Notice: We are in the process of migrating Oral History Interview metadata to this new version of our website.
During this migration, the following fields associated with interviews may be incomplete: Institutions, Additional Persons, and Subjects. Our Browse Subjects feature is also affected by this migration.
We encourage researchers to utilize the full-text search on this page to navigate our oral histories or to use our catalog to locate oral history interviews by keyword.
Please contact [email protected] with any feedback.
This transcript may not be quoted, reproduced or redistributed in whole or in part by any means except with the written permission of the American Institute of Physics.
This transcript is based on a tape-recorded interview deposited at the Center for History of Physics of the American Institute of Physics. The AIP's interviews have generally been transcribed from tape, edited by the interviewer for clarity, and then further edited by the interviewee. If this interview is important to you, you should consult earlier versions of the transcript or listen to the original tape. For many interviews, the AIP retains substantial files with further information about the interviewee and the interview itself. Please contact us for information about accessing these materials.
Please bear in mind that: 1) This material is a transcript of the spoken word rather than a literary product; 2) An interview must be read with the awareness that different people's memories about an event will often differ, and that memories can change with time for many reasons including subsequent experiences, interactions with others, and one's feelings about an event. Disclaimer: This transcript was scanned from a typescript, introducing occasional spelling errors. The original typescript is available.
In footnotes or endnotes please cite AIP interviews like this:
Interview of John Mauchly by Nancy Stern on 1977 May 6, Niels Bohr Library & Archives, American Institute of Physics, College Park, MD USA, www.aip.org/history-programs/niels-bohr-library/oral-histories/31773
For multiple citations, "AIP" is the preferred abbreviation for the location.
In this transcript, Mauchly provides background information on his youth, his experiences as a physicist and his plans for the ENIAC in the pre-World War II years.
I know, for example, that you were born in 1907, but I don’t know very much about your family. William
Well, my father was a teacher of physics. He then became a scientist and research worker in geophysics at Carnegie Institution of Washington in various departments. One of those was the Department of Terrestrial Magnetism.
You were telling me about your childhood in Washington. Do you have any brothers and sisters?
Yes, I have one sister, seven years younger than I, who now lives in Florida. At the time when I was in college, she was just finishing school. She went on to Tufts College, where she met her husband. She has pursued a kind of a career in music all these years, but she wasn’t a concert musician. My mother thought she would be. Her mother and my mother felt that I would be one too, so she tried to train me that way.
Really? Did your sister major in music in college?
I think so. I was so wound up of course in my work at Hopkins in graduate school and so forth that I didn’t pay too much attention to what she was doing except that I knew that she was off to college to study music.
Did you ever seriously consider another career for yourself?
I never seriously considered it. I joined the glee club when I went to Hopkins, my first year, but more as a social function, I think. I liked to sing in a group, but I was no piano player or vocalist, really, so that was a dead end as far as I was concerned. I was really more interested in, first of all, being an engineer, and later, in being a scientist. I never knew quite which one I wanted to be. The thing that somewhat influenced my studies by the end of high school was finances. Not knowing how I might financially get a college education, since scientists like my father weren’t paid very much, I thought it was necessary to go to a school where I could earn my way. Such a school was the one that my father had gone to, the University of Cincinnati. He’d gone there for graduate work, and knew a man named Schneider who was the dean of engineering at that time. They had started something called a cooperative plan there. The-cooperative plan would allow you to spend, say, six weeks or whatever at study, six weeks in some industrial work, and so forth. Such plans of course sprung up in lots of places — Antioch, I believe, has one, and Drexel has one and other places too. But this was one of the first to be very well tried out. So there was where I was applying, the University of Cincinnati. But some time in my senior year, an English teacher, Miss Diefendorf, said that I ought to try going to Hopkins. Her brother had a scholarship there, or had had a scholarship there. She thought that since I was a resident of Maryland, even though I was going to school in the District, as a Maryland resident, I was very eligible and had a good chance to get a scholarship to Hopkins. Well, like most people, I thought Hopkins meant medicine. But she already told me her brother went to an engineering school, and I found out that Hopkins did indeed have an engineering school, which was apparently partly supported by the state, and that’s why they gave state scholarships. If I were able to pass the examination and get the approval of the State senator of my area, it would be clear and easy to do. It turned out that way. I went over there and took the examination. That was fine. Then I met with the senator, who said he was certainly glad once in his life to have had somebody whose examination marks justified the appointment. Sometimes he was called upon to make appointments without such good results. Anyway, so, I spent two years at Johns Hopkins studying electrical engineering. But it turns out that some of the instructors in the basic courses, such as physics, which I was taking as an engineering course, were already known to me. They were people working at the greatest institution, and the Bureau of Standards, who were coming to Hopkins to get their PhD’s. This brought me back to the direction which I might have taken otherwise anyway, and that is, physics. It was much more interesting than that hybrid thing that the engineers were doing, because the first years of engineering turned out to be a very routine cookbook type of thing. So I got the United States Steel Handbook, for how to design steel, how many rivets, what the spacing should be, and all the other little details of how you do structural design. That was all basic engineering, no matter what you chose to specialize in, whether electrical or mechanical or, I forget what they call it, the design of sewage systems —
Sanitation engineering. And so on. It was very easy for me to persuade myself that the really interesting work in this world was done by scientists, physicists, and that engineers were just people who applied these things in a routine way, and really got no kicks out of life, except perhaps the same kick that I got when I used my Mechano set to build some operating device, such as an elevator or something like that. I found out that it was possible to transfer from the engineering to the physics course, not only with no loss of credit, but with lots of advantages as I saw it. For one thing, at that time Mr. Godenaugh, who was the president of Hopkins for a short while, had a plan whereby you did not have to have a bachelor’s degree to go on into graduate work. It was therefore possible for me to enroll directly in the department of philosophy right away in my third year of college. The graduate school of philosophy included the department of physics. Furthermore, as a consequence of this, I was immediately eligible for earning a stipend as an instructor in physics, because I knew enough physics to teach those first year laboratory fellows. So I did that. Then, there was another area in which I could earn a little money. I had vacated the engineering scholarship of course when I went to graduate school, but there was something called a Quincy scholarship, which was available to physics students, and so I could be partly supported by that. Another thing happened; in that transition I got happier and a lot more interested in my work. One of the things that happened was that Hopkins again changed presidents, and the man who became president then was Joseph Sweetman Ames, well known to the NASA administration, Langley Field and other places, which have Ames laboratories. Well, Dr. Ames had been the head of the department of physics at one time, and I guess he was Dean Ames when I first heard about him. He became president of the university, and I had to somehow satisfy him, with respect to any scholarships that I got. So it was sort of sad, from my point of view, to find out that the president, instead of leaning over backwards to favor the people in the physics department, he was going to lean over backwards to prove that he didn’t favor such people, I think. So he decreed that I didn’t need a full Quincy Scholarship, that he would split it and give me a scholarship. And with that, of course I forget exactly which year, but I also tried to earn some of my living working in the Johns Hopkins Library which familiarized me with the accession of books. Spending time in the catalogue room, and then pasting all kinds of book plates in them, to say where the money came from that bought that book, became very familiar with a lot of the ordinary library practices you might say. For instance, I noticed that there was a branch library system, and that a book which looked awfully interesting to me on physics might have been bought by the chemistry department. So over there, in the chemistry building across the quadrangle, was the library of the chemistry department, which was getting this very interesting book instead of the physics department, because it had a fund which bought this book — and on. Well, of course, I was going to the public library but this was so much more handy to me. We had a new physics building, built let’s say just in time for my graduate work. My early work in the engineering school had been within an engineering building, but the physics people also had to use that building, because they hadn’t yet had their building. I think Morgenthau was one of the ones whose names were attached to the building.
How did your father feel about your academic work? He must have been delighted with your decision to switch over to physics.
Well, that might have been so, except that in 1928, he died. He had a rather long illness, and nobody knew exactly what his problems were or what. The doctor who was treating him felt that he had sleeping sickness — a kind of meningitis. I think by the time I got out of college, he was very ill, and in December, 1928, he died. But this did not interrupt my career. He’d provided my mother with plenty of insurance, because he’d been a teacher, and I guess the scientists of Carnegie were also eligible for the Teacher’s Annuity Insurance [TAI]. So my mother was never impoverished you might say. She carried on. We kept the house in Chevy Chase and she rented some rooms to boarders. But as far as I was concerned, my scholarships allowed my education to go ahead. My sister’s education also went ahead at Tufts College.
How was the physics department at Hopkins?
Well, as far as I was concerned, it was very exciting. These were the years when quantum mechanics was just coming into full bloom. So there were no textbooks in English which were considered satisfactory — for some of these things. So, we had to read papers and books by Reisenberg and Bohr and other people, in German usually, because there weren’t any textbooks. I bought a lot of books then... There began to be good books in English, but it was kind of rough for me to have to learn some of these things from German books. It shouldn’t have been rough, because my father spoke German. But he never spoke it in the family. We came from a family in Ohio where German was apparently a rather common tongue, and my mother’s maiden name was Scheidermann so her family, you might say, was on the German side too, and so if they had spoken German in the family, why, I might have had that as a second language. But they didn’t do that. So I still was naturally disposed to scan or however you pronounce it, better than other languages that I tried to learn subsequently. I had no difficulty passing the language examination in French and German for the PhD requirement. But I never considered myself to really have the German so that I could take the Born’s or somebody else’s book and go right through it. I had to also learn some mathematical ideas, concepts of quantum mechanics, if you don’t know the language, to be sure that you’re getting the ideas straight, as you probably realize. At any rate, in the 1930’s here I was in the physics department, and, I started incidentally earning money in the summertime by working at the Bureau of Standards, where my father had many friends some of whom were members of the church in Chevy Chase, Sunday school teachers, and, so it was a rather pleasant thing to go down there and talk to some of these people and say, “Where can I get a summer job?” I was hired as an assistant physicist or something of that sort, in the beginning. One year I worked in what they called the mechanical laboratory, as it was with a man who was very well known in those days, Lyman J. Briggs, who was director of the mechanical division, and later became the director of the Bureau itself. And later; I worked with Alan Astin, who was one of the people who came to Hopkins to get his PhD while I was there, and he became Director of the Bureau.
I was going to ask if there were any of your fellow students.
E.M. Condon, as we know, had a sort of a sad harassment in the McCarthy Era. He and his wife, his wife particularly, had gone to cocktail parties where there were supposedly “Pinkoes” or something. He got really terribly harassed in various ways. The work at the Bureau this first year was calibrating current meters. This does not mean electrical current, but the current of a stream, a river. You would hang a fish line or something in the Mississippi or other places, to find out how the water’s going. They had to be calibrated, and it was my job, for instance, to ride on the special kind of car which travels on nice straight big tracks over a pool of water, with a hydraulic drive. This was the first I’d ever heard about it. It would maintain a constant speed, and have various observing stations down a sort of artificial canal. One could find out what this meter usually read, when it was towed at a known speed and so on. Then you’d go back and use a Thatcher slide rule, to make plots to tell the US engineers or other people concerned with the river gauges and so on, that here’s the calibration for this. There were other departments that did such work. It was no wonderful exciting thing to calibrate fever thermometers. I never had to do that. But I did almost as badly in that I had to test numbering machines which mainly were tested for the Post Office. You had a machine, stamp, stamp, stamp — and some would be set to stamp the same number twice, and then change and the next number twice, and so forth. These were long adding machine tapes which you’d inspect later to find out whether they’d really kept the right count. That sort of thing is what I did in the summer.
This was in the summer.
This is during the time you were at Hopkins?
What year did you go to Ursinus?
Oh, that was after I got my degree.
And had done a year’s work at Hopkins in addition. I went to Ursinus in 1933. Well, I guess I’m giving it in too much detail.
Not at all, sir.
You asked something about what kind of place was it then? It was exciting. There was a professor named Herzfeld, for instance, who later went to American University at Bethesda and the highly celebrated R.W. Woods, who improved the gratings of Rowland, and so it was very natural for some of us to gravitate to spectral analysis, using the Rowland grading, as improved, and so on. Then I went in for infrared under Dr. A. H. Funt. I did not know some of these people perhaps who had made their names in the physics department in earlier times, but there were two newcomers during that time. One of them was a man who came to teach X-rays, and preside over an X-ray laboratory. His name escapes me right now, but he was well-known worker in that field, and to some degree; tried to pin down the E/M ratio and things. That was a big thing too at that time. Dirac and others were saying that from mathematics they could establish the electric charge onto mass ratio should be a particular number such as 137. And so, while I wasn’t particularly attracted to X-rays, I got involved. The other fellow came in fresh from Japan, Dr. Diecke, and a Hollander who had gone to Japan on an International Research Fellowship to spend a year or so. He came to the faculty to increase their ability to handle spectroscopic work now in band spectroscopy. While Wood’s experimental techniques, and Rowland’s spectra and all those things were great, and Funt’s infrared rock salt prisms were also great, band spectroscopy was the burgeoning field. To me, that represented the only possibility of doing a thesis, actually, because it combined both the experimental and the theoretical sides very nicely. You had to know some theory to figure out what the band spectra was all about, but you couldn’t do it unless you had the experimental photographs of that spectrum, and who’s going to get it for you? Nobody but you. So you go down to the laboratory, and you arrange an apparatus, with a vacuum system usually and so on. So I got plenty of training in glass blowing, and drawing vacuums, finding the leaks etc. I just warmed the cement a little bit in the place that looks like there may be a pinhole leak, and so forth. I did all these different things, and I was one of the first people to use the 21-foot concrete grating that Dr. Diecke had set up using a Rowland grating. There was this big metal-railed, sort of spring-metal steel bands, which went all around, all over, at least 100 degrees. I guess of the room. You could melt flexible thin glass plates that you bend a little. This was the place where I settled down to do a thesis, practically all the pictures. I wanted to be in the ultraviolet, which presented some problems. No matter what you do there is always some sort of trick, you know. You can use the ammonia to sensitize plates for the infrared. You use the Vaseline or some mineral oil to sensitive plates in the ultraviolet range and so on. You go to the dark room, and take your chances with R.W. Wood. There were several darkrooms around. He had his own, in his own laboratory. But every now and then something would disappear, something would be out of place, something that you wanted, and you reached for it, it wasn’t there, and of course, you’re developing plates that are sensitive to other things. The dark room was dark... He didn’t have a red light on, there. We had many stories last year, at the Hopkins Centennial, which were devoted to Rowland Woods, about the way in which Woods used to establish whose equipment was whose, which was sometimes a bit of a problem, you know.
I don’t know whether it’s the tape or machine that did that. (Note: The distortion is such that 25 percent or more of the information is lost — EE)
Well, to repeat this slightly — I got into the question of, does the sun affect the weather, and atmospheric electric variables, such as my father had been interested in, and so on, and everybody else seemed to think I was wasting my time, but I organized some of the students to help me, and they got paid 50 cents an hour for doing that, just as I’d been paid 50 cents an hour down at Hopkins for doing these big calculations. Well, I’d take them down to Washington, and we’d copy weather out of the M Street headquarters of the Weather Bureau. We would copy down data, and take it back to the Uranus and I’d run it through my computer, and pretty soon I’d built an analogue computer, for harmonic analysis, in order to find something about barometric pressure waves. Quite a few things of that sort that went on in my laboratory of Uranus with the help of some students to push buttons and turn knobs for me. So, I was doing two things, you might say, which were not necessarily connected with the weather, but they were inspired by it. That is, I built an analogue computer, and we used it for this harmonic analysis. Secondly, I was making all kinds of experiments to try to get a digital computer which would help in the statistical analysis. These experiments I made on digital computers were, of course, pretty well influenced by the fact that whatever I built, whatever apparatus I had, I had to pay for out of my own pocket. The laboratory budget wasn’t sufficient to supply any research for me, and so I suddenly, for instance, one day discovered that there were fuses called indicator fuses at the hardware store, and they had a little neon bulb in them, and it had some interesting characteristics which then I chose to exploit. And I built a digital device, for instance, which was a code device for encrypting a message, and then deciphering it again, and so on, and, in spite of this work with rug counters, salvaged from gas tubes and so on — in spite of all that going on, in my science college days, and in spite of the fact that there were many students who knew about this and could testify to it, this aspect was totally neglected, ignored — on purpose, I’d say — in the trials at Minneapolis where I was typified and stereotyped as an “analogue man” until I met Dr. Atunesoff. Well, anyway, I was giving a paper at the Physical Society because I dared not confront the Meteorological Society with the kinds of things I was investigating because they were very skeptical. I gave a paper at the Physical Society about using my harmonic analyzer to show some influences of the sun on weather and up from the audience at the end of the meeting, came this fellow, Dr. Atunesoff, who said that he was building computers too. And as a result of this approach of his, why, I wanted to know more about it. He said, “Well, I can’t tell you more, unless you come out to Iowa State, and I’ll show you when you get out there. But we’re afraid that IBM might get hold of my ideas and so we’re very careful not to spread them around. We’re trying to get patents, you know, and all that.” So I said, “Well, I’ll try to get out there. I will keep in touch.” And one of the big lures was that he said that for something like, I think it was a dollar a bit, he could provide storage for all the data necessary for solving a big set of simultaneous equations. Great. I didn’t get out there right away, that is, some of the testimony in the major trials will show, I finally managed to get some people to pay my gas and oil out as far as Ohio, and I and my six year old son went out to Iowa State, not to Iowa State but to Ames, Iowa and I spent a few days talking to Dr. Atunesoff, partly trying to persuade him that the vacuum tubes that his outfit (this graduate students) were building were not being used at very great speed. They depended on this mechanical compotator effect, what’s now known as the ABC machine, Atunesoff-Berry computer, but they weren’t getting much out of the electronics. It was held down to the slow switching speed of the mechanical revolution. But it was clever, that just by storing the regenerating charges on a bunch of condensers, in a drum, that he was getting low storage cost, but he wasn’t getting a high computing speed out of all this. Furthermore, as it’s been shown later, why, he really couldn’t solve the simultaneous equation which was all he was trying to do, any faster than somebody else could on a desk calculator. Well, that, you might say, is another story. The thing is that I had the opportunity suddenly — they told me that with the imminence of the war, the United States involvement in it, there was something called electronic defense preparations, and a defense course was being given at the Moore School, and physicists and mathematicians were eligible to learn something about electronics. It was by signing up, getting into that, that I first met Eckert, who was just a graduate student of no outstanding reputation or anything at that time. He was a very intelligent guy who had done a lot of things, but they weren’t in newspaper headlines, of course. He’d built a device for measuring the amount of smoke in a smokestack, for instance, a nickelometer? I think they call it, and it was a rugged, reliable electronic device. He’d also gotten a patent on how to do scanning in television. A lot of television devices in those days were mechanical, you know, rotating lens systems and so on. So he was going to scan pictures by deflecting light beams through an acoustic wave tube. All of this impressed me very much, and I think it should. He was an up and coming guy who was doing things in his prep school. At the Wharton Moore School, why, he’d done a lot more, and here he was, having gotten this graduate, his engineering degree, and was working as a graduate student. He could tell me whether any of the ideas that I was anxious to see pursued were really feasible. What he told me was, he didn’t see any reason why not. All you had to do is pay attention to the design features, as to how poor the parameters of a vacuum tube, or what no were, how much variability there was in them, and he’d design you circuits so that, in spite of the variability, why, you can get a pretty good idea of when things are on and when things are off — using an on and off, all or none, procedure to get your reliability. Now, incidentally, I had been spending time in the past few years, before that, visiting people over at Bartol Research Foundation. I must tell you that the head of the atmospheric electric research, at DTM (Department of Terrestrial Magnetism), as we called it in Washington, Carnegie Institution, had been Dr. W.F.G. Swann. He was a well-known physicist in those days, and I used to love to hear him talk to the Physical Society about Michelson-Morley experiments and other things. But very soon after my father came to DTM, he was offered, and accepted, the job of directing the Bartol Research Foundation, which he continued to do until he died. And so my first application for a job, incidentally, was to Bartol and he referred me to the Franklin Institute. Whether I’d like a Job at the Museum? But as you know, I got into science and so I was in that general area and I’d still visit the Bartol Research to see what they were doing. I was still living in the Washington, D.C. area, where I was picking up weather data, with some students with me sometimes. And I also visited in those years with Merle Tuve, who became director of the Carnegie Institute DTM. Tuve and Bright and Havsted were the people who were running a Van de Graaff generator which was built on the assumption that perhaps they’d learn more about the earth’s magnetism by learning about the magnetism of the atom, and so therefore they had to have an atom-smasher, you might say. And so, with, I think it was a half million volt, Van de Graaff, why, they got in the business and did things faster than Van de Graff did up at MIT. Well, one of the first things they did was to verify Lise Meitner’s experiments on fission. And through all of that of course, I was interested in and watching the fact that my friend Havsted, who I’d gotten his PhD about the same time I did at Hopkins, — Havsted and Tuve, who was another Hopkins PhD, and Dr. Bright, who was the theoretician, they all were operating with that van de Graaff generator, things by which they measured the results were counters — electronic counters which could count very very fast. And so it was no flight of the imagination really, for me to assume that if you can count cosmic rays that you ought to be able to count pulses, even if they were generated by something which was just a keyboard in which you created pulses to correspond to numbers. Therefore, with fast electronic counters and switches, gates as they call them now, that you could do things in the way of building an electronic counter. It was Eckert then who assured me that you could certainly do this reliably, which was a necessity, of course, for mathematical calculations and commercial accounting — both require reliability. And so, if you didn’t have the reliability, why, you weren’t going to get anywhere. But Eckert absolutely gave me the greatest confidence that that was not the problem. That all this could be done. All you had to do was get somebody to back you and get some money for it. But this was ‘41, when I was down there taking the course. With the involvement in war coming along, I was offered a position on the Moore-Wharton School faculty, essentially to replace people who were leaving, like Dr. Travis, who was a Naval Reserve man, and so he was taken out of the faculty as was another man, who was a good friend of a man who was working for Alan Hazletine (not Alan, that’s the patent attorney over here) but the Hazletine who made the neutrodyne patents, had a big thing going on up on Long Island, Great Neck or thereabouts. He was manufacturing all kinds of important electronic equipment for the military, and so, Max Maclewain was the professor who was called upon to supervise these factories and manage things for the war production. So he disappeared from the scene. These were severe losses to the Moore-Wharton School. Maclewain was the guy who’d taught the electronics course that I’d once taken there in the evening, just to be sure that I knew something about electronics, while I was heading toward faster computers. So I was offered this position, to come on and be, at least for these people, during the war, a temporary replacement, you might say, for these people to help teach the students who were still coming through. They had contracts and arrangements for Navy students who had to be taught engineering. They had Army student training programs. So they were not dependent on just whatever civilian students might be around, but had these extra programs for military instruction. I got in on that, and resigned from Uranus that late summer and I became one of the staff of instructors. It wasn’t any great position. I wasn’t the head of any department any more. But I was assured that they would pay me just as much as I’d been getting in my old job at Uranus. At the same time, they had another PhD in that course, who accepted a job just as I did, at the end of the summer. That was Arthur Burks, who was a specialist in symbolic logic from the University of Michigan, philosophy department. So Burke and I roomed together for the first year, I guess because my family was still in Collegeville, and getting housing set up. Taking a job in another city wasn’t easy. I would commute to my family, but stay in Philadelphia. Burks, who was single then — and I worked not only on teaching students, but on some war projects which were called defense projects the first year, and we worked on whether we could put a sufficiently large number of turns into a big copper coil carried by an airplane, so that if it swept low over a body of water, it could trigger off magnetic mines. Well, we did a lot of calculations but I don’t think anybody ever swept magnetic mines that way. We had other kinds of mine sweepers. And the most irritating project, from one point of view, was one that the Moore School brought in from the Signal Corps up in New Jersey, who wanted to have the patterns of radiation from a dish, as they called it, a parabolic, reflector which was put around a dipole transmitter, to push some rather short wave radiation. It might hit a plane, be reflected and picked up again, and form a radar detection system. Try to identify the aircraft that might be intruding with that system, why, they wanted of course to improve the sharpness of resolution, by getting a nice sharp point to the beam and its response back again. They had to eliminate what they called the “side load” if that was possible and not have any waves go back of all this, because then they’d reflect from something else, and that would also mean that you’d be getting reflections picked up, and all that. Well, that was a vexing contract, because the people at the Moore School really had very little idea as to how the project should be done. They just conceived that maybe numerical integration could do this, and so, they had one small Fliden desk computer with a plus-minus bar. It didn’t have even the multiple buttons of a Marshant and so on. I’ve forgotten whether it was capable of ten digits or just eight digits. They made both models in those days you know, with the bank of keys. One of those, they had purchased in order to try to check up on the accuracy of the differential analyzer, sometimes on a trial run, to see whether it would come out what they thought it should. That was all that was available in the way of digital computing at the Moore School. So they took the contract, and they told me, “Here’s what you want to do — what you’re going to do for the next six months or so —" I said, “Gee, six, months?” I said, “Well, it might take longer. I don It know how I’m going to do it yet.” “Well, use a book.” Here’s Stratton or somebody from MIT that has written this book on electromagnetic theory — it had all kinds of vector notations, with deltas and bits, counter-integrations and so on. There was nothing in that book which referred to numerical integration, no digital methods and so forth, you know. It was all just pure electromagnetic theory a la Maxwell. “Well, this is the way you do it. And we got a physicist from across the way, the physics building next, which can help you.” Unfortunately, I forgot his name, no point in mentioning it now, he was a senile gentleman who had long since, I think, been retired even from teaching first year physics, and I spent six weeks or more trying to explain to him what a vector was, and how it was used in this book on electromagnetic theory, and what we were going to do, after which I asked to be relieved of that burden, and I started on by myself. As for means of computation, why, we found that the Moore School had a statistical laboratory which included perhaps ten or twelve desk calculators of one sort or another, and during certain periods when their classes weren’t using these, we could, borrow them. So we hired a dozen or so students, mostly electrical engineering students, to come in odd hours when they were available during the day, and use these machines, and I drew up the forms and we started out. It was a miraculous thing, that while we were hard at work on this, incidentally, there were other people, including Arthur Burks, working on the roof, with the actual physical radiators and aluminum dishes with appropriate cutouts and things, measuring to see whether they could check our calculations, and our calculations checked then, on what we were trying to do. And when we finally got the first thing computed, and I compared it with some curves in a book that I had seen called MATHEMATICAL TABLES, why, darned if this curve didn’t look a lot like a first order Bessel function. And later, of course, we found out there was good reason for this. With the simple dish, without the cutouts and so forth, the first order Bessel function was the right answer. Which we hadn’t known in advance. Well, that gave us the confidence, that this was really working, that I’d split the areas up into the proper small elements, and summed them properly, and gotten the students all to check things so that there weren’t any gross mistakes, and so on. So we were all set to go. Now all you had to do was cut out these things. I was subsequently advised by a Bell Telephone man who worked with Signal Corps that none of our variations that we’d studied on that contract were of any use, eventually, but that some other ways of phasing antennas and what not were more important But be that as it may, the thing is that this contract, which was supposed to be done in a few months, lasted, I think, a year or so, and it was my contention that if we’d had an electronic calculator, of the kind that I was proposing, that this would have been just a matter of some programming, of course, but maybe a day or two’s calculations. Meanwhile, there was no intention of anybody, apparently, to open their mouths about such a proposal as an electronic computer, to get government funding. I was advised by Dr. Chambers that if I wrote up the proposal, instead of just talking about it, I’d do better. But it wasn’t until 1943, that the Ordnance people sent Dr. Goldstine up to see what could be done about expediting calculations. By that time, they’d hired a hundred or more girls, acquired desk computers for them to work with and so on, to supplement what they could do with a couple of shifts on the differential analyzer, and they still weren’t producing the firing tables to make the guns useful. And so he was supposed to collect ideas, see what we can do to expedite. My former science buddy, student, whom I’d gotten to come to the Moore School to service the differential analyzer, was telling anybody who came from Aberdeen, “Go upstairs and talk to that professor there.”
Who was this?
Chapline was the man, the student. He’s still alive, and probably he’s working on technical writing nowadays. But his part in the trial out there was rather brief, and nothing was asked him, for instance, about what he knew about my digital experiments at Uranus, although he did know why. That wasn’t the subject of the trial, you know. The subject of the trial was to prove that I copied down NASA stuff. Apparently that subject didn’t get into the trial. At any rate, Chapline, who had as his hobby repairing cuckoo clocks and tuning organ and things of that sort, I thought would be a natural for keeping this mechanical monster, this differential analyzer, tuned up, and stringing the bands for torque applications and so on and it was quite a thing to do. So he worked with the girls, like my wife, down in the sometimes an analyzer room, keeping that thing going with sometimes an accuracy close to one part in maybe 10,000, or one part in 1000, somewhere in there, and here we were, upstairs, doing other things, like I’ve described, the Signal Corps project and things. If we’d only had an electronic computer! But along came this fellow Goldstine, a first lieutenant I believe then, anyway a lieutenant, assigned to see what he could do about expediting. And Chapline said, “Go up and talk to Mauchly.” After we’d talked a little bit, he said, “You ought to write this up. I said, “I have. “Well, where is it? “Well, I don’t know. I’ll look. “I looked. I’d given away all the copies that I’d made, to all the various members of the faculty but nobody had any. Nobody could locate what I wrote in August, ‘42, and here it was like April, March or April of ‘43, and so I still had the same secretary, who made very good Gregg shorthand notes, and she reconstructed the whole thing from the Gregg notes that she had, and with very little change or corrections, why, I had the whole memo all over again. Then he said, ‘Well, we’ll need more than this. We’ll have to have dollars and better estimates and so forth.” Within a very short time, a few weeks, I think, people were going down to the Pentagon with Goldstine, and getting high interest about this. And then on April 9th, which is Eckert’s birthday, that was the day we went down to Aberdeen, and presented this to the man who was in charge of the Research laboratory, Colonel, I’ve forgotten his name.
Simon, yes. Colonel Simon, who on retirement went out to Carborendun, CO, I think.
You indicated in the trial that Deiderich was presented with your proposal in ‘42, and did not think —
— that’s true —
— that the war would last long enough to justify it?
Yes. My first real conversation with somebody sent up from downstairs was with Dr. Deiderich, who was in charge of all the IBM calculation for the Ballistics Research Laboratory. Most of those calculations where done by punch card, of course. His general comment that it would probably take at least a year to build all this. Here it is 1942, in a year the war will be over. I guess this isn’t really something for us to push on, you know.
Did he change his mind by the spring of ‘43?
We never confronted him with this and asked him how he felt about it. But he was a remarkable man. He was getting on in years then, but his father was still living. He went out to California to live with his father after that, I think. Maybe his grandfather, I’m not sure. Lots of longevity in that family. Well, he was a great guy, but he was just a little more sanguine as to when the war would be over. Some other people, later, of course had to look at it differently. The situation that Goldstine was confronting when he was sent up was that there were guns delivered to the fields of battle without adequate firing tables. I’ve understood that one difficulty was that somewhere in there, the theater of war changed to the deserts in Africa with Rommel, the Desert Fox, or something. Since the ground was more resilient in Africa, and allowances they’d made then for the recoil effect were inadequate. If you’re shooting up here, and the gun recoils, why, it not only lessens the velocity of the shell, but changes the tilt of the thing, and that too is going to be bad on your marksmanship. Also, the air over equatorial deserts has different temperature and density relationships than what they normally were planning for in their firing tables. So they had to have extended corrections, I guess, for that, and so they badly needed firing tables, not only because the guns were new, but because they hadn’t covered all the range of parameters which big artillery needed. We didn’t finish the computer in time to provide these useful firing tables but we finished it in time to show that this thing was very useful for all kinds of purposes, and that we had successfully complied with the wishes of the Aberdeen people, to make this a general purpose computer, and not just something which could do firing tables. So that after the war, they could feel that their money had been spent for a useful purpose. As a matter of fact, it was used for meteorological calculations later, when von Neumann became as interested in meteorology. I think I was the first one that mentioned to him that I’d been down to the Weather Bureau and talked about the applications of computers to that. They became interested in this to the point where they found out about L.F. Richardson original work began during World War I and right after, where he tried to integrate the equations of motion of the atmosphere. It took him, say, something like a year to do a 24 hour forecast and it wasn’t at all accurate anyway. And so it was obvious that we needed faster computers, and also obvious that you needed better mathematical methods. I was providing the faster computers, which was fine. The better mathematical methods were the kind of thing that was in von Neumann’s department. He was advisor to the underwater sound people, and underwater explosions and so forth, for the Navy, as well as the Los Alamos people who had propagation of waves, on the impact of setting off a bomb, and so on.
Before von Neumann got involved, who was responsible for the mathematical methods?
Well, the mathematical methods were, you see —
— on the machine itself, who decided what on the ENIAC, I’m talking about —
Well, perhaps you’re thinking about Pademacher.
Even before him. For example.
Well, who could be before him? He was the one that was given the project, at the time we started on the ENIAC. Nobody else could have.
I thought that you had something to do with writing the original progress report that related to numerical analysis.
Well, to numerical analysis, but not particularly in the rounding off error as a separate study or anything. Going back to that, when they were trying to beef up, fortify, or justify this proposal for spending money on an electronic computer, one of the things I thought was very necessary to do, I think everybody agreed, was to show how a step by step digital method could be employed on such a machine, to achieve what they wanted in the way of integrating a differential equation. Now, of course, they’d already been integrating equations with desk calculators, which are digital, and having girls write down numbers and put them back in and so on, do a lot of good different things which are explained as Adams’ method or somebody else’s method and so forth that appeared in Scarborough’s and other books on numerical mathematics. But we found it necessary to show how, if you used just a simple method, that it wouldn’t take any very complicated device to do this. You wouldn’t have to slow the machine down by continually having human inspection. It’s doing the right thing, you know — and, now I think we ought to do this or that. So, I immediately got into a method which was written up in various books on differential equations, including Ince the author that liked best at that time. He had an explanation, an appendix or something there, on the Runge-Kutta method, and so it was kind of a fourth order thing. If you did all the things that were required in the Runge-Kutta method, why, once you got started, it was like a fourth order interpolation or fourth order extrapolation and so forth. This would fit what would be the true solution pretty well. However, these books really didn’t say much about what errors might occur under such conditions or how much you are off the true curve. There was however an idea that seemed reasonable to me and other people has thought so later. I don’t think it’s ever been distinguished by anybody’s name, that was that if you did — if you did the same integration method with intervals which were, say, half as big as the ones you used first, and then say half of that again, and so forth, you got a basis for extrapolating to what might be the ultimate effect if you could do it with the infinitesimal calculus, let’s say, with an integral approaching zero as closely as you like... If you complete it fast, you see, this becomes more feasible. So, this kind of extrapolation was later on written up and studied and so forth. That, of course, has to do with a kind of error which has nothing to do with round-off error, and so you have to worry about that too. That is, how many digits are you going to carry your numbers to, as you keep on your step by step course? Should it go to six digits, eight digits, ten digits, 12 digits? Or do you need 20? You have no guidance on that in the beginning. We were aware somewhat later in the business, when we saw Howard Aikens,” machine, that he had provided for 20 decimal digit accuracy, because he thought that would be required somehow in some of his calculations. The “how many” digits problem was something that everybody, of course, who knew anything about what we were trying to do would be concerned with, but one of the things which we brought to bear on this was the fact that the astronomers were all concerned with this sort of thing. Most of the equations and calculations they do are essentially step by step differential equations things. So Leland Cunningham — an astronomer from the West Coast — was one of the several astronomers who were brought in to Aberdeen, to help them in their war effort and calculations at the Ballistics Research Laboratory. He came to Philadelphia, and we went down there a few times, in the very beginning of the project, to confer on just what we should do with respect to various specifications. One of our concerns was how big should we make the digit capacity of one register accumulator? And so it was through discussions with him that we came to the idea that ten was probably a sufficiently good number. Nobody could guarantee that. But at the same time, as insurance if you like, Dr. Brainard, who was the project administrator of the thing, decided that here was a spot where they could protect themselves and get better information by assigning somebody in the math department the job of investigating this. They ought to know better than anybody how to investigate it. So, I mentioned the man’s name a minute ago, he was a theory of numbers man. Dr. Rademacher was given the project, to try to investigate rounding off errors in the solution of differential equations. And I don’t know who proposed it, but maybe Dr. Goldstine, but someone brought up the point that there are things called adjoint equations which they used at Aberdeen for studying what happens when various parameters change, in firing trajectories and so on. In previous years, incidentally, the people at Aberdeen relied for most of their theoretical information on F.R. Moulton who was a professor at the University of Chicago. He wrote a book on the subject of these firing tables as a result of his World War I experience, I guess, helping Aberdeen. So it wasn’t a coincidence that Goldstine was at Aberdeen, because he was a student of F.R. Moulton, and was interested in this sort of thing, the firing tables and so on.
I thought Goldstine was a student of Bliss’s?
Yes. I may be wrong that he was a student of Moulton then. He was a student of G.A. Bliss at Chicago but I think Moulton was at Chicago too, and he was the one who put in hardback form a rather thin little book, a monograph on how to compute these things, ORDNANCE CALCULATIONS, or something. So, at any rate, I guess Goldstine may have sat in on the same conference I’m talking about at one time or another. But it was Cunningham that I remember particularly that made a special effort to come. He was an astronomer who had had much experience in computing, for getting orbit parameters and things, and he was mostly interested in comets, I believe. But they’re sort of hard to do sometimes too, with the inadequacy of the data and so on. So, while we went ahead then saying, “Well, we’re going to build a ten decimal digit thing.” We would make it possible to hook two of these in pairs together, with a carry-over act on, and a cable that hooked them and so on. In this way you would have in effect, if you wished, 20 digit accumulators, without having to do any additional programming, as we might call it today, you know, to get double precision. The machines would automatically work at 20 digit accuracy, with half as many accumulators, of course, effective to the same amount of physical equipment there. So that would, you might say, provide for the kind of thing that Aiken was trying to provide for, with his large accumulator capacity, the Mark I, which we didn’t then know about. But as I say, Raddemacher was given this assignment, and a couple of months late it seems to me now, why, he was asked, “What have you done? What are the results? We’re about ready. We’d tested out our decade units, and we’d gotten, reliable ways of doing this. So it’s about the last moment at which we can specify whether it’s going to be ten digits or 12 digits or that we were going to put on the accumulator. So Cunningham came up, I think, and we had a conference at the Moore School. Raddemacher presented the same things which he thought were pretty indicative of how he was going about it and what was coming out of it, and it turned out that it wasn’t truly appropriate at because he had misunderstood. He’d never done any mechanical differential equations integration himself apparently; he was a theory of numbers man. He could answer lots of questions which might relate to numbers which were 2 to the 200th power or something, you know, plus one or two and whether it was prime or not and so on. But when it came to this raw business of round-off errors, why, this was an entirely new field to him. And so he had somehow, I think, assumed that the accumulators could just sort of expand to any capacity, and so on, you know. I can’t detail right now just what the difficulties were, but he had made inappropriate assumptions, and his results were not at that time useful. Then he went back to work, and I think later he did get something which was much more useful. And in the long run, why, Harry Huskey, whom we hired on the project later, was the man who tried to verify some of the conclusions of Raddemacher, in seeing whether the round-off error really behaved the way he said it would, in particular chosen examples. The easiest example to choose, I think Harry Huskey took, was the integration of sine and co-sine functions, because you know exactly what the correct solution is, and so you could study how it gets off from that, as time goes on.
Who decided on using the Hend method of round-off?
Well, I mentioned before that it seemed to me that when you have a fast machine, that simplicity of programming had some advantage, and that you could make your intervals smaller, and your methods less complex than the Punge-Kutta for instance, and perhaps achieve the same results or better, depending on how you arranged things. So the first thing I did was to say,” Well, what’s the simplest way in which we could measure the slope of something; so I said, “Why don’t we do it this way? This seemed to be a nice simplification, and I thought very easily it was simple enough that it could be iterated very easily many times. So it was one of those cases where perhaps it could have become the Mauchly Method, except that we had another resource available to us. Dr. Brainard decided that it would be nice to have some library research done on methods of integration. There was a student there at the time named Tom Liverman, who was available to Dr. Brainard to do this. And somebody decided that Tom Liverman should go to the Math Library, and investigate different methods of integration. He felt we shouldn’t just rely on whatever methods the people at Aberdeen had been using, because a lot of those did depend on humans making estimates and working it out on paper. We wanted something which could be as automatic as possible. But darned if Liverman didn’t come back in a few days and say, “Well, here is a method that was proposed by a guy named Heun back there in Germany.” He had the quotation and so forth. So, OK, he thought of it then, and nobody ever used it actually. So the simplification that I thought was a rather natural thing to do have this antecedent.
That’s very interesting. I didn’t know that.
So I can say that I re-invented it, if you like. But the credit goes to the guy who first published it, and that’s what happened. And the reason it goes to him actually is because I think that Dr. Brainerd wanted to have all the insurance he could. This was at a stage in the proceedings where nobody had any strong confidence, except Eckert and me that this thing was going to work. Months later, when the two accumulator tests had been passed I’d checked the crosstalk and made the two accumulators which were built integrate second order equations, such as sines and co-sines, hyperbolic sines and co-sines and exponentials and what not, the most we could do, with no further controls, the two feeding back and forth. At that time, I think there were some great sighs of relief, that at last there was an indication that this thing really might work, you know! Before that, it had been a big question mark in most people’s eyes. And so, because we were doing this in a digital fashion this way, we believed that once we had checked out two accumulators that could do this, that ten or 20 or 30 or 40 accumulators could do it, because they’d have the same interactions in this digital system that you have in analogue systems where everything is connected to everything else. Continuous variables can just roam around, you might say, and go wild. This was the situation, as we saw it that would occur at MIT, where they were trying to do a project which later became known as Whirlwind, when it became digital. I don’t know what it was called at that time, but they were going to do a lot of simulations by analogue computer. By the time we heard about it, van Neumann and Goldstine were working closely on this project of ours, closely in the sense that they knew what was going on, and Eckert and I of course somehow got invited to go along, that is to go up to MIT and talk to some people there. We all took this view that unless they go to digital methods; they’re never going to get anywhere.
Well, how did the people at the lab feel about working with Eckert?
Which? Do you mean the Ballistics Research Lab?
I meant specifically at the Moore School working on the ENIAC.
As far as I know, everybody thought that was great. There may have been somebody who felt that he was too young and inexperienced. There may have been, but I don’t know. I don’t even remember that anybody really felt that way. I know that some people had more respect for his achievements than others. I think that Dr. Chambers, for instance, who was sort of Eckert’s confidante and advisor on many things, as he was to me, Dr. Chambers had full confidence in the fact that Eckert and I were going to come through all right. Some other people may have had a little less confidence. But there was never any animosity or anything of that sort.
It was a good working relationship.
It was a good working relationship all the way through, as far as I know. And the amazing thing to me, which I ought to have asked Eckert about some time — he probably is the only one who would know nowadays — where did some of these people come from, that contributed to this big team? In the Minneapolis trial, of course, why, Brainard testified on Honeywell’s side, that this was all just a big team effort. He felt all the people on the team should have been equally credited on whatever came of it, patents or what not. And as you know, why, we’d tried to satisfy the requirements of who did what and what was novel by writing letters and getting responses and so on. But the big question I was talking about was, where did all these people come from? Well, some of them were graduate students at the Moore School at that time, I think. Kite Sharpless who died some years ago now, was at the Moore School because he wanted to do some graduate work in electrical engineering. He had been a teacher in a prep school up in New England, and I think he wanted to get further education, to measure up his salary. So Kite Sharpless was a very interesting participant in all this. Some of these people were, I think, known to the Moore School people like Brainard and others because they’d taken graduate evening courses. They might have been working at RCA, for example, and they’d come over there and take evening courses to get extra credit. Joe Cheddicker for instance, who I think worked at RCA in designing test instruments, seeing that the test instruments were properly calibrated. He had helped us in the very beginning, for instance, testing out some of the various designs for decade units, which was the very first thing we had. There were a big number of modules, you know. They had to have 200 of those plug-in units, ten for each of the 20 accumulators, and then some more things which were essentially decimal ring counters. They were a little smaller; we called them repeat units, which counted the number of times that a particular operation is done. If you wanted to add three times, you could set this for three, and that decreased the amount of program wires that had to be taken around to the various places. So there were people, then, who had been and were probably employed by RCA. They had been taking graduate work of one sort or another. Some of them were just people that had graduated from the Moore School and somebody knew where they were and just said, “Come on back, we need you. There’s important work to do”. Perhaps somebody had considered the fact that where they were then working, they couldn’t get a deferment; with us, perhaps they could get a deferment. I don’t know. That was always one of those little things that went into job looking and job changing, in a war time. So I don’t really know. I’d like to know more about where they all came from. Bob Shaw, a very good guy on logic as well as circuits and so on, dealt with a man named Gail, Harry Gail. I don’t know just exactly why he knew him, what it was all about, but Gail became his sort of assistant all along. You know, Shaw and Gail worked together on lots of things, designing function tables and other things that we were doing.
Thank you very much, Dr. Mauchly.