so that they are significantly distorted, there is an axi-symmetric equilibrium. But, they can become unstable to either m=2 two or higher modes.
Okay. So basically some types of perturbations will —
— be unstable. Now when that happens, those perturbations can interact non-linearly with the whole star.
So you get sort of a feedback?
Yeah. What happens then is that you get a bar which tends to wind up, because the inner part is rotating faster than the outer part. So then, on the inner part you are getting a bar which is going faster than the bar on the outer part, i.e., of a spiral. When that happens, you fall out of a non-linear domain. But the inner, more rapidly rotating part transfers angular momentum to the outer part. It’s just as if, let’s suppose, you had a planet in orbit around a rotating bar. There is some angular momentum transferred to it. So the outer one absorbs angular momentum and moves out. And so, now that people can do fully nonlinear time [unintelligible word] calculations, when you reach these points you actually do nonlinear instabilities, which can cause objects to grow discs. And then, they actually do carry off angular momentum in the discs.
Mm-hm. Sort of a very specific case —
So, there is in fact some truth to the original picture, but for reasons that are completely different from the original idea. What was positive energy, the energy became positive and the rotating [unintelligible word] will fly off to infinity. And then, this tends to occur for reasons of stability: not too far from the place when previously people would have said you have the mass shedding. I haven’t done this work; other people have. But if you look at the best model of time-dependent non-axis symmetric work, you do find that if you give more and more angular momentum, they tend to grow in equatorial plane, with the angular momentum being carried out in spiral waves. And this may be relevant for the formation of the solar system.
Interesting. Now, it also seemed to me that as you were talking that this would be perhaps applicable to very large systems, and I was wondering if, say, galactic dynamics?
Right. So I tried to understand — that was the argument of the massive halo work with Peebles.
Okay. I was going to ask about that, whether this sort of focus on stability of dynamical systems worked naturally into —?
Right. And so, then I discovered that the same kind of instabilities, which Chandrasekhar and others — Maclaurin and Jacobi before him — had found in hydrodynamic systems, also would in uniform density hydrodynamic systems would apply to compressible ones, polytropic ones. That they would become unstable at a similar point. And so developed a criterion which was basically kinetic energy over gravitational energy. When they had reached a certain point, they became unstable to these bar models. Now, there are better treatments of that instability now. I had used that approximate treatment, which Chandrasekhar had developed using the Virial Theorem, and applied it to the compressible objects. So it’s close to right, but it’s not an exact treatment of the instability. And I found that in the same way, they had become unstable to more modes.
So this was maybe late sixties, early seventies?
Yeah. Let’s see, what are we up to now? So, the Lynden-Bell group was '67 — “Stability of Differentially Rotating Bodies.” I was back at Princeton when it finally got published. Then the method was '68 — self-consistent field method. Then, I finally got the codes to work and did massive white dwarfs with Peter Bodenheimer in ‘68.
So that was the product of the career that you started here in Cambridge?
Right. Finished at Princeton. Not much cause for very much more modeling. So then I wrote a sequence of papers. I could then do non-spherical stars, so then —
Is the sequence the rapidly rotating stars?
Yeah. I don’t know whatever ever happened to paper 3. There’s 1, 2 and 4.
Yeah. I was wondering about paper 7, too. I was just looking through. Maybe it was in conference proceedings or something like that.
Then I did magnetic ones with the same method. So once I could do non-spherical objects, then I could do magnetic ones. And since for rotating stars, you can consider angular momentum and look for the equilibrium. For a magnetic star you can consider flux, and expand it or contract it as needed. It did the same thing. And then, since I was making them, pulsars were discovered. And I said, “Well, I can make rapidly rotating magnetic objects.” And so I had the idea — it was independent of Pacini — for the magnetic dipole on pulsars, because I had been working on rotating magnetic stars.
And so basically the equation of state was different but a lot of the —?
Well, in fact I tried to explain it as a very rapidly rotating white dwarf. And the first few were found with periods of many seconds. So that was a possibility.
And when the first millisecond pulsars discovered?
Well, the Crab was the first one that was found which was in that domain. But, my first paper — which was the first theoretical paper on it, so that’s Nature 1968 — I said, “Hey, the pulsars could be rapidly rotating stars with magnetic fields on them.”
Just an apology. I didn’t realize that your work on this was so —
Oh yeah. No, no, that was “seminal.” I don’t know whether it was before or after, but it was independent of Pacini doing the same thing. So for years, I was known as a pulsar person. And you probably didn’t know that?
I saw from your CV that you had worked on that, but —
Yeah. I think my first prize was the Helen B. Warner Prize, ‘72, from the AAS. I don’t know — if you look up the citation, I’ll bet it was for exploration of pulsars.
Okay. Yeah, I actually didn’t look up the citation. Just a question about this research. I notice that you do have a lot of references to specific objects, and I was wondering, for the observational work on that, were you ever doing observations yourself or were you —?
Not on pulsars, no. I tried to do observations on white dwarfs to see if any of them fit this; on known white dwarfs. And we found a type of white dwarf which I didn’t understand at all, which is now — I’ve forgotten even the name, but it’s now a classical type of pulsating white dwarf. They were rotating, the ones we found. They turned out to be. But see, what confused me was there were no — we did what was really the first Fourier analysis of very rapid time variations in stars. I was not at all responsible for the technology. That was Jim Hesser. But I motivated it, and so we did observations at Princeton, of stellar fluctuations, and to look for periodic fluctuations. What we found was aperiodic ones. I should look it up, but there are now known to be definite kinds of — I think they’re — convecting white dwarfs. But they show peaks in the power spectrum, which we found. And they are, I think, the equivalent of the RR Lyrae instability strip, but for white dwarfs. So it was an experimental discovery, but I didn’t properly interpret it at the time. I was the first one to find it but I didn’t interpret it correctly.
So you were doing this directly observational work. How large a component of that research, how long —?
It was a lot of my time. I always found observations frustrating, and to analyze them. So I did a little bit of it. But you’re always at the mercy of the weather, and the instruments and everything like that. I prefer pure theory. Then with Jim Gunn, I did this paper on the magnetic dipole radiation in pulsars in Nature, which was crystallizing that as a magnetic dipole model which was, as I said, independent of Pacini. So that’s what we were known for. Then I wrote —
And was Gunn at Cal Tech at that time?
He was at Princeton at that time, then as a postdoc. And then Jim and I wrote several papers on pulsars.
I noticed that he was a fairly frequent collaborator.
Right. And that was the theory of pulsars at that time. So this is way back when. So we did acceleration of high-energy cosmic rays, Phys. Rev., which was a much cited paper. I still don’t know whether it’s true or false.
So what was the main claim in that paper?
That you could accelerate cosmic rays from the radiation from pulsars, the low-frequency radiation. A superficial plausibility is that there are. The Crab Nebula is filled with cosmic rays. So it’s plausible. Whether the high-energy cosmic rays, a significant fraction of them, come from pulsars, I think we don’t know to this day. But that was the argument. So we did the acceleration of cosmic rays from pulsars at that time. So there was then the acceleration of high-energy cosmic rays from pulsars. And then “Do pulsars turn off?” was on the finite lifetime which we discovered. Basically, the spin-down ages have given you a short time, a million years, whereas the galaxy is a billion years. So something is making them die. And we did the first statistical papers on them. So then I wrote several papers with Jim on pulsars.
Just a background question. Was Jim actually your postdoc at the time?
No. We were of the same age approximately. And so, he came as a postdoc. I think I was a young professor by that time. Then, he and I enjoyed working with each other a lot. Then he went to Cal Tech. And then later on, when I was a senior person, I invited him to come back to Princeton.
And at this time, did you have many students or other postdocs that you were working with?
Well, I worked with — I’m just looking at these papers — I worked with Jim Mark, who was a graduate student, on the self-consistent field method; with David Hartwick, who was a postdoc; with Peter Bodenheimer, who was a postdoc; with Jean-Louis Tassoul, who was a postdoc; with Jim Hesser, who was a postdoc. I think they were mainly postdocs hired by Lyman Spitzer.
Okay. And he was the chair of the department?
He was the chair of the department. But I had no special funds at that time, so I didn’t have a choice. Then, I worked with Jim who was a postdoc. He was not with Lyman. I don’t know under what circumstances he came. And that was very, very exciting for me, working with Jim on the process. So that was my first, sort of, big hit in astronomy. And I think that if you look at the citation of the American Astronomical Society prize, it was probably that work with Jim on pulsars which was the standard theoretical interpretation at that time. Worked with Rich Gott on a problem. He was a graduate student at Princeton at that time. And then – this must have been from a visit to Cambridge — a paper with Joe Silk and Martin Rees in 1970.
I didn’t realize. Would you have just been on a brief visit?
Yeah, probably a summer visit.
How would you characterize your interactions with other faculty at Princeton? Eventually, you started working with Peebles on the question of dark matter halos. But before that who were your closest collaborators?
Martin Schwarzschild I did a paper with, but he was an important mentor. And so was Lyman Spitzer. Those were probably the two most important. And the most important colleague was Jim Gunn.
Okay. In what way was Schwarzschild an important mentor?
I would take ideas and discuss them with him and listen to him, and similarly with Lyman. Martin’s interest was in stellar interiors; so I would talk to him about stellar interiors. And Lyman was more interested in stellar dynamics. But since I was working on rotating stars and things like that, I was a little closer to Martin’s work.
You’ve said that you’ve considered Princeton and Cambridge to usually be the strongest places for theoretical astrophysics. Was there a general Princeton style of research, as opposed to, say, Cal Tech, which is usually regarded as more observational work?
Yeah. If you go to lunch at Cal Tech, even the theorists that talk about particular objects — you know, so-and-so is observing such-and-such and discovered something — [is such that] it always seems to be the style to work backwards from the observations, or to work on the observations directly. The style at Princeton has tended to be — I mean there are exceptions, but there is a tendency — to work from ideas to physical principles to mathematical expressions of those, to then make contact with the observations. So that I think is what Schwarzschild and Spitzer did, and that’s what Chandra had done at Chicago. If you look at these papers, I was starting with theoretical models for some kinds of rotating stars or magnetic stars, and then trying to find objects in the sky I could associate with them, rather than working backwards. The paper with Jim on —
Do you think that the case of pulsars was maybe a case of the opposite direction? Here you have new phenomena that’s been discovered…
I would say no, because I worked on magnetic stars, I worked on rotating stars, and the first paper — my Nature paper on that — was just applying that to see if I could explain — I’d worked on rotating white dwarfs and then I worked on magnetic white dwarfs, and then my ‘68 paper was a possible model for a rapidly pulsating radio source, which is a rotating magnetic white dwarf — which was, of course, wrong. It was really going from a theory towards the observations. It was close, because doing the same thing with a neutron star works. And then I did that with Jim, probably the same thing with neutron stars to the pulsars. There was one paper that I wrote with Jim, which was analysis of the observations, which was working backwards.
In general most of your observational papers from this time period are more: "here’s the type of effect that you can theoretically —”
There was a back and forth. I always tended to be happier going from theory to observations than perhaps [from observation to] theory.
I was going to ask about the work with Peebles and if there’s anything —
Let me fast forward it to that.
Is there anything we’re missing that you’d —?
So, I did more work on cosmic rays. So that’s been the theme. I worked on cosmic rays steadily. I worked with Gunn on cosmic rays; I worked with Kulsrud in cosmic rays. And much later, the current mechanism for accelerating cosmic rays was started with a paper with Roger Blandford. It’s one of the ones I’m most proud of. It’s “The Stochastic Acceleration of Cosmic Rays.” I think it’s gotten lost in the history, but —
Could you explain it in a little more detail?
If you have shocks, if you stand in the frame of the shock, you have a convergent flow. And then, if you have particles in it, then they’ll bang back and forth between the — so then if there are irregularities in the fluid, they are carried towards the converging shock. If it’s threaded with the magnetic field, then the particles will bounce back and forth between these irregularities and it’s a straightforward Fermi acceleration between converging walls.
So the shock acceleration of cosmic rays, the first paper on them, was the paper I did with Roger Blandford. It’s now the standard mechanism. I didn’t follow it up much, but if you look it was widely cited. That’s one of the things I’m most proud of. But I did other things on cosmic rays. So I worked with Kulsrud on cosmic rays. Then more on rotating stars and the stability of them, and then a whole other theme, which I wrote probably twenty-five papers on, was the evolution of dynamical systems with collisions. Evolution of globular clusters. And there, I found several effects with different people. So this first paper was with Spitzer and Chevalier — who was a graduate student at that time, Roger Chevalier at Berkeley — then… [laughs] I’m very impressed by my early work. Then, the model for X-ray sources. The one that has now become the standard one for high-mass X-ray sources, the first paper on it was with Kris Davidson, “Neutron Star Accretion and the Stellar Wind: Model for a Pulse X-ray Source.” And that was the standard reference.
So that’s an accretion disc model?
No. You have matter falling on the pole of a neutron star, and then you get pulsations because it rotates rapidly. And that’s one of the major kinds of X-ray sources: the X-ray neutron stars.
Or is it X-ray binary?
Yeah. They’re all binaries, because that’s where the mass comes from.
And [unintelligible word] accretion of stellar wind, model for pulsar — the wind is from the other star.
Okay. So you have the mass accreting on the —
Okay. Yeah, I’m familiar. Again I didn’t realize that you —
Yeah. The first paper on that is the paper with Kris Davidson.
So these were very productive years for you in the late sixties, early seventies.
Yeah. Then, the first paper on supernova input into clusters of galaxies, which has become a hot subject now, was [unintelligible phrase] pointed out that there are metals in clusters of some galaxies. The metallicity was known, that there had to have been supernova to do them, and the supernova energy would make it an important component of thermal energy clusters, which people are just sort or rediscovering now to explain the X-rays from clusters of galaxies, the profiles. So that was ‘73.
How would you characterize, say, through the late sixties the major new questions that were being asked? First of all pulsars, with the discovery of the —
Pulsars — everything relating to neutron stars was the most exciting stuff. So radio pulsars and X-ray pulsars were the hot subject. I remember, every lunch for a while was on one or the other of these questions.
And how about other questions — I mean, the discovery of the background radiation, questions of that nature?
That was a bit of a sleeper. Certainly, the people in the physics department knew its importance at Princeton because it was a thing we co-discovered — Peebles and Dicke, and David Wilkinson. But I don’t think there was widespread understanding in the astronomical community of the deep significance of this right away. The hot stuff was the discovery of neutron stars, because it was a direct prediction of General Relativity and things like that. Only in retrospect was the cosmic background radiation of enormous importance, as was light element nucleosynthesis.
But at the time you don’t think the astronomers were aware of it?
Well, I remember reading Peebles’ paper at the time when I was there on the light element nucleosynthesis. He is not given credit for that, but he should. He wrote the first paper on it.
Right. Well, it’s an interesting story in terms of – I’m sure Peebles has probably told you this — but his initial paper was rejected by the Physical Review. Alpher and Hermann had done some earlier work with Gamow that was similar calculations, but they actually thought that you could have light and heavy element nucleosynthesis in the Big Bang. And I forget if it was Alpher or Hermann, but one of them was a reviewer for the paper that Peebles submitted, and said, “Well, here are the papers that have done similar calculations earlier.” And that was the first that Peebles had heard of those papers, because they had essentially disappeared. They were written in the late forties, early fifties and just —
By the way — I won’t quote the name — but the paper on the explanation for the pulsating X-ray sources had a very strange history. I wrote this paper with Kris Davidson and submitted it, and sent out paper preprints of that. I got a preprint back in the mail a short time after. It was, [by] someone else, which was similar to this subject but didn’t have a lot of things that we had. Our paper was delayed forever in getting published. And it finally came out after the paper was published, by this other person who I had gotten a preprint of. But the paper that was published, this other person had a lot of the same things that were in our paper. Of course, no reference because our paper hadn’t been published. And these were things that were not in the preprint version of that same paper, and I subsequently heard — I don’t know with what reliability — that that was the person who was refereeing [our paper].
I was wondering if that was the —
I don’t know if it’s true. It’s an amazing story.
By the way, if you do want to name names you can always take some parts of the interview and declare those off limits.
For the inquiring historian, it would be very easy to figure it out.
Okay. Yeah. I mean, it will be easy to look at the record and see.
So you’ll see that there are two initial papers quoted, and this other person’s has precedence because it was a month or so earlier than ours. But the things in it were things that were in our preprint, that hadn’t been in their preprint. That was an experience which startled me.
Is that fairly unusual, I mean in terms of —?
That’s the only time that it has looked as if something had been stolen by a referee. I couldn’t prove it but — And I have always been very free to talk about my ideas, and I know a lot of people who won’t talk about their ideas until they’re accepted for publication.
So you were fairly free about distributing preprints.
And talking to people about ideas, etcetera. And I always figured this way. Occasionally — and it will be rare — this will happen as it did in this case.
Right. But so that’s relatively rare, so that doesn’t stifle —
Rare. So that’s the loss. And the gain is that — if I do silly things and people find out about them and tell me, “Jerry, you are a great [unintelligible word], you have made some terrible blunder” — that I’ll gain more than I’ll lose. And so, I have always continued to do it. But I have occasionally lost on that, and that was an example.
Are there any other examples while we’re on the subject?
That’s the most egregious, because I had such clear evidence.
Yeah, that is fairly clear when you have a preprint that doesn’t have specific subjects.
Right. And then mine went out and then theirs, and then mine was held up, and then theirs was published with material that was in my preprint —
That hadn’t been in their preprint.
Yeah, that seems pretty egregious.
So let’s see. Now we’re up to ‘73, “A Numerical Study of the Stability of Galaxies, or Can Cold Galaxies Survive?” And the origin of that was I had always wondered — I went through this simple argument with you that stars don’t shed mass, but they become very unstable if they have enough angular momentum to make them very flattened. But if you look at pictures of galaxies, they are extremely flattened. If they were stars and were like that, they’d be wildly unstable. Well, Peebles at this point had a very primitive, probably the first in the world, n-body code — really crude on a desktop computer. And so I was talking about this, and there had been some bad blood between the astronomy department and the physics department which I never understood then, I don’t understand now, but it was of an earlier generation. It was not Peebles and mine; it was the previous generation. And so, I always just ignored that and I was friendly with Peebles. So he said, “Why do you think that?” And I explained all the work that Chandrasekhar and other people had done, and then that wasn’t just for fluid stars, it was also true for compressible stars. And it seemed to me that it would be the same thing for stellar systems.
There is no sort of length scale that would change the —?
No, it’s just a gravitational instability.
It doesn’t depend on anything else. There are no atomic constants in it. All you have is G and M and J. You can get your units out of G, M and J.
So that’s the dimensional analysis that I learned from Chandrasekhar and Landau, Lipschitz, etcetera. So you can make a length, time and mass out of G, M and J. And when using that, if the objects you make the stars are unstable but the galaxies, you say, are stable, or you don’t say anything. So he tried to do that with my experiments, and we found that they went unstable at exactly the same point. When the ratio of the kinetic energy to total energy was more than such-and-such, they invariably went unstable and made bars. But yet you look at galaxies and they didn’t look — there were many galaxies which were not ghostly bar-like but which were very flat. So the point of that paper was: can cold galaxies survive — ‘cold’ meaning little random velocities. And the tentative conclusion was: no they can’t. Well, at least we couldn’t understand how they could. So there must be another component which is hot; so the total system is hot even if the visible system is cold.
Right, and that’s —
So we said there has to be — and the two papers got confused in Peebles’ mind — there has to be within the flat disc — this is not an external halo — within the region of the flat disc, so within 10 kilo parsecs — there had to be another component which had a mass comparable to the mass of this component, but which was more or less spherically symmetric, i.e., hot, held up by pressure. That was an argument for a spherical component within galaxies, but a dark matter component. It was dark matter in the sense that you couldn’t see it. When you looked at these galaxies, they have spheroidal components, but the mass typically associated with them was a tenth less of mass, and that wouldn’t be enough to stabilize them. You needed them on the order of half the mass. So in that paper, we argued that you can’t make a galaxy stable unless it has more mass in the hot component than our galaxy seemed to — or we’d have a bar. But, it looks like our galaxy does have a bar, but it would have a very prominent bar. It would be wildly unstable, not just a few kilo parsec bar in the other parts. So that was the point of that paper. It was great fun working with Jim.
And then that paper tends to get confused with the ‘74 paper, with —
Right. Which talks about [inaudible word] outside.
And that’s where you get the extended halo.
We wouldn’t have thought of that if it hadn’t been for the first paper, because the first one says there has to be a dark matter component inside the disc.
Mm-hmm [affirmative]. And then you started thinking: “well —"
Well why, if it’s inside the disc, why would it stop there? You can calculate its mass per unit area right at the Sun’s neighborhood, and it’s very significant. And then does it just truncate? No, it will continue. How much does it continue? Then you start looking at the observations.
Right. And were you actually looking at rotation curves for neutral hydrogen at this point? What observational signals you were looking at?
That was one of them, but we looked at basically anything that could measure the mass exterior to the Sun. That included satellites of our galaxy, rotation curves; and we quoted the rotation curve of Andromeda, which was the best known one, and it was — I’m trying to remember, I’ll remember his name. Okay. No, I don’t. He was a Lick observer. But he had a rotation curve in Andromeda which went out as far as any model — Burbidge’s or Vera Rubin’s rotation curves — did. And they found a flat rotation curve. We quote it in the paper. And I then gave a talk at the National Academy. It was very exciting to me. And he came up and talked to me, you know, staggered up to me. He was an ancient guy. His paper had been written in the thirties.
It wasn’t Zwicky that did —?
No, it was not Zwicky. It was — I’ll try and remember his name. But he found this rotation very, very early, the flat rotation curve of the Andromeda. So we quoted that, we quoted the timing argument from the fact that Andromeda is coming towards us which is one of the stronger arguments — which was not ours. That was Woltjer’s argument. So we summarized all the different arguments, and may have gotten as far as Zwicky. I’m not sure if we did or not. But all of them were consistent with more and more mass at larger and larger radius. The mass got larger in proportion with the radius; even though it was clear that the light didn’t go. So, let’s see, I’m jumping ahead.
Just a question about the reception. You have an earlier interview with Alan Lightman, in which you described the reception of these ideas as being very negative.
Extremely negative. The most vituperative was Geoff Burbidge.
Okay. Was he editor of the App. J. at that point?
I don’t know what he was. But he was really violent.
Do you have any specific examples? Did you have confrontations with him or —?
I probably haven’t looked over the correspondence, but it had surprised me. You know, I had cordial relations with him, but he was just wild with antagonism to the idea. And in general, the reception was — ‘frosty’ is probably the best thing you could say. It was just treated as an observer for a decade. I did give a talk at the National Academy. I have to find out what the date of that was. It was considerably later. No, it was 1977.
And do you recall that even at that point this was treated as a wild speculative idea?
It was wild, speculative, but by ‘77 I think it was beginning to get some attention.
What brought it more attention later? Was it your original work or were people following it up?
I followed it up in various ways, but I think the Burbidges’ rotation curves. Because in fact, it was very interesting. I may have quoted the rotation curves, but Margaret Burbidge did a lot very good work on rotation curves. And they showed a flat rotation curve. They fitted with something which had a Keplerian rotation curve. And I think I wrote to her or asked. I said, “But look. You found this and it agrees with just what we’re saying. It’s just that you fit it to something with Keplerian.” And she kind of agreed in a sweet way. But their evidence continued to pile up. They did a lot of work on the — And then Vera’s evidence. Probably, it was the Rubin and Ford work which finally convinced people.
And do you know whether they were originally sympathetic to your idea when they started doing this —?
I’d have to look over their papers. I don’t know. It was very much associated with our name for a long period.
I think it’s just sort of an interesting historical point. I think that now it’s often associated more with observational work, which seems an odd irony.
Right. Because the observations — One could check their early papers [to see] whether they thought they were confirming or refuting. I never looked them up. It would be interesting to check.
Yeah, it would be.
Because I know they knew about our work at that time.
Did you personally interact with Vera Rubin though?
Yeah, and we always had very cordial relations. Did I interact with her at that time? I think yes. I think we talked about observations. I don’t know that our paper motivated her. I can’t claim that because I just don’t remember. But she might say it one way or the other in her papers. But I think her work was many years later.
Yeah, it was at least five to ten years, maybe even more.
Yeah. And so, I was going around giving lectures on all of this and talking about it, collecting the evidence. And so I learned something from — I’m going to go to tea right away.
I decided — and this was a very interesting — it’s one of the few things I sort of learned from mentors with whom I didn’t interact. Watching Hoyle’s situation here and subsequently, I made a rule for myself which was, “Never defend yourself.” Because I noticed what happened to Hoyle. He had done one important thing after another. He did things on fragmentation of collapsing objects in star formation; he did things on nucleosynthesis: one very important matter after another. He did things on accretion. Then, he did one thing which was wrong: the Steady State Theory. Previously, when he had done things that had been correct, people had argued against them because he was always very daring. And he had been able to argue them down, and then went on to other things. Well, on the Steady State which, when it was first proposed, was as plausible as anything — I mean, no one knew or had enough information. It was interesting, I think. Well, and again everyone rejected it. Some of the arguments against it were flawed. So he defended himself. And I think he thought it would be a repeat of his previous things. It didn’t transpire that way, because it turned out that was actually scientifically incorrect, whereas his other [work] had been correct. But because of his style, he essentially spent the rest of his life defending the one thing that he’d done that was wrong. And that’s all that anyone every remembered him for. We recently had a conference here, which I helped to organize because I wanted to set the record straight. So I invited a lot of the people here for that conference because I wanted to get the record out there of all the great things Hoyle had done. And he had the bad judgment, I thought, to spend a lot of time defending himself. And so my argument was: “Look, if what you’re doing is right, it will be proven right by science and people will remember that you did it. If it’s wrong, you shouldn’t spend a lot of time defending it. You’ll be just wasting time and destroying your reputation.” Well as I say, that strategy also didn’t always work for me because it often turned out that people thought that what I did was wrong. I didn’t bother defending. Then when later it was rediscovered in some area —
It was forgotten then.
It was forgotten that I had done it. I was the one who had it ‘wrong’ the first time. And then it’s happened to me a couple of times when people say, “But you were just lucky.” And that drives me wild. I don’t know what you can say. “Because no one could have known at that time.” I thought, “Oh my God, what do I say to that?”
That’s interesting, because I think you’re right. I think Dicke is another good example where people end up being identified —
All they think of is Brans-Dicke. And he did a lot of other things.
So I chose a principle which, on average, worked well for me, although, again, it sometimes doesn’t.
Right. Maybe we should stop if you have tea.
Yeah. Yeah, I’m going to have tea, and I think we’re up through —
— your work with Peebles in Yale?
So the second paper was basically a summary of the observations, and it showed that all the observations that we knew of were consistent with the massive halo. And some of them strongly indicated it. So we had Babcock’s rotation curve over Andromeda which went very far out, as far out as any of the later ones. We had the rotation curves of the Burbidges which didn’t go far out, but as far out as they went, they were flat even though they fit them with Keplerian ones. We had the timing argument of Woltjer, which gave the same mass we now have, 2 x1012, based on the fact that Andromeda is coming towards us. We had the satellites of our galaxies and their tidal radii, which gives M/R3 as a function of radius, we had the orbit of — So essentially, we had the clusters, Zwicky argument. We basically summarized all of the data, and I think in fact we had all of the data that anyone now uses except the Cosmic Virial Theorem. All of the current arguments for dark matter, except that one, were in the paper. And Vera is a very good worker. What it did was give a lot more data on rotation curves. Didn’t change any of the other things. And they were consistent with the ones we had, Andromeda going further out, and the Burbidges not as far out.
Were there changes in the other observational work along with Rubin’s that led to a sea change, or was it — why do you think, say, five or ten years after your early work people were really —?
I think people just gotten used to it. At first, they objected wildly. If you were to take a look at the response that people had to Hubble’s expansion of the universe, I think right now there’s only Alper and Burbidge who say that the redshifts aren’t cosmological. But oh, a few years ago there was — I’m trying to remember the mathematician; maybe his name was Sebo — [someone] who was objecting to the Hubble law, and a couple of others. But right afterwards there were many, many who objected to it. And then data came in and everything was consistent, and so people just got used to it, I think. But I think people’s skepticism about new ideas — in fact this wasn’t new because Zwicky had said the same, but it was relatively new — is reasonable, because otherwise you’d slide back and forth and you want to make sure that all the observations haven’t been misinterpreted. That’s why we marshaled many, many different arguments. And I don’t think any one of them would have been persuasive by itself. It’s just that they’re all saying the same thing. So Woltjer had that argument for the timing; Babcock had the flat rotation curve. They both said the same thing: there’s a lot of mass, there’s not a lot of light at very large radii. It’s just when you put them all together in four or five pages that made a fairly persuasive case. And in fact, the other thing was that we also used the radio rotation curves which went out further than the optical ones.
Okay. Are those the neutral hydrogen climates?
Yeah. I don’t know why people didn’t pay attention — I’m trying to remember the radio observer. I’d have to have the paper at hand, but there were a whole bunch of them, and they were as good as Vera’s, and they said the same thing. They were earlier.
And they’re further out, right? So they’re —
They were further out than the Burbidge ones. They are not further out than the later ones that Vera had. But they were all saying the same thing also. So it was all just consistent. Let me go on back to where we were.
We were at about 1975. I think you were about to tell me about gravitational waves and [unintelligible phrase].
Right. Well, I did this early paper with Thuan on collapse of a rotating object which might be triaxial, and calculated gravitational waves. And I became convinced from that, one that collapse should lead to detectable gravitational waves; but two, on the basis of that work, another work I did, that Kip [Thorne] was tremendously overestimating the likely gravitational waves. Now his own estimate —
Was he overestimating the amplitude or the frequency?
Yeah. Both. But mainly the amplitude. So he was overestimating the number of sources that could be detected — by orders of magnitude.
At that time Joe Weber was — this was after he had been criticized?
Right. Joe Weber was stuff I just didn’t believe.
Okay, yeah I know that most of the people in the community didn’t, but was this —?
Well, I didn’t believe it for a very simple reason. I made a side calculation that if the center of our galaxy was radiating as much as he said – not to do the technical observation — then the mass lost from the center would cause the whole outer parts to expand.
That would cause what’s called “decay term” in the stellar motions, which was bigger than the observed decay term.
Another observational consequence which was clearly a conflict?
Right. So the galaxy couldn’t be losing that much. So he just took the flux of energy he said he was detecting from the center, and just said, “Well the consequence would be that the whole galaxy would be expanding.” Now, it’s not an historic thing. Just because it responds instantly.
I never published it because people had technical objections. But I thought Kip was tremendously overestimating. I never like to write negative papers. So I just said that to him privately, and I said that to a number of people in between. And then there was a bet with him in the seventies.
Do you recall the terms of the bet?
Yeah. I said that gravitational radiation would not be discovered before the millennium. And he paid me last year.
That was good. You won that one. So what were the terms? What did he pay you and what would you have owed him?
I think he wanted Playboy. But I didn’t think that was appropriate, and I said, “Well, I’ll give you a case of French wine and you give me a case of California wine.” And he did. He sent me a very nice case of California wine. Took a little prodding. [laughs]
So was he hoping that LIGO would come through and —?
Yeah. But just, there was a tremendous exaggeration. His own estimates have come down so far now — I think it’s one to two orders of magnitude of the likely sources. I now only disagree with him by a little bit. But I think there was a bit of salesmanship in the earlier estimates.
So do you think in order to encourage people to build gravitational wave observatories; you had to make it seem more feasible?
You said it, I didn’t. [Laughs] So let me go on. I probably have been known as a skeptic, but I always think that that’s been unfair because I’m enthusiastic about it. I just thought it was quantitatively oversold.
So you’re now excited about LIGO and LISA?
No, because I think they won’t detect anything. I think they’re still —
Hmm. So LISA as well you think —?
Oh, LISA. LISA will. LISA will.
But LIGO you think will —
No. I think it’s just a waste of money. And it’s a huge amount of money which could go to other things in astronomy or physics. And that was an astonishing project, because when it was designed and built, if it lived up to specs it would detect nothing. His estimates have come so far down. Can you imagine building an optical or radio telescope and saying, “If this works as well as we think, and if there are as many sources out there as we think, we’ll see nothing”?
That is shocking.
That was what was done. And they spent hundreds of millions on it. And that was hundreds of millions from the science budget. If NASA does it, you know they’re going to spend a small fraction on the instruments that goes to science, but that most of the money is going to go to engineering. Well, this was coming out of the same funds that would be supporting postdocs and faculty members, building instruments and laboratories. So why this huge cost? Let’s see. A paper that has had a mixed history but I’m very proud of is ‘75, “Explosive Events in the Early Universe.”
I was going to ask about that.
Now, if you look at what we did, we basically said: “if you look at the [unintelligible word] cluster, metallicity is a third of solar. You knew that from the iron lines that were seen in the X-ray spectra, even at this time. That couldn’t have happened unless there was a lot of supernova output that actually got into the cluster.” And if you then say, “Well, if you take that output and divide it by the number of galaxies,” and you say “how much is there per galaxy?” It’s quite a lot of energy output. And basically we said that that would affect the surroundings. Well, that’s just what Steidel has now found in winds from galaxies at redshifts 2 and 3. So that was a firm prediction which did turn out to be right. We went further in a way, which probably was wrong, so let me go on that.
We thought the consequences of that would make compressive shocks which would lead to further galaxy formation. So the feedback, which is the current term, that we predicted really is there. My guess is we got the sign wrong. That it reduces the density — shock heats and reduces the density — in high-density regions and inhibits galaxy formation that would otherwise have occurred, and that this overwhelms the effect of causing, then, shells which enhance the galaxy formation.
Right. Interesting. So even though the density waves might cause some compression, there is also just the negative impact due to —
Blowing material out, evacuating and heating material. I don’t think the jury’s in, but we were using this as a mechanism, for two [unintelligible word]. One is to heat the cluster gas. I think that one was right, and people agree now. The other was to enhance galaxy formation. I think that was wrong. But the basic idea has now been observed. So Steidel has written any number of papers about winds from galaxies driven by supernova.
So was your work on the phase structure of the interstellar medium at about this same time?
Yeah, I must have passed that.
Yeah. I think we forgot to discuss that earlier. There were papers with McKee that —
That’s very widely regarded as seminal work.
Right. I’m not sure what year we’re now.
Yeah, I wasn’t sure if that was in the late sixties or early seventies that those papers —
Let me just [unintelligible phrase]. Maybe I’m not up to the mark. That was a very long time in gestation, that work with Chris. I had gotten interested in supernova; so I had a paper: “Do pulsars make supernova?” The jury is still out on that one because —
So the standard idea now is that the supernova result from accretion of matter on the white dwarves?
Right, but you have a collapse. And we know after — let me back up. For low-mass stars you’ve got a white dwarf and you could get enough accretion onto it, either within the star or later separately, so that it implodes in a type 1 supernova. For high-mass stars, they make black holes. Intermediate mass stars, three to ten or fifteen, we know empirically that they make neutron stars by the counts where we find them, etcetera. We do not yet, even today, have a good mechanism. When people try to compute it, the stars tend to collapse all the way to black holes.
They don’t eject enough mass. Is that the—?
It’s not the outer parts. They can’t stop the inner parts from collapsing.
So there’s something wrong in the calculations. But empirically, we know they make neutron stars. Now, and so therefore the pulsars which you later see must have been inside those stars because the neutron stars are everywhere. And all we pointed out in that paper is that that process must help eject the envelope of the stars. Since right now what they have is a neutron star which accretes and then collapses to a black hole, they still haven’t included that process because the models tend to be axis-symmetric and non-rotating. And so the added energy from the neutron star, rotational, might be the important element that is causing the envelope to reject it. So we proposed that in “Do pulsars make supernova?” I would still say it’s an open question because the people can’t do it without the pulsars, and we know that the pulsars are made inside them just empirically, even though we can’t figure out how. So that open suggestion is still out there for people [unintelligible word] to supernova. I guess we’re not up to –
Yeah, sorry. I guess I didn’t remember the times except — I was just wondering how your work on interstellar medium was connected to this work on intergalactic medium.
Yeah. It’s ‘77, late seventies. I want to say something about the things with Scott Tremaine. Beatrice Tinsley gave a lecture at Princeton, which I thought was brilliant, in the mid-seventies, where she pointed out that ordinary stellar evolution will make galaxies get fainter. And Sandage and others hadn’t included that in their calculations. Therefore, they treated galaxies as standard candles when they weren’t. And the fact that they are getting fainter with time changes all the interpretation of the observations people have been making, of galaxies and the calculation of omega and so forth and so on. I remember, when Beatrice talked, that I was in the audience and I said, “This is terrific, but I bet I can think of at least one correction which goes in the other direction and it’s just as big. It’s all your idea; but the dynamical effects.” And so that led to the paper with Scott called “Another Evolutionary Correction of the Luminosity of Giant Galaxies.” Well, we talked about mergers and accretion, and that was the first time in the literature — it’s lost now — where people had talked about galaxies merging in accretion, which would be a dynamical effect which — Scott was a student then hich would lead to them getting brighter with time.
Because you’d have a new star formation?
No. There is the first, brightest galaxy in a cluster. It accretes another one onto it; and then the first brightest galaxy in the cluster is brighter. It’s as simple as that.
Okay. Just more stars.
Yeah. Just added stars. And you could calculate the rate at which this would happen, and it was comparable to or larger than the rate in which these galaxies would get fainter by aging.
So you have contradictory effects then.
Right, you have to include them both. So that’s why we called it “another one.” But and then I wrote papers on Peebles’ cannibalism, etcetera. So mergers are now a very, very hot topic, but this was the first time people had actually calculated them. So it was the paper with Scott.
Yeah, I know one of the things that you mentioned in the Lightman interview, which I thought was really important, is that at this point in the seventies people were starting to look at galaxies in a whole new light, partially because they were realizing that they couldn’t be used as standard candles.
And that it sort of forced the community to look at these things much more seriously.
Exactly. So the caricature of the science of cosmology from the twenties to the seventies was a science of two numbers. You would try to find omega and H, and you’d use any objects you could as standard candles to do it. And you’d just assumed that they were there, and they were always there. Then it was really Beatrice who changed it. She led the way, and her work was strongly resisted, because it showed that what everyone had ended up doing was riddled with errors.
Right, this whole observation project.
Which was huge. As soon as that happened, people said, “Oh my goodness. They weren’t there always. You had to make them.” They’re not just evolving, but they had to make them, and so people had to start thinking about theories of galaxy formation, galaxy evolution, cluster formation, cluster evolution. And then you realized there was a whole set of questions which no one had asked before. So that paper of mine was the first in that series to look at the late time mergers of them, and that led to another one — the one with Hausman, “Cannibalism Among the Galaxies: Dynamically Produced Evolution of Cluster Luminosity Functions” — which is looking at the same thing: mergers. We called it “cannibalism”. And then in between is the paper with Chris. And that was an example. Probably we worked on that four or five years. So it was probably published in ‘77.
Why was the working phase so long?
We kept on thinking of new things we wanted to do. It’s a twenty-some-odd page paper. And we wanted it to be a comprehensive theory. So we kept on iterating and in effect realizing that there was something which we were taking from observations which we should have been able to derive to make a self-consistent theorem. I finally completed my part of it in the summer we were in Mexico. My family and I had a summer place in Mexico, and I just sat and worked without any distractions. But that was very exciting. And I was thinking about the intergalactic medium and the interstellar medium at the same time. I was trying to apply many of the same ideas, which were clearly the case in the interstellar medium — the effect of supernova in doing the shock heating — to the intergalactic medium. So that was the connection. The papers ended up being published roughly at the same time, but this work on the interstellar medium had started much, much earlier. Probably, I taught Spitzer’s course on the interstellar medium once in talks, and he had given me his book to — Oh, then there’s one other thing which my name isn’t associated with, but I did the first thing on it. And that was that if you have two stars in a binary system and one expands – now what do they call them nowadays? We called them double core. There’s a new, a typical name for them now which is — I can’t remember. Because Bodenheimer did something after we did it. He and I had discussed it a lot. I think his thoughts are independent from mine. And I think his nomenclature stunk. We now know there really is a phase in the evolutionary binary stars where one gets enveloped and then they spiral together.
Oh, okay. So where one is inside the —
Right. So that ended up with a paper with Peter Bodenheimer in ‘78. But my work and Bodenheimer’s work started a whole field of these double core evolution of stars.
I want to go back to the paper on the interstellar medium briefly. When you were working on this, were there particular observational results that were really crucial in starting your thinking, or was it more that you thought there various theoretical lines of thought that —?
That’s a fair question. I can’t remember the sequence. I should know it. But the Copernicus found lots of hot gas in our galaxy. I don’t know whether that was before or after we predicted that the hot components of the galaxy would be very large. I know that the Copernicus results are the observational basis for saying that there’s a very substantial hot phase. And I honestly can’t remember the sequence — whether we knew that — because it was done in Princeton — and that led me to think about that, or whether it came afterwards. We could just look it up on the web, when was Copernicus launched. So our paper started in the mid-seventies, was published in ‘77. But then here is a paper of mine which had been forgotten. “Particle Acceleration by Astrophysical Shocks” with Roger Blandford. It’s one of the papers I’m most proud of, 1978, in which we – and this is more Roger’s work than mine — we followed up the work that I had done with Chris to try and find out the implications for accelerating cosmic rays in the galaxy. But we hit upon the mechanism — now this Fermi acceleration and supernova shocks, which everyone believes now is the standard mechanism for accelerating cosmic rays. There’s a huge amount more work that has been done on that, but that was the first time that was mentioned.
Right. So there are a number of papers related to the effects of blast waves in these explosive events.
Exactly, and this was one of them where we’d gotten — I think today people believe it is the correct explanation for — Then, I did a paper with Caldwell on the galaxy. People still refer to it because we had much more mass in the bulge. “The Mass and Light Distribution of the Galaxy: a Three-Component Model.” Since I had done this evolution of the halo compartments, I was looking at the stellar halo and I thought people had greatly underestimated the mass in that. So our model had probably ten times the mass Bahcall’s model and everyone else’s model. Now the MACHO results have confirmed that the stellar mass of the halo was much more than people had thought. We were just doing it on the basis of observations. But that other paper was right on that. Then there’s another paper with Blandford called “Supernova Shock Accelerators of Cosmic Rays.” Then, there are papers with McKee and Cowie on that. So they sent lots of papers on supernova equations.
Let me ask you a more general question about the other work that was going on at the time that was exciting. It seems like during this time, people were developing a much richer idea of the structure of galaxies, the interstellar medium, all kinds of a much more fully developed picture. Are there other components of that that you consider to be especially important? I mean a lot of your work was —
Was on that.
But you mean work of other people’s?
Yeah, or just other ideas that you recall being introduced during this time period.
Well, Princeton was pretty much of a hotbed of that, because Spitzer was interested and Chandrasekhar was interested.
Who else was in Princeton at the time that you remember? Peebles, of course, was there.
Right, but he wasn’t particularly interested in this. Certainly in my work I would, I mainly talked to Lyman. I’m trying to think of who else. I think I did talk to Dick McCray a little bit at that time. Those are the only ones I can remember. And I read the observational papers. So here’s the paper which I was most noted for being wrong on. It’s the paper with Cowie, “Galaxy Formation in Intergalactic Medium Dominated by Explosion.” This was the quote-unquote “explosion theory” from galaxy formation. So I took the work I had done earlier in the interstellar medium, combined it with the work I had done with Schwartzchild here on explosions. All of that, I think, was right. And then I tried to see if you could use this to enhance galaxy formation. And I think people thought it was an alternative to a gravitational top down or bottom up — which I hadn’t meant it for.
You just meant it to enhance?
Right. But in any case, I think it was wrong, whether it was the exclusive theory or an amplification. My guess is that it’s a minor effect, and I’m still not sure of even the sign of it.
Yeah, you mentioned earlier that you think the sign might have been wrong.
Yeah. So, but that —
What led you to change your mind about it?
I guess the evidence isn’t there that you have these cool shelves, which is what you’d need. You’d have more of a higher column density in the damp Lyman alpha systems and other things. That, plus the successes of the standing gravitational approach. Hierarchical theories, mainly just abandoned that. I’m looking at it again with students to see if it’s a minor but significant — were we normalizing it on Steidel’s observations. Because the effects are still there. The question is: what’s the sign of them and how big are they. Because what we said about explosions and winds etcetera, from galaxies are really there.
So now it seems like just in terms of shift and focus — I mean at this point — it looks like you’re starting —
I’m moving extragalactic.
Yeah. You’re moving extragalactic and you’re starting to move towards large-scale structure.
Exactly. So then are the papers with Ikeuchi on Lyman alpha clouds, and [I] did papers on gravitational lensing. That’s been a minor interest of mine, but a strong one, all along. And I think we emphasized early on, with Turner, the amplification bias: the fact that as you have lensing, it makes things brighter and then faint objects come into the field of view and you’ll think that there are many more than you would have thought. So you always overestimate the number of lenses.
Oh. So you think that the MACHO searches, where you were looking for a characteristic amplification curve, could just be due to a faint object moving in front?
Yeah. Well, that’s what they think too. But there is nothing that we have to say that people don’t know now. What we stress there, there’s something called amplification bias. If you have a rapidly falling number of sources, and you boost some —
— so now I compare those with this number — Let’s see, do I have it right? Just a second. Let me do it differently, show it differently. Right. Here is the number versus flux of some kind of source. And now, I take some of these and make them brighter by one magnitude. So this one moves to here. Now it has the same brightness as this one. But, I’ve boosted — it may be only one percent of these that I have boosted —
Oh, I see. Okay.
But these lense ones may correspond to 10 percent as many of these. Because there are so many fewer of these.
Right. Because your number is dropping off so quickly.
Right, so rapidly. And this is now called amplification bias. Now Ed Turner had stressed it also, and then we made all the calculations on it. Ed Turner was at Princeton.
I also want to ask you about related developments. So we’re in the early eighties by now?
Yeah. The mid-eighties, yeah.
What was your initial response to inflationary cosmology?
Complete ignorance. I realized the problem, and I thought this was a clever solution, but I didn’t know enough particle physics to understand whether it was a reliable calculation or not. I thought it was an interesting solution to a very important problem, but I just couldn’t assess it. I still can’t. And when I talk to the particle people they say, “Well, there doesn’t seem to be any good theory of inflation.”
Yeah. That’s my impression as well.
Right. And it wasn’t skepticism about being wrong. I just didn’t understand enough. So I’d say that’s still an unsolved problem: how you get the flatness and homogeneity and isotropy. And that was, I thought, a very clever idea; I just don’t know whether people worked it out well enough to consider it firm.
So you will consider it firm when it’s tied to a specific model in particle physics?
Which other particle physicists agree with. As I say, I’m not expert enough to understand it, but I know that no one has produced the calculations which other people agree with.
Right. Yeah. I think Mike Turner calls it a paradigm without a theory, and that there are, I think, a hundred different candidates for it.
And no two people agree. So that’s why I was interested in this recent work of Steinhardt and Turok, because they have a completely different approach to try to solve the same problem. So I thought the exciting thing really was that they realize it is a problem.
Yeah. I’m very interested in that work as well. It seems — it’s tied to String Theory work, that alternative to inflation.
Just one more question on inflation before we go back. Were you surprised by the amount of research effort that has been devoted to inflation over the past twenty years, or do you think it’s natural?
There are fads. And in astrophysics, I mean, I can remember back when everyone was calculating [inaudible word] waves with slip, with viscosity, with I don’t know what! There was a period when you couldn’t go to a Tuesday luncheon without people doing accrete neutron stars observations and theory. Most of these things pass after a while. Sometimes they leave something, a body of work which is used, and sometimes they just evaporate where people don’t find them useful. So this was a bandwagon on this. And what happens is young students begin and they see a lot of people working on something, and they feel they can make a calculation on it and they want to do it. And if it turns out that it’s something important, well, and then they participate in a group effort. What I’m saying is it’s a sociological phenomenon, not a scientific phenomenon. And I think it’s just normal, because there was a period — and we’re just at the end of it now — where most of the best theoretical postdocs were working on CMB fluctuations. And it’s preposterous! It’s a linear calculation. Anybody with an IQ of a bit over 100 can do it. And all of the best people were doing it, and they were competing to see whose code was better, [unintelligible phrase] and the other. Everyone was getting the same results because it’s an extremely straightforward calculation. And you really couldn’t tell if people were trying — we weren’t, but people trying to hire them — you couldn’t tell the difference between one person and another because they had all written all the same papers. And it was just a bandwagon. Now that I think, the summary of that will be incorporated in the literature because it’s fundamentally correct and important. That may not be true of String Theory, but what drives it has nothing to do with the correctness or importance. It’s just a collective human phenomena.
Right. So now you say we’ve started to reach the end where all these postdocs are doing the CMB calculations. So they shifted earlier or later in terms of the calculations? So it seems later you get into very complicated nonlinear calculations.
I think that it’s shifting at the later kind of things mainly, and I think that’s mainly because they get jobs in a number of places where there aren’t observers. And so they want to tie things down to other observations. I mean, the good thing about this was it’s a linear calculation; you can tie it to an observable. So I don’t fault it, but it was just a vast number of people all doing exactly the same thing, and doing exactly the same —
That’s an incredibly rich phenomenology too, and plus with the satellites the next —
Yeah, but the papers were just identical. And so I made a point of — I just did it to be ornery — that I can think of a new effect every few years that they hadn’t included, which was different physics, and was bigger than the difference of different people’s calculation or the precision of the calculations. So the last paper I wrote on this with David Spergel was just published on the Rayleigh scattering, which no one had included. It’s only a 1 percent effect, but it’s bigger than the differences between what different people were getting for the same thing, and the supposed accuracy of the calculations. So my guess is there are still other physical effects that they haven’t included.
Right. And that people should be working on that.
Which people should be working on. Let me go on because we’re running out of time.
“Standing Shocks in Accretion Flows on Black Holes,” there’s a whole series of things on this. But I got very interested early and I’m still writing papers on the feedback. Most of what people do when they do models for quasars is they do with gas coming in, and there are classical disc solutions and there’s ADAFs, CDAFs and all these different kinds. And then they calculate the radiation emitted by them, compared observations. What they don’t tend to do is [look into] what the effect of this radiation on the gas itself [is]. And what I found in a series of papers, that I know has been particularly interesting to me, is they qualitatively change the solutions because the gas preheats.
And there is so much radiation coming out.
Right. And that it completely changes things. So I’ve been the skeptic in the wings. I just talked to Monica Brenner when she was visiting here last week. I said, “Shouldn’t you put this in? Won’t it change things qualitatively?” She said, “Well I’ll look at it,” and then I got email from her; she said, “Yes, you’re right. We’re going to do it.” But I’ve heard that again and again. So the field has essentially ignored this, and I guess my first paper was 1985 on this, with Karmen Chang’s student.
And so at this point you were also getting very interested in large-scale structure.
Right, right. So that’s a whole other side topic that I got interested in. Continued working on evolution of n-body systems, and one of the things that I stressed was that the effect of the tidal fields qualitatively changes things. Because people had always left them as isolated systems. And I did this mainly with students, so one of the early papers with Hyung-mak Lee —
It’s like the tidal effects of a nearby system on the —?
Right, but just the tidal effect of the whole galaxy on a cluster.
The moral, I think, of it is you don’t have to evaporate source to infinity, just to a tidal boundary. And you might think it’s a minor effect, but it qualitatively changes the evolution of the system. And then in fact they go through the plane of the galaxy. I walked into Lyman’s office once and said, “Did anybody ever include this?” and he said, “No.” And I ended up writing a paper with him on this. It’s a tidal shock which heats the cluster. So there were several papers on that.
And these were n-body simulations, computer simulations?
Mainly they were using [unintelligible word] codes and things like that, but it’s a statistical approach to the same thing. Then, a paper which was very widely quoted — “The Proximity Effect: Quasar Ionization of Lyman Alpha Clouds.” This was with Duncan and Bajtlik. The idea is simple. If you have a quasar here along the line of sight, and you are here, right around the quasar there is an H2 region; so everything is ionized. At very great distances from the quasar, let’s say halfway, it’s in the meta-galactic radiation field. Well, there is some place along this line of sight where the two are equal. And then from there on in, the quasar dominates and there will be many fewer absorption lines because the gas is all ionized.
So if you look at a quasar absorption spectrum and see where the transition is, you can tell what the metagalactic ionization flux is. And this has become the standard way of getting the cosmic background radiation field. So the quote-unquote “proximity effect” of quasars – that was an idea I had in ‘88 with Bajtlik, and now it’s used by everyone to get the metagalactic radiation field. Let’s see. What am I up to? One paper with Simon White; then there were more papers on self-consistent spherical accretion around black holes, where we were trying to include this effect, and more with Park — “Spherical Accretion of the Black Holes: A New Higher Efficiency Type Solution” — again trying to include the outflow. I had written this earlier paper with Jim Gunn on pulsar populations, and then I wrote another one with Ramesh Narayan, where we said there were multiple populations of pulsars, and I think that’s held up pretty well. That’s the only paper I wrote with Ramesh, 1990. In — 1990 is the first paper on cosmological hydro simulations.
And I had gotten very interested. I had done computations since I began, and many of these were computational astrophysics problems. I got the idea that people had that n-body results, but no one had included hydrodynamics at this point in cosmology. And I couldn’t understand why they didn’t, because the codes that people had done — and it was beautiful work by Davis, Frenk, Efstathiou — whom I am missing — there are four: Mark Davis, Carlos Frenk, George Efstathiou. I’m missing one. Oh, Simon White. They really started the whole field of cosmological populations, but they only did dark matter. And that’s clearly very important and it’s the dominant thing, but you can’t see dark matter. So you might as well include the components that you can see, either in absorption or emission like gas. So I had this idea, and I had done a little bit of hydro in various different ways. So I thought, “Well, let’s put that in.” And there were no codes; so I got a code from a civil engineer, Anthony Jameson, who was a very famous hydrodynamicist. And we coupled that with a theory that existed for doing the same things for dark matter. And the first paper was with Rennie Cen, who was a graduate student there, and Jameson and me in 1990. And that was one of the first papers on doing cosmological hydro.
So what sort of effects did you see immediately that were really different than —?
Well, you would shock heat the gas. And so you could produce X-ray clusters and things like that.
Okay. And I know later — this is from your ‘91, the book on large-scale structure — you also emphasize that you were working on nonlinear positive energy perturbations?
Is that the same —?
It’s the same point again. The paper with Simon White was on that. And I still think that this is something that people have omitted. I got to thinking about supernova calculations, and that’s a positive energy perturbation in these explosions.
Whereas most people’s image of how you get cosmology was the [unintelligible word] picture of collapse of objects. That one is certainly right. But if you take Gaussian perturbations, which is now what is the standard picture, there are as many negative density ones as positive density ones. There is as many positive energy ones as negative energy ones. If you do a numerical calculation, you automatically include both because you use Gaussian initial conditions. However, so to speak, all the heuristics, all the explanations always look at all the collapsing parts, not the expanding parts. And so we found the first expanding one; that I did with Yahil. And then Ed Bertschinger did in his thesis with me on that. That was his void solution. So I had always stressed that there was a parallel and equal importance to the positive energy negative density perturbations in cosmology.
Right, because you do see voids.
Right. And then that produces shells. If you take two spherical shells that intersect, you get a curved filament. Well, in the simulations that’s what you see largely. And then a filament and a shell intersect and you get a lump. Well, a large part of the structure you see is better described this way than collapsing objects, and it just hasn’t gotten into people’s vocabulary. It is, of course, in the simulations. And so I’ve stressed that, I would say, to almost no effect.
So there is sort of a disconnect between what people are actually doing with their models and then how they —
Describe them in words. And so I stress the importance of that other description. That’s half of the problem.
That’s interesting that it’s not only how you model but how you read your models and think about them.
Yes. So, the first paper I did on that, which was the first expanding one, was — and that was before Bertschinger’s thesis. It’s way back. With Ikeuchi and Tomisaka, I did the hydro one in ’83 — “The Structure and Expansion Law of a Shockwave in an Expanding Universe.” It was the paper with Schwartz and Yahil in ‘75, and that was the first positive energy perturbation. We actually presented the solution there, which Bertschinger had in his thesis. But we just quoted without results — the power law expansion. Okay. I’m up to ‘91.
So when did you become interested in superconducting strings and what led you to be interested in that?
I guess I passed that.
You mentioned it in your ‘91 book.
Right. It was clear that you couldn’t really make it work quantitatively to do much on galaxy formation with feedback from supernova.
Was it because there just wasn’t enough energy?
Wasn’t enough energy. The effect is there. As I say, it’s now been found. But it wasn’t. And Ed Witten had written papers about superconducting cosmic strings, and he had written one paper which talked about the cosmological effects and how you would see it in a microwave background and whatnot. And I saw this paper and I thought, “Hey, he missed the most important effect – the fact that you are dumping in huge amounts of energy, and that this will dramatically blow holes in the medium.” And I did not have [enough] knowledge of particle physics. And Witten is a great physicist. I said, “Ed, if this is right, it will have incredibly important consequences.” He said, “Well I don’t know if it’s right.” I said, “But you’re proposing it.” So, I then wrote a paper with him where we tried to find out the consequences. Now, you can turn around and say, “Those consequences aren’t there.” So if there are superconducting strings, cosmic strings, they have a less mass per unit length, etcetera, than he was proposing because the ones he was proposing would have dramatic effects. So basically, I worked out the astrophysical consequences of that. And the fact that these consequences weren’t observed meant that it’s not there. I have often used this as an example because I often think: you never find the theologian who said, “I was wrong.” But I think that in physics, you can be wrong. You can propose some things such that “if this is true, then the following should happen.” Then you go out and look for the following. And it didn’t happen. Then you have to say, “I was wrong.” And then if you have a physical theory that can’t be wrong, that there’s no experiment which could prove it wrong, then it can’t be right — it’s empty. And this is my own human test case, where the work I did with Witten — I mean it was crank mathematics, but it was wrong. We were proposing something which didn’t happen.
And I think more broadly, the other topological defect theories ended up being wrong.
Yeah. They’re just not right. And I did want at this point — it’s interesting. “A Hydrodynamic Approach to Cosmology: Texture Seeded CDM and HDM Cosmogonies.” I tried to work out what they would be with Cen, Spergel and Turok.
Then, this was a very interesting paper which has been very widely quoted. Whoops. Whoops.
Shall we close?
Yeah. Then in the same way with Peebles, I had tried to summarize the evidence, and with no new real science but just a good hard look, and had come up with the conclusion that you need a lot more dark matter than people had been thinking about. I did something similar with Paul Steinhardt.
This was in 1995?
Right. Because I had written papers with a variety of people on tilted cold dark matter [unintelligible phrase] this and that. But basically, I would take a model and put it through the machinery of our simulations to see how well it agreed. And I had come to some fairly strong conclusions about what was viable and what wasn’t. Paul is better at statistics and also had an independent physical view on this, and could understand where you could get cosmological constant, i.e., quintessence, from. So I very much enjoyed working with him. And we summarized the evidence, very strongly, for a flat, lambda-dominated universe at that time.
So this was like your work with Peebles in that you didn’t go away with strong presumptions about this, and did a more or less neutral survey to see —?
Right. And tried to be unbiased on what was the best evidence and what it indicated without saying what we would like.
And it came out very strongly. It indicated that omega was pretty close, in a manner was pretty close, to what I had with Peebles. With Peebles it was .2. We came up with .3 I’m now drifting back to .25. So it’s always been in a fairly narrow range. And you solve an awful lot of problems if you have a cosmological constant or quintessence. There’s another paper. Quintessence is much better motivated physically, but it’s phenomenologically extremely similar. So we were among the early people to say that this was really a viable, and presumably the best in terms of just best fit, model.
And so what was the reaction to this before the entire supernova 1a results started coming in?
Negative on the part of both the astronomers and the physicists. The physicists said, “This is an impossible value for lambda.” I remember Ed Witten objected to it. He just said, “How would you possibly get it?” I said, “I don’t know. That’s your problem, not my problem.” [Laughs] But I felt that, since Paul was an extremely competent physicist, that if it was really nonsense he wouldn’t be writing the paper with me. I was just doing the fitting to the models.
Well, quintessence is essentially a scalar field that’s displaced from its true minima.
Right. And then the astronomers were, on the other side, skeptical. The physicists tried to defend omega equals one matter. And you know, especially people like Carlos Frenk and Joel Primack and others. It was the Bible. And the particle physicists wouldn’t contemplate lambda at a lower omega. And the astronomers didn’t like it just because they said, “Well, look. If omega is low, then you just have an open universe, that’s that.” And they didn’t seem to be troubled by all of the arguments which indicated that the cosmological constant fits better. For example, then you don’t have a problem with the stars [being] older than the universe. All the other arguments that we put in that paper, they just thought, “Well, the uncertainties are big enough that it’s tolerable.” And we really felt that that was the best fit, and that there were some attractions to a flat universe which could be achieved with this.
So do you, at the moment, have a strong preference for particular physical source for the omega-lambda component, or is that still something —?
No. It’s a black box. I certainly like quintessence more than a cosmological constant because a cosmological constant has always felt to me like action at a distance. Quintessence is a real force field which is compressible and acts in a proper way. Because a cosmological constant is a force pushing two particles apart, dependent only on their distance in a way which doesn’t depend on anything. If one moves, the other one seems to know about it right away. It seems to me acausal. It just doesn’t seem like a very natural kind of physics. So I like quintessence. And then the question is, is cold dark matter right. And the history of this is that you always perceive by successful approximations. And my guess is that we’ll find that the present cold dark matter picture is not quite right. I have explored two variants of it, one of them being collisional dark matter, the other being warm dark matter. I am not sure the cold dark matter really needs that revision. It looks to me as if it does. And of those two the warmed up matter one seems the more attractive — the most attractive of the variants that reduces back to the other one for most cases. I would not say it’s compelling, but I’m open-minded to considering minor variations because of the way science in this field has always gone. It’s always gone by successful approximations. You keep most of what you had, but you realize there is some new element that you need. And my guess is that’s the case now as well.
What’s your take on biasing in cold dark matter models? Do you think there’s a good physical grounding for this?
Well no, I think it’s nonsense. In other words — let me back up. I was actually in the room when the word was invented. It wasn’t my invention. Mark Davis, who had just done this CFA survey of galaxies, and was among the very earliest ones to make computer models for dark matter, noticed that the dark matter — the pictures on the screen; I was at Berkeley at the time — didn’t look like the pictures that he had gotten from the surveys, because the voids are much emptier. It really strikes at empty patches in the real universe, and then gray gradations. He said, “Well, we just have to turn up the bias.” You know, a vacuum tube to get higher contrast, “and it will work.” And that is delta n over n to some power. I mean, if you take the density of dark matter to some power is the other, it turns out its multiplicative if you do the linear domain. And so he said, “Why don’t we try this?” etcetera. And that was the origin of the term, and became widely adapted. At that level I thought it was a good idea. But then it’s become a Bible. So they had nonlinear bias, stochastic bias, all kinds of other things. It was fundamentally flawed. You would never say that stars are only a function of density. [Star formation] is a function of density and temperature and magnetic fields and various other things. There’s a lot that goes into the physics. So why would galaxy formation be only a function of the density of the dark matter?
It doesn’t make any sense. In our simulations, as good or as bad as they are, it’s a function of the density and the temperature, because the Coma Cluster has incredibly high density of gas in it. There is no galaxy formation in it because the gases tend to be [unintelligible word] degrees. All the bias schemes have just said, “What function of the density of dark matter is the galaxy?” And then, when it doesn’t work they say it’s stochastic bias, i.e., there is a random component. Well, if you have one of the right variables but not the other ones, there is always going to be a random component, because it’s not as if it’s quantum mechanical. It’s absolutely deterministic and what appears to be stochastic is just because you don’t have the right variables. So I have not been enthusiastic about this approach. I just think it’s wrong-headed from the get go. If you do it correctly — and I’m not saying we’re there — it’s going to be a function, certainly, of both density and temperature.
And maybe a few other —
Maybe other things which we don’t yet know. But what happens is, people get a hold of things and initially it’s just a kludge. You know, you count the number of legs; if it’s four it’s a mammal and if it’s two, it’s a bird. And that’s a pretty good way to start but you’re going to make some mistakes. But then, people get very, very serious about it and do high falutin’ mathematics on it as if it was given on Mount Sinai.
And it doesn’t make any sense. And that’s what’s happened in many fields in astronomy. And it happens more when there are more people in it because you have enough to make a bandwagon which all charges around. And then other people make refinements on the first people’s calculations, people make refinements on their calculations and it becomes like medieval theology.
Could I ask you what you think — and I this is I think still very speculative as far as I know — but what you think the most likely population 3 objects were?
Well, I think everyone agrees, even if it’s speculative, even though they could all be wrong. And the first ideas on it were Martin Rees’ and others’, the first calculations were mine with Nick Gnedin’s. And that is, when you have molecular cooling like you had your cooling, you can get an early population of stars which can cool just because of this, where the gas can cool and make stars. Then, they turn on and the soft UV of hard optical from them kills the molecules, destroys the molecules. And then you stop until you can go later in more massive objects, which can then form by cooling on [unintelligible phrase] ordinary processes. So that’s the idea. I think everybody who has worked on it has found the same things. Could be wrong. Now, it depends on the model, whether or not that happens. Just to take an example. If you were to start with PIB, which was Peebles’ idea for a primeval isocurvature, then things start very early and you could have a very substantial population 3. If on the other hand, warm dark matter or some variable of that is right, then you start late enough so that you pass this phase altogether.
Then you won’t have any stars until you’ve already got galaxies.
Right. More massive things than — but in the standard hierarchical models, there would be a phase which you’d go through which would be this.
And how massive would these be?
It’s about 10-4 of the total mass forming at the small clusters which fragmented the stars. You know, hundreds of thousands of stars. It could very well have happened. I think the evidence is inconclusive now, about whether or not. It’s interesting because it tells you something about cosmology. If you change the force law between the dark matter particles, you can make this phenomenon occur or not occur.
You’ve said that a lot of things in your early hydrodynamical curves, a lot of it was just limits on computing power.
How do you see things progressing in terms, first of all, of more powerful computing? But do you think there’s any chance that the physics will change, or do you think it’s partly that you had to make assumptions in order to do the computations that you won’t have to do, that you won’t have to make in the future?
It will certainly get better, but as Peebles pointed out it reduces to the previously unsolved problems of star formation. It’s one that my daughter works on. Now, if we can get heuristic recipes, either from really good simulations of star formations or from understanding observations — so we could say that under these conditions of temperature, density and magnetic field, you’ll form this and this population of stars; and I guess turbulent velocity would be another ingredient — then you get that in this coming out, either a very careful understanding of star formation or just understanding the observations of getting a cookbook. Then I think we’ll be able to put that into the cosmological simulations. But as of now, you’re kind of stuck. I think we’re not doing badly because basically, what we say is at some point the gas collapses. That we can follow. And since there is no evidence that things collapse and just sit there as molecular cloud, you don’t see that in nature. The assumption that it forms other stars, with approximately the distribution of stars that the Orion or other things have, at least is based on matching to the observations on this, but it is unsatisfying at some level that you don’t have a better theory for it. Now the other changes that may come in, though, is — I think it’s entirely likely that cosmic rays, magnetic fields are going to turn out to be important, and that the specific particle physics of the dark matter is going to turn out to be important. None of those are in there right now in any of the simulations.
Another question about dark matter. Basically, one of the frustrating things about it is that we have only detected gravitational effects of this. Do you see any prospects for seeing it in other ways?
I am not enough of an experimental physicist to know if there is a really good chance of that. There are a lot of efforts. I wish them well. But right now you could say — you know, and I said at coffee — maybe it’s all massive black holes. Nobody could give you an argument against that right now. So it’s that far from being understood. If I said I had a dream and the dream told me that it’s all 105 solar mass black holes, primordial black holes, there is no one who could give me an argument pro or con at the present time. That’s how badly we understand it. It’s not even elementary particles at all.
Right. So is there anything that we’ve missed?
I think we got up to the present. Oh, there is one project I missed when I was talking about things and it’s a big one; and that’s the Sloan Digital Sky Survey.
Oh right, I was going to ask you about that.
Yeah. I have to go, but we have a couple minutes more.
It really all started — I was just thinking of unappreciated scientists I know. Jim Gunn is certainly one of them. He’s an extraordinarily intelligent person. He could have worked in mathematics, in physics or astronomy. But where he’s even better than other people was as a technologist, as an engineer. So he was very instrumental in building the first cameras for the space telescope in his part of the team. And when he came into my office one day and tried to describe the prospects for building a digital Schmidt sky survey, I thought this was fabulous.
So did he have the idea of the huge CCD area then?
Right. He said it could be done from his work on that, and that you could do much, much better than the Schmidt Sky Survey. You could do multiple colors; you could do linear detectors, etcetera, etcetera, better time resolution. It took minutes to convince me [because], one, this would be terrific science and, two, it could be built. I just believed him. His record was so strong. If he said it could be built, it could be built. So I said to him, “Look, Jim, if you can build it, I’ll try to find the financing and the organization for it.” And we worked together on that.
So when was this that you originally proposed that?
You should really ask him. He actually wrote it up recently, and he in fact dated the time he came into my office. But I can’t remember it now. It was certainly a decade ago, but it’s maybe a bit more. Maybe ‘89? I’m not sure. Then I got some seed money from Dellers [spelling?] and some money from Princeton University. Harold [Furth], I wasn’t provost then, but he was very cooperative. Because I realized it was a big project, I had to find partners, both with the technical skills and then the scientific skills to analyze it and for the funding for it. I think it was frustrating for the scientists involved because there was always element of perils of Pauline — would we have enough money, could we get through the next quarter? I was always confident. I would say, “Look, you just worry about building the thing.”
Were there any particularly surprising parts of the project that ended up being a lot more difficult than you expected?
The software. That was a genuine error because it wasn’t Jim’s specialty, and I just don’t think we realized. And we underestimated the cost probably by a factor of ten. A difficulty.
What was your data transfer rate from the telescope once it was — in order of mag — Was that part of the problem?
Well it generated terabytes, but just designing all the software — the pipeline that takes the data, the analysis software, all of it was just a huge job. I had no input to it at all. There were a lot of very good people, including Robert Lupton in Princeton. But as our awareness of the needs of this developed, I had to scramble to get more partners and more funding for it. That was probably the biggest weakness. It made me very skeptical of many national projects. I remember I once wrote an op-ed piece in the New York Times about Star Wars, anti-Star Wars. I don’t know if that got on my CV. But when I read over an earlier generation of it — that the complexity, the number of lines of code was more than the airline reservation system and fewer than the US telephone system, it was intermediate — I thought, “And it has 15 minutes, and it has to work the first time? What’re the odds of that happening?”
[laughs] That’s just —
This was from National Academy analysis, and everyone knew that it was just preposterous.
A few thousand warheads, and you’ve got to sort them all out.
And you’ve never had a chance to test the system properly under the circumstances. I mean, just imagine some group of government workers who built the airline reservation system, and then in 15 minutes — If a reasonable fraction of the United States population wants to get an airline ticket, what’re the odds that it’s going to work? And it has to work at — since if you stop 10 percent or 90 percent of incoming missiles it’s a failure — it has to work at .999? You just ask yourself, “What’s the chance of that happening?” So I’ve rarely become overtly political, but that was an op-ed piece on that. And so I’ve been on the fringes of some of these debates. That was in the New York Times early on.
Have you continued to be involved in the Star Wars debates or was that —?
A little bit on the edges. I don’t think I’ve been as prominent as that, and that was a decade ago. It was an earlier generation.
Because I don’t know if it’s a union of concerned scientists or groups of that sort that have continued to criticize it.
You know, I’ve talked to Garwin about it and he’s very acerbic and funny on it. But it’s always been a completely crazy idea. And it’s just done, as far as I can tell to feed the aerospace industry, and for political purposes, to get people elected.
Yeah, that’s been my impression as well. Yeah, it’s an easy sell as well. It’s kind of hard to —
Right. Would like to correct everybody?
But, that’s been my only venture into real politics. [Laughs]
So, could I ask you a few more general questions?
One is just about the relationship between physics and astronomy. And historically, these have evolved as very distinctive fields with distinctive cultures. But obviously, as we approach the millennium they have interacted a lot more. And how do you see the changes from your own career and maybe —?
As I say, there have always been people at the interface and I don’t see it as a qualitative change. That is, that you know if I look back at my predecessors — Chandrasekhar and Spitzer — or at their predecessors — Eddington, etcetera — they have always been people who, on the one hand, paid close attention to the astronomical observations; and on the other hand, thought in terms of the highest-level physical theories and trying to find the confluence between these two. It’s not been something which has interested most physicists and it’s not been something that has interested most astronomers. But I don’t think there’s more of a coming together now than there ever was. It’s always a small fraction. And that I think there’s always a tremendous opportunity for interaction.
Another question about the development in the field. Do you think people will shift to doing multi-wavelength studies with particular problems, rather than developing an expertise in a single wavelength and pursuing that?
Some people already do: the observers. The theoreticians automatically do it. There are some observers who will simultaneously study radio and X-rays of some cluster of galaxies, etcetera. And so I think, if observing was hands-on and you had to be there at the telescope, you learned one set of skills or another. But insofar as you are sitting at a computer console thousands of miles away and the software is becoming equivalent, I think you have more and more people who will be doing virtual observing at many wavelengths.
Finally, I just wanted to ask you about what you think the most dramatic changes have been during your own career and our overall view of cosmology and other parts of astrophysics that you’ve been involved with?
I would say it’s a maturing of the field, that at any given time you look across fields and it’s hard, intellectually, to see that some of them are much more mature than others. Medicine is so much less scientifically done than economics, which is so much less scientifically done than chemistry, etcetera, etcetera, until you get to mathematics — in terms of standards of rigor. Well, if you take within a discipline, cosmology, when I started it, was really like theology and it was unappealing to many people. Many people of the older generation don’t realize how much it has changed. Stellar atmospheres was one of the first things to become rigorous, with good physics and good observations combined to try to understand things. Then stellar interiors, then interstellar medium, stellar dynamics and cosmology only in the last thirty or forty years, where there are well-defined theories which can be right or wrong, where there are good observations, good mathematical techniques where you try to put them together and see what’s correct and what isn’t correct, and try to approach it in a relatively unbiased way. And there are a lot of fields which are not yet there. Star formation is one. I think AGNs are in the same position. I think there’s a lot of dogma there and not very good theories. So those are two very important fields which are not mature. But I think cosmology is approaching the state of a mature science, where there are clearly formulated theories and then there are really good observations; and then you can make the theoretical comparison between the two. It doesn’t mean that we have the answers yet, but at least it seems more sensible than it was. There are always dogma and bandwagon effects in all fields. But as they shift on the spectrum, it becomes less Don Juan. So I think that’s one that’s in a, getting to be in a really, really good state. And of course, the most interesting ones are always the ones that are not quite in that state. When cosmology gets to that state, then maybe it will be more fun to work in one of the others. I think AGNs is an extremely interesting field now, just because — I have often thought if I hadn’t been an astronomer I’d be in economics because economics is in a really sorry state through my observations; where the predictions always come out wrong and they can’t tell the difference between wanting something to happen and thinking that it will happen or having evidence that it did happen.
Right. Or another way of putting it is some predictions always come out right, retrospectively.
[Laughs] Yes. And so, those fields seem like fun, in fact, for applying the scientific method. But I’ll probably continue with cosmology because I only have so much of a career left, and I’ve developed all these tools now, working with people on the — And I hope to apply them. And right now, most of the work is to try to refine them to the point where you can really say which cosmological models are right and which ones aren’t right. So I think it’s a good time to stop.