History Home | Book Catalog | International Catalog of Sources | Visual Archives | Contact Us

Oral History Transcript — Dr. Herman Goldstine

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.

Access form   |   Project support   |   How to cite   |   Print this page


See the catalog record for this interview and search for other interviews in our collection



Interview with Dr. Herman Goldstine
By Nancy Stern
At Institute for Advanced Study, Princeton
March 14, 1977

 
open tab View abstract

Herman Goldstine; March 14, 1977

ABSTRACT: Family background; graduate study in mathematics at University of Chicago in the 1930s under Gilbert A. Bliss, Marsten Morse and Eliakim Hastings Moore; faculty position at the University of Michigan, 1939-1948; war work at the Ballistics Research Laboratory and at the Moore School of Electrical Engineering in Philadelphia; ENIAC and counter technology; John Von Neumannís involvement with computers at Princetonís Institute of Advanced Study; mathematics after World War II. Also prominently mentioned are: Paul N. Gillon, T. H. Hildebrandt, Martin Schwarzschild; Aberdeen Proving Ground, Ballistics Research Laboratory, and University of Pennsylvania.

Transcript

Stern:

Letís begin with some questions about your youth. Then weíll work our way up.

Goldstine:

All right.

Stern:

I know you were born in 1913.

Goldstine:

Right.

Stern:

But I donít know anything about your family.

Goldstine:

I see. You want to know that sort of thing?

Stern:

Yes.

Goldstine:

Yes. My father was a lawyer who practiced in Chicago. My mother was a housewife. And I was born and raised on the North Side of the city of Chicago. I went to a public school called Stephen K. Hayt School, and then I went to the Nicholas Senn High School in Chicago; in my third year to the Los Angeles High School, then back to Senn and finally to the University of Chicago. I started at the University of Chicago in 1930, and I got my Bachelor of Science degree in 1933. Next I got a degree in mathematical astronomy, in fact a Masterís degree in mathematical astronomy, under Walter Bartky. Then in Ď36, I got my Ph.D. in mathematics, under Lawrence M. Graves at the University of Chicago. I next spent a year as research assistant to Raymond Walter Bernard. I helped him to bring out the memoir of a man named Eliakim Hastings Moore, who was the first chairman of the mathematics department at Chicago. I then became a research associate -- I guess that was the title -- and I was the assistant to a man named Gilbert Ames Bliss who was the chairman of the mathematics department. He and I brought out his magnum opus, which was a book on the calculus of variations. That took me to about 1939, but then I kept returning. He brought me back to Chicago each summer for several years. In that period, while I was at Chicago, I was also an instructor at the University College of the University of Chicago. In Ď39, I went from Chicago to the University of Michigan, and taught mathematics there as an Instructor.

Stern:

Just a second, I want to go back over some of the things you said. For example, when did you decide to go into mathematics?

Goldstine:

Right away. I think that was my plan when I started at the university.

Stern:

So while you were in high school or even earlier you made that decision?

Goldstine:

Certainly while I was in high school, I certainly had it in mind that I would study mathematics. Yes.

Stern:

There was never interdisciplinary kind of orientation to your early schooling? Because Iíve noticed that about computer people. They often seem to have had an interest in philosophy or engineering or something.

Goldstine:

No, I think I knew I was going to be a mathematician. So I donít think that was the problem. I took a number of courses outside of mathematics, but that was just for fun, not --

Stern:

-- like what?

Goldstine:

I took art history, and I took a course on Gilbert and Sullivan -- I mean, all kinds of things of this sort. Anthropology, economics. I donít remember what all, but you know, they were the sorts of things one took to fill up oneís elective requirements. Also I audited a number of courses. For example, in the time when I was at Chicago, Thornton Wilder was on the faculty. I audited his courses on the DIVINE COMEDY, and I guess on the ODESSEY too. I audited a course that Morris Raphael Cohen and Bertrand Russell gave in philosophy just because they were famous people and it was fun to see them in the flesh and blood, and hear how they sounded. That was the sort of thing I did with my leisure time.

Stern:

I know in terms of the story about ballistics and the areas of applied mathematics, that Gilbert Bliss was a very significant figure. What was he like to work with?

Goldstine:

Well, he was a very nice person. On the political side, he was a Republican conservative gentleman who lived in the suburbs of Chicago. He was formal but he was very kind to me. I suppose I viewed him like an uncle. He had a suite of three offices. The outer office was the secretarys, the middle office was a formal one in which he used to interview people, or to talk with people on business, and then there was an inner sanctum, the third office, which was the room where we actually worked. We used to start about 9 or 9:30 in the morning, and work essentially till about 5 each evening. Our mode of work largely was of this sort: I would write during the evening -- after work -- drafting the next section of his book, or books, because I helped him also write a book on exterior ballistics. Then Iíd bring my draft of what the next section might look like, and he would then either disregard it in its entirety, or use it as a kind of skeleton, to write what he wanted. We would then argue and debate the resulting revised draft, and write the material again. The final version was the result of this sort of interchange. So that was the kind of collaboration it was.

Stern:

This was in the calculus of variations, as well as the exterior ballistics, youíre talking about?

Goldstine:

Yes, thatís right. The most important thing to him was the book on calculus of variations since it was a book which marked the culmination of his career. The book was written and finished just about at the time he retired. So, it really represented his last great contribution. He produced the book on exterior ballistics because the war was coming along. He had made a great contribution at Aberdeen during the First World War, and after the First World War he had written some papers which were very significant from the point of view of ballistics, and I think he felt that by writing this book, he would be doing a service to the country. So he was anxious to do it.

Stern:

Your dissertation was in the calculus of variations as well?

Goldstine:

It was -- yes. I tried to study extremal properties of functions, real-valued functions, defined on abstract spaces. I tried to generalize the calculus of variations, to such spaces -- Banach spaces -- and ones of this sort.

Stern:

You only worked with him after you finished your dissertation?

Goldstine:

Thatís correct, right.

Stern:

The kind of focus at Chicago at this time, would say it represented the new, with respect to this, or more the old?

Goldstine:

No. Thatís a good point. There were two schools of thought on the calculus of variations. One of them was a sort of conservative classical school, of which Bliss I think really represented the last important member. And then there was a modern school here at the Institute for Advanced Study, which was headed by Marston Morse. In fact, at one time during that period, and I donít remember the dates but it was in the thirties, Morse visited Bliss, and while he was there he invited me to come here to the Institute for Advanced Study to be his assistant. I debated this attractive offer with myself, and then decided I would stay on and help Bliss finish the book. But itís quite true, Blissís work was classical. It was concerned with the calculus of variations, with properties of extremal curves in the small. Bliss did not see, I think, or appreciate the significance of Morseís ideas, which were connected with properties in the large. Bliss really finished up the classical subject, pretty much. Well, he wound it up at least until modern times, when nowadays I understand things like control theory use the classical calculus of variations, making it once again an important subject. But thatís a good point, that you raised.

Stern:

Were there other mathematicians working in this at the time?

Goldstine:

Yes, there were. You mean in the calculus of variations?

Stern:

Yes, at Chicago.

Goldstine:

At Chicago, yes. In fact, Chicago was an unbalanced school at that time. The mathematics department was unbalanced, because there were too many people in the calculus of variations, and too few people in some other important branches of mathematics. There were two subjects in which Chicago in those days was impressive. One was the calculus of variations, and the other was number theory, under Leonard Eugene Dickson. These subjects were taught almost to the exclusion of other things which might better have been taught.

Stern:

Like what?

Goldstine:

Topology or Morse theory. There was no topology taught at Chicago in those days, even though it could be seen to be a coming, vital subject to mathematics and probably to physics too. I think Chicago was at the end of an era at that time. The department originally had been created by E.H. Moore, whom I mentioned before. It was in its time the most influential department in the United States. Mooreís thesis students included George David Birkhoff, who became the great figure at Harvard; Oswald Veblen, who was a great figure at Princeton, and then later here at the Institute. It included T.H. Hildebrandt, who was the chairman of the department at University of Michigan. Chittenden, who was the chairman at, I think it was Iowa or Iowa State, Iím not sure now. It included R.L. Moore, who was the great figure in mathematics at the University of Texas. And it included many others. The influence of the University of Chicago at that time was probably predominant in the United States. E.H. Moore was still alive when I went to the university in the thirties, but was in poor health. He was a very secretive man; secretive in the sense of not being willing to publish his researches. He was always perfecting his things, and he never quite reached the point where he was willing for them to appear.

He worked on abstract space theory, and could have made a very great contribution to the world if he had published regularly. But instead, he was always throwing away his previous versions, because he saw more elegant ways to do things. And very largely, although unbeknownst to him, he was really in competition with von Neumann, who came a generation later; but he was in competition with him because both of them were trying to discuss operator theory in Hilbert space. His technique for handling Hilbert space was very awkward, compared to the way von Neumann eventually formulated the theory. The result was that when Moore began to study operators, the whole apparatus was so incredibly complicated and constrictive that he never could see how to proceed to unbounded operators, with the result that when von Neumannís theory of Hilbert space came along, it just swept E.H. Moore away -- in fact it swept everybody else away, too. But it meant that the work at Chicago, at least insofar as it related to Hilbert space theory, got washed out when von Neumannís work came along. There were many people at Chicago who had worked in calculus of variations. Bliss represented, as I said, the culmination of all this work.

His book represented his own research plus that of his predecessors, and that ended the field. Dickson had been a great figure in algebra, up until around 1930, at which point he got tired of the subject, and told A. Adrian Albert, who was then a young assistant professor in the mathematics department that he was tired of algebra. He had just finished a book called ALGEBRAS AND THEIR ARITHMETICS, which was kind of classic. At least the German edition was in the subject at the time. He turned the field over to Albert saying he was tired of it, and he was going to work on number theory -- on the Waring conjecture. This is a problem, proposed by an English physician and mathematician back in the 1700ís, which asserts that every number is the sum of four squares, nine cubes, etc. By 1936 -- just about 1936 -- I believe, Dickson retired. At the last spring meeting of the American Mathematical Society, which in those days was always held at the University of Chicago, Dickson had listed as a paper that he was going to present, something entitled, ďAnother Result of Waringís Conjecture,Ē or Waringís Problem, however he phrased it. When he got up to speak, he said, ďMr. Chairman, Iíd like to amend the title of my paper to read, ĎThe Complete Solution to the Waring ProblemĒ -- which was very nice, and everybody cheered and applauded, and Dickson then showed how he had finally finished this problem. But at any rate, the point of all this excursus is that by about 1936, the University of Chicagoís mathematics department had nearly reached the end of the road. Dickson was the first Ph.D. the University of Chicago ever produced at all, and therefore the first in mathematics. Bliss was a man who had got his Ph.D. early on at Chicago too.

So, by Ď36 or thereabouts, the mathematics department at Chicago had gradually gone downhill, to the point where it was just about at its nadir. In fact von Neumann once told me that he felt that the life of any department at the university was one generation, and that one had to do something very important, very profound, in order to give it life for another generation. In fact Chicagoís Math Department continued to sink somewhat after Ď36 and gradually, when Bartky became dean, Hutchins rejuvenated that department, bringing in Marshall Stone, Saunders MacLaine and a number of other people, and it became the great department that it is today.

Stern:

You were a student by this period, 1930-36. Were there any graduates that really went out to become illustrious mathematicians?

Goldstine:

Thatís what trying to remember. Of course, there were very --

Stern:

-- besides yourself, of course.

Goldstine:

Iím not trying to duck out of that one. Perhaps the two very best were Adrian Albert and E.J. McShane. Let me try to explain to you Blissís point of view on Ph.D. students. Bliss held the view that the United States was going to need a large number of people who could teach undergraduate mathematics as distinguished from doing research in the field. He knew that there was no degree available other than the Ph.D. So he set as a conscious policy that he was going to produce teachers of mathematics. Accordingly he gave many thesis topics to undergraduate teachers who worked as his students and who came back for summer quarters. These were people who were teachers at schools such as the city colleges of New York. Bliss saw to it that they got Doctorís degrees, even though their main interest was in teaching. He spent, and I helped him, considerable time caring for and nurturing these people to get them through, so that when the time came, there would be an abundance of well-trained teachers. I think Bliss saw, and that was much to his credit, that the population growth had dropped off substantially; but that as the Depression receded, there would be an increase in college attendance. As it turned out, there was a tremendous influx after the war, as we all know.

He couldnít have foreseen that but he could, apparently, foresee the long-term need for teachers. At any rate, that was the situation there, I think the most outstanding people who got Ph.D.ís in mathematics while I was around there was A.A. Albert and E.J. McShane. McShane is now a retired professor at the University of Virginia. Heís written extensively on the calculus of variations and on integration theory. The other was Adrian Albert, who became eventually a Distinguished Service professor of mathematics at Chicago and then the Dean of the Physical Sciences Division. In later years he was a trustee of the Institute for Advanced Study. He also played a big role in the IDA, the Institute for Defense Analysis. He was a very important figure in mathematics and in public policy, particularly as far as mathematics is concerned. Those are the ones that come to my mind at this point.

Stern:

One more question about Chicago. You mentioned that Bliss was able to foresee the great need for mathematics that would come in the next generation or so.

Goldstine:

Yes.

Stern:

On what did he base that, on the war needs, or on some aspect of mathematics?

Goldstine:

No, he couldnít have known about the war needs because this was in the thirties, the mid-thirties, when he was talking about this. I think he sensed the trend for more and more people to insist on getting college educations.

Stern:

Just on college in general.

Goldstine:

Yes, college in general; I think so. Now, what he based this on, I donít know. The only thing I could conjecture is that he saw that, with increased prosperity in the United States during his lifetime, the whole thrust was for more education for people. I donít remember, when he was a young man, what the obligatory age for attendance at school was, but Iím sure that even in my lifetime, it has moved forward. I think that in those days the junior college was just coming into being. Hence I believe he could have sensed that this was going to be an increasingly important thing.

Stern:

I see. After Chicago you went on to the University of Michigan?

Goldstine:

Yes, right.

Stern:

Can you tell me a little about your decision to do that, as opposed to going to the Institute?

Goldstine:

Well, actually I had the option of going either to Michigan, and I think the salary was $1800, or to another school, which I didnít think was nearly as good a school of mathematics as Michigan, at $2500; and I decided that I would take the $1800 offer, rather than the $2500 one, which I did, and I never regretted that, either. In those days teaching was a very complicated business. I taught 16 hours a week, which was a horrendous load, and it always consisted of an 8 oíclock, 9 oíclock, 10 oíclock and 11 oíclock class.

Stern:

Wow!

Goldstine:

Michigan is a kind of curious enclave on Eastern time, sticking into Midwestern time zone, so when you got up in the morning to get to your 8 oíclock class, it was absolutely like the middle of the night. It was very hard, in that way. I enjoyed teaching, and I taught there until the war came along. When I had been a student, the University of Chicago had field artillery ROTC, and I took that, I guess for two reasons. One is, I didnít like gym. The other is, the ROTC in Chicago in those days consisted of horseback riding, which I did like; we rode and we played polo and we did things of that sort. There was another plus to the ROTC, in those days at least. I donít know what things are like now; but they had theoretical courses, for which you got full university credit. They were, you know, absolutely cinch courses, so they helped. I got through Chicago in three years, and one thing that helped me get through in that time was these courses that didnít take any work. So thatís how I was in the ROTC, and when I got out, I got a commission as a second lieutenant in the Reserves. When the war came along, I first got deferred for a while because I was teaching. Then, I guess it was July of Ď42, I got taken into the Army, and for some reason they didnít want people in field artillery at that moment, but they wanted me in the Air Force. I guess it was called the Air Corps at that point. And I was sent to a base in Stockton, California. Simultaneously, Bliss had got in touch with Veblen about me. Veblen at that time was the chief scientist of the Ballistic Research Laboratory at Aberdeen Proving Ground. He had been --

Stern:

-- thereís a lot I want to ask you about that, but I wanted to ask some more questions about Michigan. What was the situation in terms of getting jobs as a mathematician in a university?

Goldstine:

Oh, it was very poor. This was the Depression. There were very few jobs. Thatís why there were few people who had Ph.D.ís.

Stern:

How did you happen to get this job?

Goldstine:

Well, Bliss got it for me. He was very kind to me, and in effect, by having me as his assistant for those years that he did, and for summer jobs, he managed to keep me going financially. He was a good friend of Hildebrandt, the chairman at Michigan. They had gone to Chicago together, and Hildebrandt worked in the same sorts of things that I did. In fact, Hildebrandt and I collaborated on one or two papers. So we knew one another, and I guess Hildebrandt liked what I worked in since he was always nice to me. He got me a couple of raises while I was there and had my teaching load dropped from 16 hours to 15 hours. The job market was just about zero at that period. There were many people whom I knew, who did get degrees, and then just couldnít get jobs.

Stern:

Mathematicians, you mean.

Goldstine:

Yes. Physicists, mathematicians, you name it. The job market was very poor, much poorer than anything today. People now think itís bad but it was incredibly worse then, because there was absolutely no government support. That really made a difference. The whole economy was terrible. When a person couldnít get a job at that point -- I mean an academic job -- there was no industrial or governmental work. A job, in those days, meant maybe running an elevator or driving a taxicab, or working in a butcher shop or some such thing. It was different from today. To some extent, today, it means people canít get a job in the university they would like, such as Harvard, Yale, Princeton, Chicago or what not. In those days, it meant you couldnít get a job period. It meant you even get a job in a little girlsí school in Keokuk, Iowa, either. There just werenít jobs. In those days, maybe around one or two jobs would open up each year. Probably in the United States there were only a very modest number -- half a dozen as a guess. So it was a very, very thin market. I guess that answers your question.

Stern:

I suppose many had to go to high schools to teach at that time.

Goldstine:

I donít think it was very easy to get high school jobs either. I think it was very much as it is today. I think there had been a baby boom after the First World War, just as there was after the Second World War, and lots of people had been hired, in the public and private schools, to cope with that baby boom. That wave then moved through with the velocity that these things do. The Depression saw a drop in children. Maybe there was an improvement in birth control in that period, I donít know. At any rate, there werenít jobs -- many jobs -- in schools. I donít think it was easy to get a job teaching in a public school. For one thing, even in those days, having a Ph.D. didnít qualify one to teach in a public school. You had to have an education or teaching degree of some sort, and I donít know what that entailed. I think very few people, having sweated through a Ph.D., were then willing to go back to college and study to get a teacherís certificate. I think it was just a very bad period.

Stern:

Did people tend to view this situation as more of a temporary thing? Today people tend to have the idea, this is the way itís going to be from now on.

Goldstine:

Well, thatís the great thing, of course, that Franklin Roosevelt did for the United States. Up to Ď32 when he came in, there was no particular optimism about anything. In many ways, the great thing that Roosevelt did was to bring in a feeling of optimism. You know, there was his theme song ďHappy Days are Here Again.Ē Although that was all a part of the hoopla, it actually did mirror what people were feeling. I think there was a lot more optimism. And then, of course, from our point of view fortunately, the Allies began to buy armaments in the United States, which began to stimulate the business economy of the United States and jobs began to be a little more plentiful. But there were few people, there were very few working for the Ph.D. in mathematics. When I was teaching at Michigan, I can remember, Charles Rickert whoís now a professor at Yale; Jimmy (James) Savage, who was also a professor at Yale. And there was a Chinese named Fan, I think. I donít know what ever happened to him. I think those were the only three students in that period that I remember at Michigan. So there werenít many Ph.D.ís either.

Stern:

These were students of yours?

Goldstine:

No, I was on their thesis committees. But maybe the whole department at Michigan turned out about that many. Departments were incredibly smaller then than they are today. While Michigan had a big department in those days it probably had a faculty of 20, 30, 40 persons as compared to, at a guess, probably 100 today. The main function of those of us in the department at Michigan then was to teach engineering students. Mathematics was compulsory for engineering students, and that was how the mathematics department earned its bread and butter.

Stern:

Just basic courses were taught?

Goldstine:

Just basic courses. Another thing Hildebrandt did for me was to get me out of teaching those cursed undergraduate courses, and got me a graduate course to teach.

Stern:

In what?

Goldstine:

I donít remember now. Thatís the trouble. Itís too long ago. It wasnít anything very exciting, but it was better than dull courses. Because if you taught two courses in analytical geometry and then two courses in calculus in a row, the excitement was pretty low, you know. But it was the way one lived. That was typical, because both in mathematics and in physics, for that matter, this was the period in which there was no government money. There was virtually no private money. So research was what you did on your own. You were expected to do it, if you were going to get promoted or even kept on. You jolly well had to do research. But there was no money. I mean, there was no money for students. There was no money for secretaries, for trips, for anything. So everything you did was on your own, and I know in my case at least, the summer was the time that I really used to do whatever I was going to do. Something about that came into my mind, but I canít think what it is. Anyway, if it comes back Iíll tell you about it. But at any rate, this was just at the beginning of the change in the governmentís attitude towards research, in terms of massive government funding of projects and individuals and all the rest of it. The one or two things that did happen, happened under the auspices of agencies such as the American Mathematical Society. Actually the first time I met von Neumann was at such an affair. The Society organized a symposium on Modern Theories of Integration, and it was sponsored by the University of Michigan. It was a weekís symposium, to which all the leading figures were invited, and the Society asked me to be the rapporteur. I was therefore expected to write a one or two page account of each of the talks -- each talk being a few hourís lecture. One talk that I can remember with particular clarity was by Norbert Wiener. I remember that because it was the only talk which was so bad that neither I nor Hildebrandt nor anybody else who was in the audience could make anything out of it; it was just an absolute disaster for me to have to try to write a page or so account of this talk.

Stern:

What was it on?

Goldstine:

I donít even remember. Well, it was supposed to be something about integration theory, but it was such a rambling thing, it was so obvious that Weiner had done no work in putting it together. I donít think it was just my fault. Of course, thereís always that possibility. I would have felt that way, perhaps, if it hadnít been that a half dozen other people were in the same boat as I was. They just didnít know either what he was talking about. And this was a subject I knew a lot about, because Iíd written a number of papers on integration theory. But the other person I can remember very clearly was von Neumann, who gave a superb, beautiful lecture, in the way he invariably did. It also was hard to write up because, like all von Neumannís lectures, one always fell into the trap of sitting there just listening to this beautiful presentation, which seemed so tremendously lucid and clear and easy that one didnít bother to take notes. It was only after one got home and tried to recreate the material that one realized that he had missed the essence. So that was a lot of work, too, but it was an absolutely gorgeous lecture. I met von Neumann then. He actually didnít meet me. It was one of these things where he shook hands with me, but Iím sure he had no recollection of it at all.

Stern:

When was it?

Goldstine:

Well, it was some time before Ď42, thatís all I know. Ď40, Ď41. I donít know. I would have to look in THE BULLETIN of the Math Society to see when that really was. But that was an interesting encounter for me, because it really made it very clear to me what a superb mathematician von Neumann was. Everything connected with that talk of his was just an order of magnitude better than others which never were however very good. And at any rate, thatís the beginning of that. Letís see, what else? I started to tell you about Veblen. During the First World War, the ballistics work in the United States government had been divided up between Washington, D.C. and what in those days, at the very beginning, was the Sandy Hook Proving Ground, which was located at the end of Sandy Hook Harbor. Presumably that was one of the harbor defenses of the United States in an earlier era. Now, the Washington office of the Ordnance Department was the dominant office. That was the office of whatís called the Chief of Ordnance, the boss of the Ordnance Department. In the First World War, the man who was, so to speak, the chief scientist for the Ordnance Department was an astronomer named Forest Ray Moulton, who had been professor of astronomy at the University of Chicago. He had a number of brothers, as a matter of fact, one of whom was professor mathematics at Northwestern University, and a brother, Harold, who may have been the founder of the Brookings Institution. It was apparently a distinguished family. At any rate, at Aberdeen, the leadership in ballistics fell to Oswald Veblen, who was in uniform as a captain and then later a major, I think, in the Ordnance Department. In the early days, about 1902 or Ď03, when Woodrow Wilson became president of Princeton, he had the idea of forming a group of people who were called preceptors.

Among the people that he brought here to Princeton, probably the most famous mathematicians were George David Birkhoff, Gilbert Bliss and Oswald Veblen. I think Henry Norris Russell, who was a great leader in astronomy, was also a preceptor. Birkhoff stayed here for a while, and then got a call to become professor of mathematics at Harvard. There is correspondence at Chicago or here, I forget where I saw it, in which he asked advice as to what he should do. He decided Harvard was the place where he would go. Veblen stayed at Princeton, and Bliss went to Chicago. When the Institute for Advanced Study was first founded, the first two professors were Einstein and Veblen. Veblen played an absolutely decisive role, both at Princeton University and at the Institute for Advanced Study. In fact his contribution in establishing great mathematics departments at both places was probably greater than any personís. I think thatís probably why there are superb departments at both institutions.

Stern:

Veblen had connections with the Ballistic Research Lab during World War II, did he not?

Goldstine:

Yes. Since he was still alive and very active, he was invited to become the chief scientist. It was his great contribution in the Second World War, to recruit a first class group of scientists -- not just mathematicians but scientists in general. In general he did a superb job of this. For example, he brought into the laboratory Edwin Hubble and Martin Schwarzschild, both of whom are first class astronomers. Rubble is dead now, but Schwarzschild is still alive and very active here in Princeton. Schwarzschild left in Proving Ground after a relatively short while, because he was very anxious to fight Germans and didnít want just to do scientific work. His father had been professor of astronomy and head of the observatory at Gottingen, and so was in a great tradition going back to Gauss. But Martin, as I say, wanted to be out in the field. In addition, letís see, who else? There were many people. For example, Gregory Breit was one of the people Veblen brought to Aberdeen. He brought many scientists including mathematicians into the place.

Stern:

Mathematicians and astronomers, then for the most part?

Goldstine:

No, he brought a lot of physicists too. Well, Breit is an example. I was trying to think of who else. Thomas, Llewellyn Hileth Thomas.

Stern:

Cunningham too.

Goldstine:

He brought Leland Cunningham to Aberdeen. I forgot about that. Another one he brought in, perhaps even more well known, was Ted (Theodore E.) Sterne. Theodore E. Sterne from Harvard. And there were just lots of people. He must have brought in 50 or 75 people, maybe more. Perhaps the most exciting thing he did, in some ways, was to form a Scientific Advisory Committee to the Director of the Ballistic Research Laboratory. That was a committee of very impressive scientists to insure the excellence of the work that was done at Aberdeen. I can remember some of the names of the people. There was, for example, I.I. Rabi. He was a very impressive figure there. He still is. George Kistiakowsky, whoís another impressive scientist. Johnny von Neumann. A man from General Electric named Albert Hull who was most impressive. Well, I donít know exactly what his total contribution to GE was, but he had an influence on the development of the magnetron, and he was just a superb experimentalist. He could look at something and tell you how the experiment should be done. He had just a tremendous feeling for this sort of thing. Let me see who else. There was Hugh Dryden who was a superb aerodynamicist in the government. There was Bernard Lewis, of the National Bureau of Mines. He knew a great deal about explosives. Joe Mayer, the physical chemist and others.

Stern:

You came to the Ballistic Research Lab in 1943, was it?

Goldstine:

Ď42, I guess.

Stern:

The decision with respect to the ENIAC was made in April of Ď43. Is that correct?

Goldstine:

Right. Yes.

Stern:

Now, from my readings of the documents, it seems to me that that decision was made rather quickly.

Goldstine:

Yes, it was made very quickly. If I may back up a bit, during the period between wars, Vannevar Bush had developed what was called the differential analyzer. This was a realization of an idea that Lord Kelvin had had, but didnít succeed in building, because he lacked a technological part, namely, a means of amplifying the torque or the output of an integrator. (This whole tradition of integrators really goes back to Kelvin and Maxwell.) But at any rate, Vannevar Bush had the wit to see that development by some man -- I forget his name now -- which had a key role in the steel industry, was applicable to the differential analyzer, and he made a working machine. Both the people at Aberdeen and at the Moore School of Electrical Engineering at the University of Pennsylvania saw the great use that each of them could make, using such a machine. In the case of the Moore School, electrical engineering had progressed by that time, the early thirties, far enough so that engineers were able to write down with a certain amount of facility, differential equations which would describe the behavior of electrical circuits. Their big problem was to integrate these equations. As a result of the work of Bliss, Moulton, Veblen and others, in the first World War, the people at Aberdeen were able to write down the differential equations for motion of projectiles. Their problem also was to integrate the equations. So, when the Bush machine first appeared, both Aberdeen and the University of Pennsylvania went to Bush and asked to be allowed to make copies of his machine.

He put the two groups together, and somehow they collaboratively produced two sister machines. The result was that when the war broke out, the people at Aberdeen turned immediately to the Moore School and rented their machine for the duration of the war. They also were having great difficulty in recruiting people to work at Aberdeen, because it is not close to any city. It is perhaps 40 or 50 miles out of Baltimore, and even further out of Philadelphia, and so it was very difficult to get people to work there. The Army didnít have adequate housing for its civilian employees and so the director of the laboratory formed a substation at the University of Pennsylvania, where the Moore School organized what were in those days ESMWT courses -- Engineering, Science and Management War Training. Lots of universities were teaching such courses to retrain people for useful war work. When I got to Aberdeen, which was in the summer of Ď42, I was taken to Philadelphia by Col. Paul N. Gillon who was in charge of the part of the Ballistic Research Laboratory which handled all the calculational work. We looked at the training program that was going on at the university, and it was a disaster. I shouldnít call it a disaster, but it was. The courses were being taught by some old gentlemen who had been retired for years from the teaching of mathematics. They were really just too old and too tired. Moreover the administration of the differential analyzer group hadnít yet been worked out very well, and the Armyís relations with the university hadnít either. It was clear that what Aberdeen needed was somebody to be in charge of the Philadelphia substation. Fortunately (from my point of view) I was picked to be the man. Thatís how I got related to the University of Pennsylvania. I then spent virtually my whole time at Pennsylvania, with a few periods in which I did some other things for Aberdeen. We turned the training program around at the university. We brought in my first wife, Adele, then John Mauchlyís first wife Nary, and Mildred the wife of a professor of Sumerology, Samuel N. Kramer. Adele, Mary and Mildred, formed our basic faculty.

Stern:

What is the relationship between this training youíre talking about and the ESMWT?

Goldstine:

These were the teachers for the ESMWT courses.

Stern:

Now, those courses included electronics courses too?

Goldstine:

No. There were also electronics courses at the Moore School. But our courses were dedicated to training programmers.

Stern:

The human computers.

Goldstine:

The human computers. Sorry, I said programmers, but at this time the word didnít exist.

Stern:

Aside from the obvious reasons about wartime needs, why was it that almost exclusively women who were hired for those positions? Or was the war the main reason?

Goldstine:

You know -- I donít know. Itís a good question that you ask. We never, as far as Iím aware, had an applicant who was not a woman. I think it was just like, perhaps, bricklayers, you know. You always see men in those jobs. Or you used to anyway. Maybe there are women now. Well, I did see in the newspaper just yesterday or the day before about a couple of women who run a house painting business. But I mean, that was an era in which there were jobs that were menís, and there were jobs that were womenís.

Stern:

This was then viewed as a kind of clerical thing? Would that be an explanation?

Goldstine:

Yes, I guess thatís right. I donít think it was something admirable. There were some men working at Aberdeen, in lower administrative jobs, having risen up from of being computers. But in general, the computer did dull work. To do numerical integration or numerical calculation of almost any sort, is inordinately dull. Itís like doing your income tax every day of your life, from morning till night. And so, I think youíre right. I think it was a kind of job which a man would not willingly go into, and therefore, the vacuum was filled by girls. Philadelphia turned out to be a good place. The original intention of the management at Aberdeen was to train the girls in Aberdeen and ship them from Philadelphia to the Proving Ground, but in fact, virtually none of them would go. So, what I did was to build up a group of a couple of hundred of them in Philadelphia where they lived. That worked happily. There was a small group, a subset of that bigger group which I organized to run the differential analyzer, and they were a very good group.

Those, plus the best of the people who did the hand calculating, formed the group out of which we chose the first programmers for the ENIAC. One of those was the woman who became the second Mrs. Mauchly. (The first Mrs. Mauchly died tragically by drowning. They were vacationing at the shore somewhere, and she drowned.) But at any rate, that was the beginning of the computer business at the Moore School. The next thing was related to the Dean of the Moore School. The Moore School in those days was a semi-autonomous unit of the University of Pennsylvania. If I understood it correctly, it had its own endowment and its own board of trustees, and had a kind of nominal relationship to the University. Subsequently it became an integral part of the University, but in those days, the Dean of the Moore School was a real potentate. His name was Harold Pender. Pender was a Southerner from North Carolina who had received his Ph.D. in France under Poincare, I think. (I know I was always conscious of being amazed that he had had this kind of an education, because it was such an improbable one for an engineer.) At any rate, the big thing that he did was to found the International Resistance Corporation, IRC, which made very superior resistors. In those days, so far as Iím aware, there were just really two big companies in America that made them, Allen Bradley and IRC. And I guess each of them was a small company, selling parts largely to radio hams until the war came along.

Stern:

I was curious about April, Ď43, that the initial memoranda that you suggested he write up took almost no time to be approved by the Army.

Goldstine:

Thatís right.

Stern:

Was that an unusual thing during the war?

Goldstine:

Well, I canít really answer that question, because my experience was limited to a small subset, you know, of the military. I think the situation was like this. Paul Gillon was an extremely competent, energetic officer, who understood a great deal about technical matters. He had been a student at the University of Michigan, at West Point, and then at MIT. He knew a lot, and he was a person who, if he put his faith in you, he put it in unreservedly. He didnít just go 50 percent of the way, and then make life miserable. He was convinced that the computing load at the Proving Ground was not going to be eased by any change of factors of two or three, or things like that. He saw that what really was needed was something like an order of magnitude, and he was prepared to go forward with almost anything, if it was only reasonable. I believe he formed a high opinion of Brainerd and of me. He was prepared to back that opinion. He was also a West Point graduate. The one thing that I admired during the war was how very effectively the Old Boysí Club of West Pointers operated. It was a beautiful thing to watch. These men trusted each other, or at least they knew what von Neumann used to call the ďmendacity factorĒ of each one of their colleagues. They knew how much to trust each person, and if they believed in a man, as convinced they believed in Gillon, there were no problems about getting money and approvals. The normal military tempo was slow, as all government bureaucracies are. The reason this thing went very fast was because Gillon was immediately convinced by us. He had been looking quite hard for a solution to the computing problem at Aberdeen and to solve it had brought the Moore School into the picture. It was he who put me there. He was determined to do something to break this log jam, and it took very little persuading of Gillon that something dramatic should be done, and here was a dramatic possibility. The amounts involved are really minute, compared to the total expenditure of the Ordnance Department per year.

Stern:

Your role and Brainerdís role seem to me to be focal, because Mauchly had this proposal eight months before.

Goldstine:

Of course. Of course. One of the things youíve got to realize is that Mauchly was never a person who was good at persuading other people. Mauchlyís whole career, I think, has always been almost one of a dilettante. Iím not using the term necessarily in a pejorative way. Iím just trying really to describe his career. I think, if you look through the standard journals, in physics and in electrical engineering, youíll find few papers written by Mauchly. My guess is that there arenít a half dozen. Some years ago there was a famous probability theory man at Princeton University named Sam Wilks who knew Mauchly for many years. Wilks said Mauchly would occasionally drive down here, and spend the day talking to Wilks about his ideas on applying methods of statistics and probability theory to meteorology, and then would just disappear, and nothing would come out of it in the form of a paper. This is just not, as you know, the normal academic method. And again, Iím not saying this in any disrespect of Mauchlyís abilities. Iím merely commenting that he was not a person who was good at starting something and seeing it through to fruition. In fact, to the contrary, he was bad at that. Thatís his great weakness.

I think thatís why Mauchly could have written many proposals to no effect because he always did them perhaps to entertain himself. I think they were an end in themselves. I mean, for him to reach the conclusion that one could do something probably satisfied him. Brainerd was a much more responsible, more mature, engineer. Well, in fact Mauchly was not an engineer. Mauchly is a physicist, and I donít say that disrespectfully either. I mean, his thing was not seeing projects through, but rather to have bright ideas. Brainerd looked at it differently. I think perhaps that it would be better to talk to him than to me, because I donít know what goes on in his head or anybody elseís head. But at any rate, once it got to us, I think as you say it went very rapidly, because there was very little argument. There was no doubt in my mind that it was only by a change of one or more orders of magnitude in computing speed that Aberdeen would get out of the hole it was in. I had very considerable respect in fact for Brainerd. I also had a very high opinion of Eckert, and I think it was these things which made it seem reasonable to me to get Gillon excited, to get the funding and go ahead. And as I say, it wasnít a big financial matter for the government.

Stern:

In terms of the funding.

Goldstine:

In terms of the funding, thatís right. From Brainerdís point of view, it was a big deal. I donít know, did he talk to you about this?

Stern:

Not about this.

Goldstine:

About his relation to Gillon? No, not about funding, but I mean about initiating the project?

Stern:

No.

Goldstine:

Well, it was a big deal, because I believe Pender was very dubious about it. Iím in no doubt that Brainerd literally put his neck on the chopping block. If the project had not succeeded, Pender would have chopped his head off. I think thatís the real great contribution of Brainerd. The greatest contribution that Brainerd made was to bring that project into the Moore School and give it a chance to exist.

Stern:

Now, initially, according to my records, RCA was to get involved in a formal way, with an aspect of the project, with respect to the counters.

Goldstine:

Right.

Stern:

Now, they subsequently decided not to do this.

Goldstine:

Right.

Stern:

Could you shed some light on some possible reasons why they made that decision?

Goldstine:

Well, I can tell you this. I can remember two visits to RCA with Moore School people to talk to Vladimir Zworykin and Jan Rajchman. They both were very remarkable people, and I have the highest regard and liking for both of them. One visit was made at which we saw a digital fire control director, which had been developed and built by RCA under a contract with the Frankard Arsenal, which is a branch of the Ordnance Department. That machine had in it a read-only memory, a network of resistors, the idea for which was originally described in an MIT thesis by a man named Perry Crawford, Jr. who is now with IBM. That memory worked very well, and became the idea for the so called function tables, of the ENIAC.

Stern:

Rajchman claims to have invented that independently.

Goldstine:

Oh, I should have mentioned that. Rajchman I believe without knowing about Crawford had done it too. Rajchman has had a number of things which have been (unfortunately, from his point of view) like this. The magnetic core use in the computer was another one. No, Rajchman is a very creative, ingenious and brilliant man. At any rate, that was a very impressive visit. RCA also had very good electronics counters, and indeed, the counters for the ENIAC were developed out of the ones at RCA. Thereís just no question about that. Eckert started with the RCA counters, then modified them and improved them to the point where he was satisfied. The interests of the Moore School and RCA were just too divergent to make anything like collaboration work. Eckert had everything to gain, I believe, by going it alone, and RCA had little to gain by going into it. I donít think that the administration of RCA particularly believed in this project, so why should they go into it? I certainly donít think that they had at that time any picture that there would be any market for electronic computers in a commercial way, so why should they go into it? I also donít believe that Rajchman and Eckert would have got along. Thatís another reason. I think from Eckertís point of view, what he wanted were these ideas, which he got very quickly, namely, the function table and the counter business. I donít think that there was ever really a very serious idea of collaboration or that chance was ever very high that is would ever happen.

Stern:

I see. In addition to investigating the RCA counters, there were reports from the NCR.

Goldstine:

Thatís correct. Eckert was or is a very good engineer, and he studied all the things that he could study, and NCRís reports were among them. There were other things too.

Stern:

The Lewis Counter as well, I understand.

Goldstine:

Yes.

Stern:

Would you say that was a kind of pro forma thing, to investigate, all those, or did he really feel that there might be some --?

Goldstine:

I think Eckert knew a lot about the subject before he started. He had studied many things very deeply while he was a student. I think he went into the field the way a good scientist should, namely, to survey the literature, or what amounts to the literature of engineering, the circuitry that exists. Then perhaps he was looking for whatís wrong with past ideas as much as he was looking for whatís right with them, you know.

Stern:

Thatís a good point.

Goldstine:

At any rate, even though he doesnít like me any more -- Iím sure he dislikes me heartily -- I had great respect and regard for him, and still have high regard for him. I think his contribution was enormous. He was a very young man. By all odds he was much, much too young to be the chief engineer for the project. But in fact, he was the only possible chief engineer for such a project. It took a person of great ability, which he had. I think itís just remarkable that he made it go. I donít believe anybody else could have, because he took incredible care to produce something which would be really reliable. He had to do this in the face of all the best professional opinion of the Hazens and the Caldwells, etc. of MIT and nearly everywhere else.

Stern:

Letís discuss NDRC and that group. Aside from, as we said, perhaps the additional motivations that the Moore School got from this hostility that appeared, were there any, either negative or positive effects of NDRCís hostility? That is, did their holding back of the reports initially have any effect?

Goldstine:

Oh, you mean effect on the ENIAC project, not on any other thing.

Stern:

Exactly.

Goldstine:

I donít think it had any effect at all, except perhaps to make the Moore School group a little more insular that they might have been otherwise.

Stern:

What do you mean by that?

Goldstine:

Well, I mean, had the MIT people been more cooperative, very likely the Moore School people would have made frequent visits to MIT. Because MIT is the natural centrum for engineering, particularly in those days, electrical engineering. The fact is that the Moore School existed as if MIT were in the middle of Frankfurt or Hamburg or someplace, you know, in enemy hands. At any rate, there was no relationship, at that point. There was some relationship during the EDVAC period, but not in ENIAC period.

Stern:

Do you think had there been more -- I assume you do, based on what youíre saying -- had there been more collaboration, if MIT and Moore School developments might have proceeded more rapidly?

Goldstine:

No, I donít think so. I really donít. I donít believe they could have proceeded more rapidly. The Moore School went forward at a tremendous pace. You could take that as a model of how quickly you can develop something. The manufacture of the machine took longer, and that would not have been helped if they had gone to MIT either, because thatís something that, you know, had to be done at the Moore School. But I think that the influence of the Moore School in the electrical engineering world might have been a different thing if there had been more collaboration. Yes, I think, as I tried to say, I think the Moore School became, or maybe was, insular, and the MIT people were insular too. My own opinion is that Jay Forrestorís greatest contribution was breaking down the analogue computer tradition which MIT had fostered for so many years, and bringing the digital tradition into MIT. I think that also kept the two schools far apart, to the detriment of MIT. But, as I say, I donít think that collaboration would have speeded anything up. Eckert was not going to let that machine out of his hands. And there were other people who did make other contributions, which should have been acknowledged. But, I think that the man who kept everything at his fingertips, and who welded the whole thing into a whole, was Eckert. No doubt about it. And thatís not to belittle the accomplishments of (Arthur) Burks and Sharpless and Shaw and all the others.

Stern:

Could we just shift a little bit? Iím interested in the mathematical considerations, with respect to the ENIAC and with respect to later computer development. Was there significant attention given, during the development of the ENIAC to questions of numerical analysis, things like that?

Goldstine:

Well, yes. Both Cunningham and I were much interested in this part of it, and Burks was too, at the beginning. Burks actually did, as I remember, write a rough program to see that the ENIAC was of a size which would fit the ballistics problem. Then the Moore School hired a man named Knobelauch who did some comparable kind of work. Also, the university gave a contract to a famous analytic number theorist named Rademacher. Rademacher was a distinguished emigree from Germany, and he and Harry Huskey did some studies on rounding errors.

Stern:

This was after the development stage.

Goldstine:

Right, this was after.

Stern:

What about before?

Goldstine:

Before, it was Cunningham, me, Burks and that was about it.

Stern:

So that initial progress report, Iím thinking of the December 31st, 1943 progress report, the first one, in which thereís an extensive write-up on rounding errors and other aspects of numerical analysis. You wrote that, is that correct?

Goldstine:

Iím not sure. You know, I donít really remember. Iíd have to look at it again and see if I could remember, if I was the sole author, part author, or what. But I suspect I was the author. Youíd better let me see it, before I would say thatís the case.

Stern:

Did Mauchly provide any input to that aspect?

Goldstine:

I donít think so. My remembrance is that he did not. I think he was interested, mildly, in that subject, but I donít remember his being interested profoundly. I think that the amount of work on the rounding error business was minimal. I guess we all felt confident that the mathematical problems would be small as compared to the electrical, to the engineering difficulties.

Stern:

But later on it became more of a consideration?

Goldstine:

It became more of a consideration, sure. Well, Iíll tell you, at the very beginning we had the experience from the First World War and subsequent to that of people like Bliss and Veblen, etc. who had done a lot of work on these problems. What we were going to do was very much the same thing, only very much faster. That was our original picture of it. So we always had this feeling that if you didnít do anything else, you could just do exactly what was done before, and just keep a few more decimal places, you know. You were bound to be pretty safe, because the things were just obviously stable.

Stern:

People like Rademacher pointed to the fact that that just was not true, judging from his Moore School lecture in Ď46.

Goldstine:

Right. Right.

Stern:

When did this shift in attention to mathematical problems occur? Did it occur when von Neumann got involved?

Goldstine:

Not only von Neumann. As we got on this thing, such problems began to become clear, at least to Cunningham and me. Maybe Iím not giving Mauchly proper credit, but I donít think he or Eckertís interest was ever very much in what we were going to use the machine for. They had some interest, but I donít think a deep one. It began to become clearer to us that the machine, even though it was being designed and built to solve ballistic equations, was a much more versatile instrument. Then, and only then, did it become gradually clear that there would be more and more mathematical difficulties connected with things such as rounding. The entire literature on rounding error at that time could have been put into oneís eye and it wouldnít have hurt.

Stern:

I know, Iíve been trying to uncover the material. It hasnít been easy.

Goldstine:

The only material that I really know of thatís any good is in Gaussís book on THEORY OF THE MOTION OF THE HEAVENLY BODIES. Letís see if Iíve got a copy here. He does write on the subject, and he does show how careful one needs to be in calculating the motion of objects, you know, celestial objects, how you have to be careful how you set up the problem. Here it is. THEORY OF THE MOTION OF THE HEAVENLY BODIES MOVING ABOUT THE SUN. There is a discussion in there of rounding errors. But thatís about the only discussion that you can find, except for perfunctory ones. The kind of standard books of that time. Hereís one of them, Scarborough -- NUMERICAL MATHEMATICAL ANALYSIS. I think he was the man who taught, yes, taught at the Naval Academy in Annapolis.

Stern:

When was that book published?

Goldstine:

1930. But you see, there are things like ďaccuracy in the evaluation of a formula,Ē ďgeneral formula for errors,Ē ďabsolute, relative and percentage errors,Ē ďrelation between relative errors number, significant figures,Ē ďeffective error and the tabular functionĒ -- So there was some discussion there. Then there was another book which was the numerical analystís bible of the time: Whittaker and Robertson, which must be around somewhere. And thereís a very modest amount in there, even less in there perhaps than in Scarborough. But that was the literature of that period.

Stern:

How about the Southwell book?

Goldstine:

I donít even know that there is a book of Southwellís. I know who Southwell is, and I know Southwell worked on relaxation methods, but I donít remember he had written a book. So I canít answer you.

Stern:

There are two books that Iím familiar with. One, RELAXATION METHODS IN THE PHYSICAL SCIENCES, and one for engineering, that has been referred to, but I didnít know whether it was being used.

Goldstine:

I didnít know about Southwell until considerably after the war.

Stern:

I see.

Goldstine:

So thatís all I can answer you on that. Let me see what Hartree says. Letís see if he mentions Southwell. Thereís a subject index, thereís an author index -- name index -- Southwell, yes, there are a lot of references to South -- well. Let me see if he quotes a book or books of Southwell. Yes, he quotes two books, RELAXATION METHODS IN ENGINEERING SCIENCE, Oxford, 1940, and RELAXATION METHODS IN THEORETICAL PHYSICS, Oxford, 1946. Well, I think Iím safe in saying that we werenít aware of Southwellís books.

Stern:

Yes. In fact, I recall Cunningham sending a report in, in which he gives a review of the literature, and most of the material was really dated.

Goldstine:

Sure. I donít remember talking with Cunningham ever about Southwell. And I think astronomers, by and large, were in the tradition, coming down from that work of Gauss, with modest improvements since. But by and large, the feeling was that if you have to add a few more decimal places, OK. I believe Iím safe when I say that the only one who really understood the kind of exponential growth of rounding error was Gauss. He clearly knew it and saw it, and he described it in his book on THEORY OF MOTION. But people like Cunningham, I donít think ever saw that or if they did, they didnít recognize it as likely to happen.

Stern:

In the computer error, who would you say is the first to appreciate that problem?

Goldstine:

I think, von Neumann and I. There are indications of it in studies. I gave the contract to Rademacher, so I know that I was thinking about the problem. And Rademacher did a beautiful job. No question about that. But I think it was von Neumann and I who first got concerned about it. Whether it was von Neumann or I, I donít know any more. But I think we were probably, in modern times, the ones who took up again the tradition of Gauss, and who saw as a real live thing that the rounding error business could in fact build up, and that it just wasnít enough to take two more decimal places, chop the answer off to five places and call it quits. I think that may be one of our great contributions to that subject.

Stern:

Thatís very interesting. With the involvement of von Neumann, in Ď44, wasnít his attention immediately to the use of computers, that is, mathematical use of computers? Did he feel for example that this was something that would have to be justified to the scientific community?

Goldstine:

Well, at that point in history, von Neumann was a consultant to a variety of places. Aberdeen was one and Los Alamos was another, as you know. And at Los Alamos he was the intellectual leader of the implosion group. As you know, Los Alamos, pretty much through his urging, had rented a number of IBM machines. They were doing these so-called implosion calculations like crazy. So he was completely embedded in the computing business, and he had thought about all these problems at great length. His primal interest in computers was to use them; he was a great consumer of computers. Just because he was a superb scientist himself, and had interests in all kinds of problems, he was also interested in the computer as a piece of physical apparatus. I think he had very little interest, perhaps, in the computer as a piece of engineering. But as a piece of physics, he certainly was very interested. I guess, I would say his chief interest was to use the computer, second interest was in designing better computers, and probably as a poor third would be the engineering aspects.

Stern:

What about the idea of having to make a pitch for computers, I donít mean in terms of the Institute, I mean in terms of the scientific community?

Goldstine:

Sure. Well, --

Stern:

-- that became a problem, from what I understand. Many scientists did not recognize the computer as being a tool other than --

Goldstine:

-- sure. Everybody who had anything to do with the atomic energy program at all recognized, first of all, that von Neumann was one of the greatest figures in the business. And secondly therefore, that the computer had to be one of the most important things, because of von Neumann. So I donít think he had a lot of problem to persuade people. For example, very early, he talked to Fermi about these things. I donít remember the date any more, but Fermi had me come out to the University of Chicago, and he spent two days or so talking to him about computers. I think part of the time at least Sam Allison was there. But the only one I remember with tremendous clarity was Fermi, who was at one end of the table, and I at the other, and he just quizzed me steadily for the two days.

Stern:

You were talking about Fermi.

Goldstine:

Yes. Well, Fermi spent two days. My memory may be at fault somewhat, but I remember his tremendous intensity. This was my first real meeting with Fermi, and I was just enormously impressed. The penetration of his questions was fantastic, and he was so totally unlike von Neumann. Conversation with von Neumann would be a conversation mixed with anecdotes, you know, jokes, all kinds of things, periodically to lighten it up. With von Neumann you went through intense periods, and then a joke, and then back to work, sort of thing. But with Fermi, at least on that occasion, it was just like a doctoral examination, you know. It was just a steady pushing of questions and answers, and following up in whatever directions the questions led. I just thought it was incredible how brilliant this man was, and how profound. At any rate, let us return to von Neumannís influence. I think that the intellectual community was maybe divided at that point. The people who understood von Neumannís greatness, and therefore accepted his judgments, such as the whole atomic energy community, for example, started to build copies of the Instituteís computer. There was no lack of confidence, I would say, in that community, in von Neumannís ability. There were places as far away as Israel, Australia, many many places all over, academic institutions as well as government ones, that wanted to make copies of computers.

Stern:

What were some of the reasons he decided to build his own computer. As opposed to remaining as consultant on EDVAC?

Goldstine:

You know, when the war came to an end the ENIAC project flew apart. He and I had a long and very serious conversation about what we should do next. Thatís the way to put it, I guess. What should we do next? The possibilities were, for him and for me to attach ourselves in some way to the Moore School. That did not seem attractive to either one of us. It probably wouldnít have been attractive to the university either, but Iím not sure. Pardon?

Stern:

What are some of the reasons why it wasnít attractive?

Goldstine:

Well, it wasnít for us an exciting institute at that time. Both of us would have looked at the place from the point of view of the mathematics department, really, and at that point, its mathematics department was in poor shape. It had one person that I can remember, John Klein, who was secretary of the American Mathematical Society. He was an utterly lovable fine person, but Iím not even sure that John was still alive at the end of the war. He died shortly after that. And there was nobody else that I could think of, except Rademacher. While there was certainly nothing wrong with Rademacher, we didnít want to be a department with perhaps only one first-rate colleague. I donít think either one of us really felt that we wanted to be just a consultant to somebody else. I think both of us felt that we want to impose our own ideas, and thatís not really something you can quite do if just a consultant, you know. Besides, as the situation stood at the end of the war, I was on leave from the University of Michigan. I had been promoted in absentia to assistant professor. Under normal conditions, I would have gone back to Michigan0 Von Neumann would have gone back to the Institute for Advanced Study. The whole concept of consultant anyway was a problem. The picture was of returning to the pre-war days, which meant no money for anything. It just didnít seem to be sensible. So then the possibility was to go somewhere else and take the whole business with us, and thatís when von Neumann discussed the matter here at the Institute with his colleagues, and persuaded them to establish a project here. Thatís how we got the controlling interest, so to speak in the thing, which we wouldnít have had otherwise.

Stern:

How easy was it to convince the administration?

Goldstine:

Well, I donít really know that first hand. I know it only through von Neumann, since it was he who initially dealt with Frank Aydelotte, the director of the Institute, and his colleagues. Now, thereís little doubt that whatever Veblen may have felt about the importance in the computer, and Iím not sure he had much interest in it, he certainly thought von Neumann was just a marvelous mathematician and person. Von Neumann was, in some ways, Veblenís intellectual son. Veblen had no children. Johnny was, I guess, the person he would have wanted, if heíd had a son. So he had a tremendous interest in von Neumann. There was Jim Alexander here, who had been a student of Veblenís, and was a great topologist in his own right. Alexander had a lot of interest in electronics. He just loved electronics. And so Iím sure he felt fine about it. I donít think the others of von Neumannís colleagues were opposed to it at all. Marston Morse, who was another colleague, had in the Second War filled the role in Washington that Moulton had filled in the First War. So I donít think he was against the idea.

Stern:

It just seems it was a very atypical project for the Institute.

Goldstine:

It was. It was atypical, thereís just no question about it. As time went on, the atypical quality of it was the seed that caused the destruction of the project, in some ways. I guess I quoted in my book the statement of Aydelotteís to the trustees, about comparing the thing to a huge telescope opening up new knowledge and all that. In the long run, it was so atypical that it couldnít continue as a piece of this place, because this place is too small. And also, that was the time in the history of the world when it should have gone to the industrial world anyway. And it did. So that all worked out well. But I think that in retrospect, it might have been better had we taken it to some other place. Well, even if it had gone to some university, a big university where it could have fitted easily, and where we could have had complete control of it and all that, I think in the long run, what I just said before is really the crucial point. There was a point in the history of that subject when it was crucial that it should be in the university, and then there came a point when it was crucial that it should be out of the university. The point, that crossover, came at about the mid-fifties somewhere. That was the time at which the future of the computer had to lie with industry, and it eventually did, no matter what happened in the university world. So I donít think, in looking at it, even if it had gone to a university, which would have opened its arms up and continued to hold them open, it would not have gone to industry. The great excitement eventually would have been industrial excitement and not academic at all.

Stern:

With respect to this relationship between academic and industrial -- von Neumann offered Eckert a position?

Goldstine:

Thatís correct.

Stern:

-- on the project, and then rescinded the offer, on the grounds that Eckertís commercial interests were too overriding.

Goldstine:

Right.

Stern:

Now, considering the fact that von Neumann himself, a few years later, was a consultant to several industrial corporations --

Goldstine:

Right.

Stern:

It doesnít seem likely that he was against a person having commercial and academic interests.

Goldstine:

Well, I guess you have to understand this in context. We made an offer to Eckert. We did not make an offer to Mauchly. Eckertís parents were very much opposed to leaving Philadelphia. He lived near them. They wanted him to live at home -- Philadelphia.

Stern:

Was he married, Eckert?

Goldstine:

Yes, he was married. His wife worked for us on the project as a draftswoman, and she was a very nice person. She had a very unhappy ending. But she was a very nice person. We were very fond of her, my wife and I. At any rate, there was great pressure put on Eckert not to go to Princeton. We went many times for dinner to Eckertís house. His father told me, at dinner, that he wanted Eckert to go into business and, in Mr. Eckert Seniorís words, ďto make a million dollars,Ē and then to quit the industrial world and do whatever he wanted, in, say, the academic world, but with this backlog of money. So Eckert was being torn by Mauchly, who of course wanted to stay associated with Eckert, because Mauchly really needed Eckert. He was torn by his wife, who wanted to stay in Philadelphia close to her family. He was torn by his parents, who wanted him to stay in Philadelphia. And he was torn by his own desire to make a lot of money. He consulted various people at the Moore School. I can tell you one thing which I remember, that I donít want to attribute, but there was a man at the Moore School who advised Eckert that Einstein was a very greedy person who stole ideas from young people! That was one of the arguments Eckert gave me at one point for being reluctant to come here. So at any rate, his ambivalences about coming here were enormous. It wasnít that he wanted to have a consulting arrangement with somebody, but he couldnít ever decide where he really wanted to be. It was a different thing from the picture youíre evolving in your mind of a relationship, of consulting occasionally for industry. Thatís just not the way it was. Von relationship to industry was a consultant, which meant he would consult perhaps one day a week, or a couple of days a month or something like that. It wasnít like that at all.

Stern:

I would suspect, although this is just a guess, itís not something Iíve documented, that after the war there was a lot of feeling on the part of people about commercializing on wartime pursuits and on government-funded research of some sort.

Goldstine:

Yes.

Stern:

And I was wondering if von Neumann had some biases against people who wanted to do just that sort of thing?

Goldstine:

No. I donít think so. What we wanted was a chief engineer who would be for real. I mean, you know, thatís a fulltime job. That was the thing we wanted. Both of us watched this whole crazy rigmarole at the Moore School over the patent issue, and we were determined we going to have such a problem. So we were determined we were going to have a patent policy, and, in fact, we retained a patent lawyer from Washington to draw up patent arrangements here, so that we wouldnít get into such a boat. Von Neumann at that point certainly believed that the fruits of a project which was funded by the government ought to belong to the public. I think we were naive in that respect, but I have the same opinion now. I think the thing we didnít understand is that if you put things in the public domain, in a certain sense, while everybody gets the benefit of them, that means nobody really does get the benefit. That is to say, I think, that our technological society is based on people taking industrial risks, only if they see prospects of great gain. Thatís the advantage. I guess thatís why patents are important. I didnít see it, and I donít think von Neumann saw it at that time either. I think both of us thought that you put things in the public domain, and just let everybody scramble for them. Now, actually people donít scramble for them that way. They perceive otherwise. I think maybe that was a misconception we had. But itís true, that was our comprehension of it at the time.

Stern:

Just one last questionÖ I read Bochnerís and Halmosí tributes to von Neumann. Both claim that von Neumannís computing interests were not really indicative of his importance with respect to mathematics. That is, it was not a very important speech of his life.

Goldstine:

Well, of course both Bochner and Halmos viewed, in a way, von Neumannís going into the computer field as a departure from mathematics. Itís a question of what you define as mathematics, and thereís a tremendous difference between physicists and mathematicians, on this whole subject. Itís a real typical one. Some mathematicians have a very special concept of what is mathematics. Physicists have a much broader concept of what is physics. A man can shift from high energy physics, letís say, to developing a laser, and be viewed by his colleagues with just as much respect as when he was working in whatever his prior subject was. Because I think physicists have a tremendous regard and appreciation for equipment and apparatus, and the possibility that this particular piece of apparatus, which one may not be interested in today, may tomorrow become just the decisive tool that one is going to need. And also, physicists have a feeling for apparatus, so that even if theyíre not interested in the thing, in itself, they can still see the technical problems, and perhaps rejoice in the other fellowís accomplishment. Mathematicians arenít like that at all. They donít see this. The computer is to some of them a whole business in technology, thatís outside the realm of mathematics, as perceived in the 1940ís and 50ís. Of course, it caused a tremendous cleavage, so that thereís now a whole new discipline called Computer Sciences, because it couldnít find a home in the mathematical world. Bochner and Halmos both viewed von Neumann as having stepped outside the mathematical world, to work with the computer. I think it was an enlargement, or would have been nice to have viewed it as an enlargement of mathematics. So I think that Bochner and Halmosí evaluation are those of pure mathematicians, viewing a pure mathematicianís departure into another area.

Stern:

They acknowledge von Neumannís contribution to what we normally refer to as applied mathematics but yet this is another issue, this computer.

Goldstine:

Thatís right. Applied mathematics was already a subject which I think Bochner and Halmos would have accepted for von Neumann, but this wasnít. This was more like technology, you know, their perception of it. At least at that time. And you know, itís a question of what you want to name something really. You see, at one point the school of mathematics here included, among other people Niels Bohr, Dirac, Pauli, Frank Yang and T.D. Lee. Then the Instituteís School of Mathematics split into two, and the School of Natural Sciences got formed. Now, under that, itís not clear to me, you know, among those original people, whether they would have felt more comfortable in the School of Mathematics or the School of Natural Sciences. Whether Einstein would have been in the School of Mathematics today, I donít know, or whether he would have been in the School of Natural Sciences. That isnít peculiar to the Institute. At some point there in the fifties, or whenever it was, that was a movement in the mathematical world, which was a movement to retain pure mathematics. I mean, so much applied mathematics developed during the war period that I think some pure mathematicians became afraid that they would be engulfed by the applications, and that a whole tradition which had been built up over a couple of thousand years would get swept away, So they stood firm against any change, and that caused lots of bitterness. I guess all thatís gone away now. You know, now that a new equilibrium has been established. All the turmoil I think was a result of a steady state changing to a transient one and back into a new steady one again, but during those transient days, things became very hectic.

Stern:

Itís interesting that these mathematicians did not consider computers applied mathematics but technology, as you said, despite the fact that it was so clearly, from von Neumannís writings, a mathematical pursuit, at least to some extent.

Goldstine:

Yes. I think maybe you have to realize that what von Neumann was interested in was not the entire computer field, you know. The computer field is an enormous one, which goes from electrical engineering and technology of all sorts, all the way over toward the very theoretical subjects.

Stern:

People like Hartree and Lehmer and Rademacher were all into the same sorts of aspects.

Goldstine:

That was because in the beginning von Neumann attracted many people from mathematics into his orbit. But in the long run, when it went into industry and became a big thing, then the need for technologists, as technologists, became very great, and so you see an enormous growth of that sort of thing.

Stern:

They moved away, then, from the mathematical point of view.

Goldstine:

Right. Right.

Stern:

Well, thank you very much.