Welcome to the Center for Oak Ridge Oral History
|
small (250x250 max)
medium (500x500 max)
Large
Extra Large
large ( > 500x500)
Full Resolution
|
|
PIONEERS IN SCIENCE AND TECHNOLOGY SERIES ORAL HISTORY OF DR. JOHN V. ATANASOFF Interviewed by Clarence Larson Filmed by Jane Larson March 13, 1984 Transcribed by Jordan Reed DR. ATANASOFF: …computing. I had an interest in calculation from my earliest days. MR. LARSON: Yes, incidentally, where were you born, Doctor? DR. ATANASOFF: I was born in Hamilton, New York. MR. LARSON: Hamilton, New York. DR. ATANASOFF: My father was born in Bulgaria, came to this country at the age of 13 and was left alone here at the age of 15 and managed himself somehow through Colgate University where he met my mother, and I was born in the hill behind the University at Hamilton. Have you ever been to Hamilton? MR. LARSON: No, I never have. I have been in that general area. DR. ATANASOFF: Yes. I, well I’ll start this way. I was five years old when I started school and I would soon be six on October 4th. After about a month, my mother said, “Let me hear you read.” My mother heard me read and she says, the following, “I see I will have to teach you myself.” Those were her exact words. And she did for about a month. She switched me from a Lexile reader to a phonics reader and then told me that you should always use phonic reading. But it’s the best we have, but it’s very poor in English and when you get to spell you will have to memorize every word. MR. LARSON: That’s a very interesting point there. DR. ATANASOFF: She’s a sharp woman and only a month, after a month I took off on my own, following her advice and at the end of the year I could read almost everything. MR. LARSON: Well that’s a remarkable story there. DR. ATANASOFF: It’s a remarkable story because I think if everybody could learn to read equally, rapidly, that we would have a lot more intellectuals in the world. MR. LARSON: Of course reading is the foundation for almost every career. DR. ATANASOFF: Almost everything and I’ll pass in a hurry until, I won’t take up every step of my life until I was nine years old and my father had just got a job as an electrical engineer in a phosphate mining company which was just opening in Polk County, Florida. MR. LARSON: Oh yes. That was very early in the days of phosphate mining. DR. ATANASOFF: It really was. There were a few mines around, but not very many. This mine was just being started. There were no houses there, but they built a few houses and we took one of the houses and when I tell you, for 1913, it just comes back to my memory of that house and whatever I did I could see the surround of that house and I could tell it was during 1913 when I was nine and at the end of the year 10 years old. My dad failed being an electrical engineer because he didn’t know the slide rule. He really didn’t need one because the principal job he had was organizing the repair of the motors which were continually being burned out by the lightning. Lightning in Florida is about 100 times greater than it is in many other places. MR. LARSON: Oh yes. I have noticed that in my trips to Florida. DR. ATANASOFF: I don’t know, I think they have done some things which mitigate the effects now, but at that time it was terrible and Dad just kept his men hard at work rebuilding those motors. That’s what he was doing. So the slide rule was left to me. It was a Deci-Lon slide rule. MR. LARSON: Oh yes, a time honored name. DR. ATANASOFF: Yes it really was. It had a nice booklet that came with it and I could read. At that time I had read everything in sight. I had read, I remember that I read [Edmund] Spenser’s The Faerie Queene at that very early age. MR. LARSON: Yes, well, that’s very advanced. DR. ATANASOFF: I could read it and in a week or two I could do any ordinary problem on the slide rule. But this didn’t satisfy me. I had to understand how that slide rule worked and of course logarithms commenced to play a role. Now here I almost stopped baseball and I spent my whole time, we hadn’t commenced school yet because they didn’t have a school there at the mine. I studied, read, got Mother to help me at the age of nine. My mother was first class in algebra and read on and on and on until I understood logarithms, the meaning of logarithms. I memorized the definition for a logarithm. Then I learned how to use tables of binoms [binomials] by myself during that period and I thought I was content with life for a moment and then I realized that I really didn’t know how to calculate logarithms and I started on the project of calculating logarithms. I didn’t know who to go to and strangely enough Dad didn’t have much of a role in these events. MR. LARSON: Yes. He had other things. DR. ATANASOFF: I got a hold of a book and read about logarithms and I have the book here with me. MR. LARSON: That’s remarkable. DR. ATANASOFF: The very book. The very book. MR. LARSON: What is the title of the book? DR. ATANASOFF: The title of the book is College Algebra. MR. LARSON: Oh yes. DR. ATANASOFF: College Algebra has meant a lot of different things in different places. It’s quite an elaborate book and analysis. And gave the very subjects of algebra. There is a chapter in here on logarithms, then there is a chapter, a further beginning on the calculus in this book. I was, I read the elementary definition of logarithm and I was reaching into calculus and memorizing derivatives. MR. LARSON: That’s remarkable and this was all in a book on algebra. DR. ATANASOFF: I did all this in my ninth and tenth year. The definition of derivative is here and it’s inscribed on my mind. MR. LARSON: Yes. I believe you mentioned, do you have the name of the author of that book by chance. DR. ATANASOFF: The name of the author was Taylor. MR. LARSON: Taylor, yes. DR. ATANASOFF: Taylor was a professor of mathematics at Colgate University and this is a book that my father had used. This is his book. MR. LARSON: That’s remarkable. DR. ATANASOFF: I learned how to do infinite series and to test infinite series for termination and I actually calculated the logarithm of five to the base 10. MR. LARSON: Well that’s remarkable. DR. ATANASOFF: And that’s the time I left the book. MR. LARSON: Fine. That is really a wonderful story there. DR. ATANASOFF: And you know I had only had elementary mathematics up to that time. But my mother helped me with the algebra parts. There wasn’t much to learn, it didn’t seem to me. I had that head start of my mathematics and it’s followed all my years. MR. LARSON: Yes, well with regard to your work on calculating logarithms and so on, you certainly got into the theory of numbers which is so important in your chosen field. DR. ATANASOFF: I thought that if I read anything I could learn it and this has proven to be useful throughout my life. MR. LARSON: Fine. Very good. DR. ATANASOFF: I did a lot of other things. I studied physics and chemistry during this period and one other thing that I should perhaps mention is that Mother had a book. It had something to do with computing, you’ll see. Mother had a book on arithmetic and it had a chapter in it and it said, “Numbers to other bases than 10”. MR. LARSON: Oh yes. DR. ATANASOFF: That was the title of the chapter and I studied that during this year and when I, ever since when I came in contact with the need for numbers with bases other than 10, my memory from that furnished stuff. It was quite a complete treatment of that subject. I haven’t had to go to any book since then in regards to numbers and as well say later when I commence to wondering now will a, should a computer be built to a base of 10 or should it be built to some other base and I think I was the first one to go into that subject, thoroughly. I decided to use a base two. This is where my number theory came from. MR. LARSON: Yes. Well that was very fortunate that you had that base of number theory because without it also then the matter of applying electronics to the base two is certainly a lot better than the base three, or ten or anything else. DR. ATANASOFF: Sure. I, later on when I decided to investigate what numbers, I mean, what base should be used in, for computing. I went into the subject fairly, fairly overtly. You’ll find it in my bulletin there, my manuscript, without any question base two was overwhelmingly more advantageous than any other base. Now another thing that happened way back in 1913, the company had a Monroe in it’s office. MR. LARSON: Oh yes. At that early date? DR. ATANASOFF: Oh yes. I wondered if that could be so. My memory told me so, and it told me exactly how it was. My memory of the machine is in detail. I knew it was a Monroe, but when I got to writing about it, I didn’t want to make a mistake and I got a hold of someone that was connected to the Monroe company and he said they first commenced making Monroes in 1910. MR. LARSON: Well, that’s a very interesting point. That dates the availability that apparently Monroes and Marchants were used tremendously as the only asset that… DR. ATANASOFF: Those machines copied European machines, a similar kind. I didn’t have much contact with European machines, it was mostly Monroes and Marchants mainly that I used and the Monroe served us for many years in calculating and it was a very good machine and now, you could unhook the carriage, unhook it, you pull a couple of levers here and you could slip the carriage back and you could look inside and see how the mechanism worked and I did that almost immediately and I knew exactly how it worked and what was done. I was learning about computing machines clear back to that day. Now I’m not going to be so detailed in my life, I had, however, when I was in the tenth grade I decided, I remember again, I could tell you the shape of the room where I first got this idea. It was a room in the high school where I was attending high school and the, I had plenty of time, I didn’t have to work very hard. I had plenty of time and I was wondering what I would like to do. I decided right then that I would like to be a mathematical, theoretical physicist and that’s what I came to be. MR. LARSON: I think that explains your choice. DR. ATANASOFF: When, I’ll rush ahead to college. My father continued, he didn’t think that electrical engineering was right for me. He thought I ought to have chemistry. I like chemistry very much. I read a book on the subject at that time, a university textbook on the subject and knew it pretty well, but when I got to the University of Florida where I talked to the Dean of Engineering and he said the electrical engineering course is the most theoretical course on the campus. I decided I would take electrical engineering and did so. I had a lot of things that did me a lot of good. The machine work and stuff of that kind, the formulas and things with computing, but I didn’t have any electronics because electronics had started to be used in the universities of the United States, but only MIT and Cal Tech had courses at that time. I didn’t have a bit of work. I knew of vacuum tubes and I knew roughly how they worked at that time from my general reading, but I didn’t have any electronics. I sent out, as I neared the end of my year, 1925 when I graduated from the University of Florida in Electrical Engineering, I sent out some brochures and I got one from Iowa State College in Ames, Iowa. He just wrote faster than the other people. I later got one from Harvard, but I took, I promised the man at the University of Florida that I would take his scholarship and I did so and I went to the University of Florida and went to the Iowa State College in the fall of 1925. During my first year at Iowa State College, we were studying the theory of real variables, I was going into the mathematics department in order to have a theoretical background for my subject. I was studying the subject of real variables and there was a proof in there that had to do with the base two and I was the only man in the class that had ever had experience with base two. I remember that. Everybody else had to worry about that subject, and I knew all about it. MR. LARSON: That’s a fascinating story. Not to the base e, but to the base two. DR. ATANASOFF: Base two, yes. You can’t use e as a base very well because it has to be a whole number. It has to be a whole number as a base for a number system. It has to be a whole number and equal to two or greater, but anyway a whole number will do as a base. Well I can easily work out the proof. I majored in mathematics there and minored in physics and in 1929, I, in the middle of the spring semester, it was the end of a second quarter at Iowa State College. They had a quarter system there then. That was in the middle of the spring semester I went to the University of Wisconsin and took up residence there and started taking three courses. It kind of worried the professors that I was coming in in the middle of the second semester and starting to take three courses. Well, professor March knew I had an aptitude of various things and already knew I had some work in elasticity which he was teaching. One of the other professors was teaching quantum mechanics and I had had, I had just read something about quantum mechanics, but I had had it. Didn’t think I would make it, couldn’t think I would make it. This was Dr. J.H. Van Vleck, and I would ask a question and he would say, “We finished that last semester.” So he would push me down that way. MR. LARSON: Oh yes. Now quantum mechanics was still more or less in its infancy there. But really the big development was really about that time. DR. ATANASOFF: That’s exactly right. We had it direct there during the year. People come around them and say that we think that the professor ought to answer you because we don’t know the answer to that either, all the members of the class. He had 25 members in the class and finally we came up to the final examination and all but five of the members quit. They just resigned from the course. I was left with the other three or four, four or five, which ever it was, and wrote the examination. Professor Van Vleck doesn’t know what to say. He said, “Atanasoff,” he says, “you did pretty well on the examination,” he says, “I suppose I ought to tell you, you did the best than anyone on the exam.” MR. LARSON: Well that’s remarkable. DR. ATANASOFF: Afterwards he became a major professor. MR. LARSON: Oh yes. DR. ATANASOFF: However, I must tell a little about my thesis. The subject was the polarizability of helium where to take the helium atom and apply an electric field and see what electrical moment it had. MR. LARSON: Oh yes. DR. ATANASOFF: I understood the subject all right, but you see this wasn’t a simple atom. It had two electrons and it wasn’t easy to calculate mathematically. We used the wave functions that had been described for helium by Hilderhoff, a German physicist. I was able to convert these so they could be used for computing the polarizability of helium and Van Vleck was in Europe during the next year, but I went ahead with a professor who was from Zurich whose name escapes me for the moment. He was just as much help as Van Vleck would have been and I finished my thesis and got a Ph.D. in the middle of the summer of 1930, got my Ph.D. Afterwards I had left memories at Iowa State College, and I went back there to teach. MR. LARSON: What department? DR. ATANASOFF: I went back in the mathematics department, but after a few months they made me assistant professor of mathematics and physics, in both departments and then in a year or two they made me associate professor of mathematics and physics and I worked in both departments. Of course the mathematics department was an applied department being in a school at that time so that it was a pleasant period of my life. MR. LARSON: Well very good. So you went right into the academic profession. DR. ATANASOFF: I did that yes. I was there only a year or so and I commenced giving, teaching graduate courses in both mathematics and physics, and pretty soon I had post-graduate students who wanted to work with me. It was a wonderful period of life to be in, with just having had a Ph.D. and having students come around and asked. People liked my courses and wanted to work with me. I had as many students as I could handle. During the next six years or so I had eight post-graduate students and all but one got the Ph.D. MR. LARSON: Well there weren’t very many students or professors of theoretical physics at that time. DR. ATANASOFF: No, there weren’t. Now I was thinking about what, one thing I forgot to talk about, you know when I got out of school in 1909, in the summer of 1910, I had a very good teacher and I want to name her because she was extraordinary. Her name was Gertrude Arthur and she drove everybody to the absolute limit, including me. MR. LARSON: Oh yes. That’s the role of a good teacher. DR. ATANASOFF: That’s the role of a good teacher. When I worked on the polarizability of helium of course we had to use approximations on this subject and you couldn’t get the exact answer. You had to use approximations and my first ever; I got an accuracy of five percent, starting with an atom about which we knew very little at that time. I was able to nail it down and get an accuracy of five percent. Later I made a recalculation and got an accuracy of two percent. MR. LARSON: Well considering that you were delving into an absolutely new field there, that was very fine accuracy. DR. ATANASOFF: I was pretty enthusiastic about it. You know, about this time when I got back to Iowa State College I knew what I needed to know, that I hadn’t had. I hadn’t had electronics and I commenced myself into reading electronics. I read about the first book ever published on vacuum tubes by a man in Africa, in Africa. His name was Vanderweil. MR. LARSON: Oh yes. DR. ATANASOFF: Of course he was, he came from Holland. His family came from Holland. MR. LARSON: Oh yes. DR. ATANASOFF: And he wrote a book on which you will find in old libraries today, Vanderweil and I read a number of other books and I also started to work it with my fingers. You would think that I mostly sat and thought, but I had started a lab with my father and he always worked me into doing things with my own hands and I’ve always found it easy. As I stayed around Iowa State College, why, pretty quick I was directing experimental theses. It didn’t mean a thing to me. I could move into theoretical, into experimental work just as quickly as I could theoretical work. I had, I was going to tell you something about my theses were about, because that’s what I had to do. I got it here somewhere. MR. LARSON: Well, you’ve given us a good description of the work. I think that’s perfectly understandable. DR. ATANASOFF: I had one man who was doing graduate work in quantum mechanics. He was working on lithium. Other people were working on the theory of approximating the solutions of partial differential equations because you understand that you had to deal with the partial differential equation in theoretical physics and that’s what I had done with my own thesis and I have given it a lot of attention as to how to proceed with that. I commenced to study new methods, new methods for solving differential equations and I and my students who invented it, a rather new method, which I don’t know where it stands in the whole hierarchy in the methods used for solving partial differential equations today. I’ve got to find out from somebody. I’m going over… MR. LARSON: Yes, well partial differential equations are very essential in quantum theory and also advanced thermodynamics. DR. ATANASOFF: Oh, absolutely. You bet. MR. LARSON: They are absolutely necessary. DR. ATANASOFF: You bet… MR. LARSON: And not an easy discipline. DR. ATANASOFF: Thermodynamics was one of the subjects I taught in those days and I’ll tell you something else about it, but I better shut myself up for a moment. Let’s quit for a few. [Break in Video] DR. ATANASOFF: I was telling you what subjects my graduate students worked in: quantum mechanics, quartz, you know the crystal quartz. MR. LARSON: Oh yes. DR. ATANASOFF: We cut them and analyzed them and you know they have six coefficients of elasticity. It’s a complex crystalline substance and we, I provided the method for computing all these from measurements, and my students calculated, my students actually made these calculations. I believe to this very day the most complete collection of the elastic coefficients of quartz was done by my students and I. And now, you know me, when I did a thesis, I analyzed the partial differential equation by using the Rayleigh-Ritz process. You’ve heard of the Rayleigh-Ritz? MR. LARSON: Oh yes. DR. ATANASOFF: Well I invented a new method to replace Rayleigh-Ritz. And then I made it simpler so there was almost no mathematical work to be done. The only trouble was that as you attempted to solve a partial differential equation you had to solve a large system of linear algebraic equations at the end. And the problem with solving a large system, I’m going to talk a little about that subject, of course, later. Now, I studied whatever computers there were around and this included the IBM tabulator. MR. LARSON: Oh yes. DR. ATANASOFF: It was a great big machine and cost the university a lot of money and I felt that I ought to apply that equipment onto some of my own objectives. I got a man in statistics named Brant, Dr. A.E. Brant to help me and thought we would find a way of, a physics problem that we could solve by using a tabulator. MR. LARSON: Oh yes. It used the IBM punch cards? DR. ATANASOFF: Exactly. It used IBM punch cards. Well I thought maybe the thing to do was to use it to solve complex spectrum. And then I analyzed it further and then I found out that it wouldn’t do it. It was a dumb machine. It wasn’t very smart and then I thought maybe I could add some additional equipment to the IBM tabulator and make it do it and we did so and published a paper. MR. LARSON: Well that’s remarkable. DR. ATANASOFF: And the, strangely enough IBM was not very happy at all. We said nice things about IBM, but they weren’t very happy at all and years later the courts forced IBM to draw every piece of paper that had Atanasoff on it out. I found out that there is a piece of paper in IBM that says, “Keep Atanasoff out of the tabulator.” MR. LARSON: That’s fascinating. DR. ATANASOFF: We didn’t hurt the tabulator at all and in the end it worked as well as it ever did. MR. LARSON: Of course that was a good introduction into your beginning to think about computers then. DR. ATANASOFF: Perhaps I should say that IBM wouldn’t give me a copy of the internal drawings of the computer and I just sat down and figured out how I would have done it and proceeded accordingly and it worked fine. MR. LARSON: Well wonderful. DR. ATANASOFF: Now we’ve got this, my students had all these linear algebraic equations, just ordinary linear algebraic equations, but they have eight or ten or more unknowns in them. MR. LARSON: That means about a week of calculating. DR. ATANASOFF: Oh yes, but pretty quick it was several weeks. My students spent several weeks on those calculations and we knew that what we needed to do was to get the results more accurate, we had to double or triple the number of unknowns in them and that looked as though it would take most of a lifetime to do. MR. LARSON: You didn’t use determinants to solve them. DR. ATANASOFF: It wouldn’t help. MR. LARSON: That would get so complicated. DR. ATANASOFF: No, no, the determinates are fine in doing a theory of such problems, but had nothing to do with solving them. Solving them, you just have to do the numerical work. And my students were making a noise about this subject, and this is the greatest impulse that I ever had towards computing. I commenced to examine all the computers, all the methods of calculating that were available. It took months, but in the end there was nothing. MR. LARSON: Yes, well there were the Marshant calculators and the IBM tabulators, that was about all. DR. ATANASOFF: I tried to us an IBM tabulator, but it didn’t have the capacity, didn’t nearly have the capacity. It was slow too. And Marshant, you know how long that takes and I used that in my own calculations of helium. Oh, I commenced to wondering, to think about analog versus digital computers. I commenced to wondering if I should go into analog computers for solving this. A little analysis showed me that an analog computer, you know [Vannevar] Bush developed an analog computer called a differential analyzer. MR. LARSON: Yes, that’s right. I believe that was the most advanced calculator as of that particular moment. DR. ATANASOFF: That’s right and he spent a $1.5 million on that computer and he got this money from the Rockefeller Foundation. Warren Weaver gave him the money. Do you know Warren Weaver? MR. LARSON: I met him. As a matter of fact, Jane met him at your home. DR. ATANASOFF: Warren Weaver was a major professor of mine in college. MR. LARSON: Oh yes. DR. ATANASOFF: And I’ve had some associations with him since. MR. LARSON: Well that, of course, that work that was concentrated in analog computers, I think probably held MIT back a little bit from going into the digital. DR. ATANASOFF: Oh yes. You know why? I’ve argued with Bush and his assistant, Sam Caldwell. Do you know Sam Caldwell? MR. LARSON: No, I’ve heard that name before. DR. ATANASOFF: Sam Caldwell, I argued with them both. At this time I was getting sharper on the subject. I said, you know analog was my own invention. MR. LARSON: Is that right? DR. ATANASOFF: It was my own invention. MR. LARSON: What did Bush call it? DR. ATANASOFF: He called it differential analyzer. MR. LARSON: A differential analyzer. Analog is a… DR. ATANASOFF: You see he hadn’t sharped up the difference between an analog machine and a digital machine. The word “digital” was not of my making, it wasn’t my word. I called it computing machines proper. MR. LARSON: Oh yes. DR. ATANASOFF: I called a calculator a computing machine proper. MR. LARSON: Oh yes. DR. ATANASOFF: It didn’t use an analog; it went directly at it in a mathematical sense. That was my word and other people called it and I missed my word. That’s somebody else’s word. But the word analog came out of my mouth. It’s kind of amazing isn’t it and I think I can prove it. Now, I’ve argued, I was in contact with Bush and the last time I saw him was in ‘16 and… it was down in Washington. Can’t recall the place. Well, I looked the thing over and I knew that I had to build a computer and I was shaken that that was what my future was. I wasn’t enthusiastic about it. I didn’t want to do it, but I knew I had to and I knew I had the means. Somewhere I felt that I had the internal means to do that. So I started in to examine the whole subject of calculation. Now what did I do? I just sat down at a desk and just started making pictures and thinking. Making pictures and thinking. This went on for months and months, making… and I worked on a number of subjects and that included the subject of memory that you spoke about. I knew I had to have a relatively fast memory. I set my goal as to provide something that was several times as fast as an IMB tabulator. MR. LARSON: Which was all mechanical. DR. ATANASOFF: Which was all mechanical. That’s what I set as my model. I didn’t think about working faster. As a matter of fact I made my computer machine way too slow because all the means I introduced to the computer were very rapid. One of the methods that I induced for computing was in memory. Now I knew that mathematically speaking that depending on what base I was going to use, it would change the method of memory that I would probably want to use. I could see that. So on the subject of base came to the floor. So I decided I would try to find out what base to use first. I decided, you know I’ve already said I decided to use digital machines, what we call digital machines, which I call computing machines proper. Well when I examined the base I have a mathematical theory which is contained in here and I presume it isn’t very good to try to give the theory verbally, but the theory showed me that if you’re doing ordinary arithmetic, the lower the base, the better. The lowest base you can use is two. MR. LARSON: Oh yes. DR. ATANASOFF: One of my calculations showed me I should use the base e, but it turns out that that was a mistake because it turns out… MR. LARSON: It has to be a whole number. DR. ATANASOFF: No. MR. LARSON: No? DR. ATANASOFF: It theoretically gave me the base e, I knew I could use it as base, so I had to use either two or three and I examined which would be better at that stage, but it turns out that there was an error in the theory that gave me the base e. When it comes to multiplication, the base two is best. The base two is really the best, better than e. Better than three. When it comes to division, theoretically from a theoretical point of view, you could use a base of three or a base of four, but there are a lot of other things that enter this field and when it comes to a question of memory, either you punch or you don’t punch, that’s two. If you attempt to apply that to a base e, I mean base three or any other base, it causes a complication of like the base 10 card of IBM’s. It’s something like that. You know how that worked? MR. LARSON: Oh yes. DR. ATANASOFF: Well that’s an illustration of a man using a punch or no punch problem which really is quite suitable for base two. You try to use it for base 10 and it uses up more card than is proper. So, the base, we are driven again the base two. When it came to the logic system, which I will talk about later, the base two is overwhelmingly better. When it came to memory, it was much better and so I was pretty much driven to the base two. When we’re attempting to get a memory for base two, you either punch or you don’t punch, you know, you can, you know if you use a relay computer it automatically works out for base two. MR. LARSON: Relays and diodes… DR. ATANASOFF: That’s right; however, I knew that the cheapest thing I could use would be a condenser. MR. LARSON: Oh yes. DR. ATANASOFF: So I commenced to use condensers and I used .003 micro [inaudible], a very small condenser. I didn’t go anywhere near the bottom. I could have used much smaller condensers. I did use as a matter of fact. The time before, let’s talk about the condensers that I did use. It would take 40, 50 seconds for the signal to die off in a condenser of that size and now… MR. LARSON: What kind, were these coil or electrolytic? DR. ATANASOFF: They were on paper. Any ordinary paper condenser would work. I didn’t have any kind of trouble with any kind of condenser. Everything worked, of course the, I knew that this was really complicated because the condensers would die off and I had to find some way of keeping them from dying off. Now my original purpose would be to use a string or something of the kind that could snap into one of two positions, have a little potential here, then higher and then a little potential here and you could snap it from that position to this position, or that position to that position. And change from zero to one and one to zero. That requires a system with two minimums of potential energy. You see what I’m drawing. MR. LARSON: Oh sure. DR. ATANASOFF: They, that was what I first thought I would do. And then I was driven by the little boy going to the grocery store. The little boy going to the grocery store… MR. LARSON: Your condensers were… DR. ATANASOFF: Like the little boy, it peters out. Runs down. The little boy’s memory runs down and he doesn’t remember what his mother told him to do. So once the little boy goes, as he starts to the grocery store, he repeats what his mother told him. And he repeats it at such frequent intervals that he doesn’t have time to forget between times. MR. LARSON: Oh yes. DR. ATANASOFF: Having had this practical idea in my mind I immediately commenced to put this into electronic shape and I milked condensers into circuits that had this property. At frequent intervals they would test what they were. And put them into the most desirable state, pull themselves up. MR. LARSON: So you have to, you had to read what was… DR. ATANASOFF: In it. MR. LARSON: …on the condenser at the same time… DR. ATANASOFF: Regenerate. MR. LARSON: …regenerate the charge because it went down. DR. ATANASOFF: You got it. MR. LARSON: You had to do that within the 50 seconds. DR. ATANASOFF: Oh yes. As a matter of fact I did it at one second intervals, but our present computers, do exactly the same thing. They have a little condenser in there and it will wash out, but they have refresh cycles which will continue to flip it through the system and are bringing them back to the original state. MR. LARSON: I didn’t know that. DR. ATANASOFF: That is the way they are made. That is true of all memories in modern computers. I’m telling you something out of place a little bit. You know the magnetic memories wouldn’t die off because of time. When you, I could say that if you reach in and take out the charge, why then you have to recharge it in that state in order to have it retain its original value. That’s a complex situation on magnetic, on magnetic cores. Magnetic cores were like that. Well. Now, I worked on a lot of things here, which could possibly be, for months and months, I guess for a year and half to two years, I was working on just little giblets of elements that could be put into a computer. I hadn’t in mind how I could build a computer. I didn’t have in mind how I would ever reach my goal. I went, I’ll tell you what happened. I went out to my office one evening, in the middle of the winter of 1937, ’38, and I sat down at my desk and attempted all my stuff around. Tried to work on this, tried to work on that. I became extremely irate with myself and I was upset to an extreme degree, then I did something that I have only done two or three times in my life. I went out and got into my car and started to drive. It was 20 degrees below zero, but there wasn’t a bit of ice on the pavement. It was all dry and I drove hard for several hours. Driving hard enough so I couldn’t think about computers. I had to think about the road and I was able to mitigate my unhappiness that way and at last I said, “Atanasoff, this has gone far enough. You have got to do something.” And then I glanced over and I was crossing a bridge 189 miles from my starting point. MR. LARSON: Wow, you really traveled. DR. ATANASOFF: I traveled at 80, 90 miles an hour in a Ford V-8 of the era and I had a good heater in there and I was comfortable. It was extremely cold, 20 degrees below zero. Somebody analyzed my situation and they said I was looking for a drink. Well I wasn’t looking for a drink, until I glanced over and saw the water. Then I said, while you’re going to Illinois, of course you could get a drink in Illinois. I went on and there was a honkytonk on the side of the road within a quarter of a mile, there was a honkytonk in Illinois and I drove in. I parked my car and walked in. I had a heavy coat on, very heavy coat. I remember the strength I used to raise it to the hook. It was plenty heavy on my hand as I raised it to the hook and I sat down and ordered a drink. All of the sudden I realized that my mental capacity had improved remarkably, that I could think things through. I knew immediately I could think things through and I knew immediately what I had to do. I sat right down and started working on the computer and I worked on it continuously for three hours. People say did you have any paper? I don’t remember any. I didn’t need any paper. Everything I did I remembered perfectly. And I got up and I drove home at a slower rate. I had come to four conclusions while I sat there in the honkytonk. MR. LARSON: That’s a remarkable story there. What were those conclusions? DR. ATANASOFF: I ought to find it here because you know it’s like me to forget one of them. Yes, in the first place you might not think this is much of a conclusion, but I decided I would not do a mechanical system, but an electrical system. This may not seem as though it was a great step forward. I had some letters from the members of the IBM staff and a member of the IBM staff told me they would never build an electrical computer, never build an electric computer. In spite of custom, I would use base two. I was going to translate the problems that I had with base ten, do the calculations with base two and translate it back. I would use compositors for memory, but I was going to use what I told you about the little boy. There was going to be a mechanism there that would test what it was at frequent intervals and restore it to its original form. I would, this last one came originally from these other things that I had thought about before, but I was finalizing it in my mind as I was sitting there. I would compute by logical action and not by numeration. MR. LARSON: Oh yes. DR. ATANASOFF: I was going to build a logic system into my computer, I didn’t know how to do it, I didn’t have the smallest idea, but here’s a box and it has leads going in and leads coming out and into it you put the initial quantities and out would come the final quantities and the machine was a logic based computer. MR. LARSON: Oh yes. DR. ATANASOFF: And now, when I got home I had these four items and I commenced examining each of them in detail. This went on not merely for a month, it went on for a number of months because I could do parts of them, parts of them I already had answers for, but when it came to the logic system, I had to start from the beginning and build a logic system and that took me months. As I look back over that era, I realize that these four, four steps that entered my mind at the honkytonk are in every computer today. MR. LARSON: That’s right. They are pretty fundamental things. They have to be. DR. ATANASOFF: Pretty fundamental and now I was delighted when I found that out. I didn’t find it out until two or three years ago that every computer, every one of them as long as they have magnetic cores, that doesn’t quite fill the bill, it would be off by one, but as far as all the rest go. The dynamic memory as low capacity and the refresh cycle is exactly according to my principal. MR. LARSON: Oh yes. That’s amazing. DR. ATANASOFF: I then knew when I had gone through these consequences and my trip to the roadhouse had actually, now can you imagine me sitting down and drawing out vacuum tubes without ever putting a vacuum tube to a circuit. I finally got it to where it would work and that was the state it was in. I hadn’t ever tried it mechanically. I hadn’t. I could have very easily. I was good at the arch, but I didn’t feel I needed to because I knew exactly how those vacuum tubes would work with that circuit and so all I had to do was sort of draw out the circuit and plant it. Then I decided I couldn’t go any further by myself. I had to have some help. I commenced approaching at Iowa State University, College at that time. I was walking across the campus and I saw a member of the electrical engineering department named Harold Anderson. I can show you within a foot of where I stood at that time. I said to Harold, “This is what I’m up against. I’ve got to have some help. Can you make any suggestions?” Harold Anderson took a minute. Always a very careful man. He took a minute, maybe a minute and a half, thinking it over and he said, “I’ve got your very man.” He said, “His name is Clifford Berry.” MR. LARSON: Oh yes. And that’s how you got Clifford Berry. DR. ATANASOFF: That’s how I got Clifford Berry. I suppose we had to ask Clifford if he would agree, but there wasn’t any problem. I had Clifford Berry. And that was in the spring of 1939, I decided I could employ Clifford Berry and he couldn’t start work until fall. He started work in September and before the end of the year, October, November, December, he had the circuits working. MR. LARSON: He really put them together fast. DR. ATANASOFF: Not an electrical computing machine, just the rudiments of a computing machine. When he showed that my precepts of how it should go together were right. MR. LARSON: That’s remarkable. So then you actually had a bread board, an arrangement there. DR. ATANASOFF: A bread board. We have a picture of it in here. MR. LARSON: Yes. DR. ATANASOFF: This picture of a bread board in here. MR. LARSON: Fine. I was wondering if you could open it up to that picture and then hold it against your chest and we’ll zero in on it. DR. ATANASOFF: Can you zero in on it? MR. LARSON: Oh yes. DR. ATANASOFF: I can do that. MR. LARSON: Zoom right in. DR. ATANASOFF: Here it is. MR. LARSON: Oh yes. Sure. Oh, very good. DR. ATANASOFF: Can you see it alright? MR. LARSON: We got a very good record of it. MRS. LARSON: Yes. DR. ATANASOFF: Now that was, we didn’t have a picture of it. This is my memory and it was done by an artist who put things together, but he did it according to my memory. Really this is what a computing machine is about. MR. LARSON: Oh yes. And you’ve got all those elements put together there. DR. ATANASOFF: I was able to actually assemble them at that stage. MR. LARSON: Yes and you were certainly able to test the various components of them. DR. ATANASOFF: As far as putting a number in there, you just had to touch a wire to put a number in. it would only do that. I didn’t build any equipment to put numbers in or take numbers out. And then when I wanted to see what was in there I had to test the condensers by means of a volt meter or a vacuum tube volt meter or something. That’s all it took at that stage. MR. LARSON: That’s a remarkable story then. Fine. Then what was your next step after you had done all these and tested and so forth. DR. ATANASOFF: You know my purpose was to make a machine solve system, linear solve system. MR. LARSON: Oh yes. DR. ATANASOFF: And that was to build a machine specifically for that purpose. MR. LARSON: Oh yes. DR. ATANASOFF: And we did that. MR. LARSON: Oh yes. DR. ATANASOFF: And we, I should also, as long as you’re looking at the picture of that machine, why, we ought to, here is the circuits that did the jogging. You ought to take, make a… MR. LARSON: Oh yes. Zero in on that as far as you can. MRS. LARSON: I got it. MR. LARSON: Fine. Good. That’s very readable. DR. ATANASOFF: I had to build a logic circuit to go on that. And here’s the logic circuit. This is exactly the way I drew it before I even had Clifford Berry there, this is exactly, I drew that. MR. LARSON: It’s very complex. DR. ATANASOFF: It’s complex and this is for the base two and if you attempted to build a logic circuit for the base five, it’s much more complicated. MR. LARSON: You’d soon run out of vacuum tubes. DR. ATANASOFF: Right, you’d soon run out of vacuum tubes and this one took 14 trials. Thirteen trials I think it was. I got 14 here in the picture, but my trials came two in a bottle so I had to have 14. Over on the next page is some table which shows how the thing works. I have had to try to put together for a core procedure, a treatment of this subject. Did you know that such circuits are put together today by using complex algebra which is called, what? You know what it is. MR. LARSON: Boolean algebra. DR. ATANASOFF: Boolean algebra, that’s right. You know I already studied Boolean algebra but it didn’t help me in that. MR. LARSON: Oh I see. DR. ATANASOFF: I just didn’t make the connection. MR. LARSON: I guess the use of Boolean algebra came a little later, but it was such a valuable tool, once that is, in fact used today. DR. ATANASOFF: And it’s the very thing that I was perfectly capable of doing, but I just didn’t do it. I… Clifford… MR. LARSON: So Berry worked right with you there. DR. ATANASOFF: Yes, he did. You understand we were on a meager budget. We didn’t have any money and here we are doing it. Nobody else had put a computer together like we did. MR. LARSON: So you then built the machine to use the computer principles for the specific purpose of solving these linear equations. DR. ATANASOFF: The lawyer said that is a special purpose machine and should not be considered with the general purpose machines, with which you are comparing it. My opposing lawyer was saying that to me. I’ve got a story to tell you about that. Maveridge worked a good many years on computers. MR. LARSON: Oh yes. DR. ATANASOFF: He didn’t build much of anything. He just did what I did, that is he made pictures. And after he was dead, the parliament wondered if they hadn’t mistreated him. They knew he was a great man and they wondered if they hadn’t mistreated him. So they got together a committee to decide, at that date if they shouldn’t build a Maveridge Machine. MR. LARSON: Oh yes. DR. ATANASOFF: So that committee droned on and finally decided that the drawings were not sufficiently accurate so that they could build a machine. It wasn’t like my machine which was in vacuum tubes that Clifford could build. It was something more complicated. MR. LARSON: A very complicated gear system. DR. ATANASOFF: Then the following meeting took place. Someone rose and said, “If that machine would only solve linear algebraic equations we ought to build it because those, that subject is the greatest importance to man.” MR. LARSON: Oh yes. DR. ATANASOFF: They agreed on that subject, but they couldn’t see how the Maveridge Machine would ever solve linear algebraic equations. MR. LARSON: Oh yes. DR. ATANASOFF: But I immediately turned my machine the moment I had a grasp of how it would go together. I immediately turned it in that direction. MR. LARSON: Oh yes. DR. ATANASOFF: I didn’t even know this story at the time I did it. I just blundered in. This is what I’m going to tell you next. Let’s cut off. [Break in Video] WOMAN OFF-CAMERA: Did you take the time? DR. ATANASOFF: No, I’m alright. MR. LARSON: Well to continue this, you’ve brought up some very fascinating points and so if you would just expand on that I think it great. DR. ATANASOFF: Do you realize that motivation, my motivation to study computing machines was to solve large systems of linear algebraic equations? This is like X plus two Y equals seven, X minus three Y equals 15. That’s two unknowns in two equations, but when we go to the place where we have 10 equations and 10 unknowns and there on up, we have something that can’t really be done to man’s satisfaction. It takes too long. I, we got to have the use of machinery. I was aiming in that direction, aiming at the desire of the people who investigated Maveridge that we would have a machine to do such processes. This is a kind of complex part of my story. The first place, we will just say a few words about linear algebraic equations. You know what we ordinarily do in solving such systems is to use the equations which we have in pairs and eliminate one unknown between a pair. We can always do that. Now it consists of this, you have a pair of equations by a proper combination of adding or subtracting the coefficients and the constant term in a specific way it will eliminate one of the unknowns. MR. LARSON: Yes. DR. ATANASOFF: That’s the process. MR. LARSON: That’s the standard process. Works very well until you get to three or more, it gets very laborious. DR. ATANASOFF: Very tiresome. Now I guess the main part of it, you know way back I thought I would attempt to solve this problem, I didn’t mention it before, but I thought I would solve it by using, say, 15 or 20 Monroes. They all have a common axis. I geared them all on a common axis. So they all turned together. That looked as though it was a possible approach to this subject, I could have done it, but the problem with getting the numbers into and out of the Monroes was too complex and also had to be recognized before we could get anywhere. Now, I suppose we had a pair of equations and the initial coefficient was two in one case and three on the other. We would multiply the first one by three and the second one by two and subtract them and the first equation, the coefficient on the first term would become zero and it would become eliminated. We would have the problem solved. MR. LARSON: Yes. Well that’s very simple for two equations and two unknowns. DR. ATANASOFF: Yes, it is. However, let me say that you could take any linear combination of any two equations and it would have the same answers as the original system. MR. LARSON: Oh yes. DR. ATANASOFF: Any linear combination that you please. I had to try to bring that to realization. Now, you know in computing, we all know that multiplication requires successive additions and changes of the, changes of the power to which we have to raise 10 for which we raise 10 numbers to get that particular element. I spent a long time thinking how I could do it. Strangely enough I came to the following answer. I would do it by division. MR. LARSON: Oh yes. DR. ATANASOFF: By division. I would take the first equation and divide it by the second one and as I did the division what would also be left over would also be an equation. If you take all the way across the array of coefficients, would also be an equation with the same coefficients, divided properly, the division would go to zero and hence would eliminate one unknown. MR. LARSON: Oh yes. DR. ATANASOFF: You’re understanding it perfectly and I did this yesterday. I applied this over again yesterday as to how I would say it. I’m succeeding with you and that pleases me immensely. MR. LARSON: Well that’s fine. DR. ATANASOFF: Now I ought to tell you this, that ordinarily in computing, you subtract the divisor off of the dividend until the dividend carries through. You know, you’ve gone one step too far. You add it back in until it was just before it was carried through and then you would shift to a different power of 10. Do you see what I’m driving at? MR. LARSON: Yes. DR. ATANASOFF: Then you would subtract and subtract until it would carry through. Then you would add it back in until it would, now this process is what is called ordinary division with computers. Ordinary division with computers. I decided that this process was too complicated. I invented a new process and this was an invention of mine. I would subtract until it carried through and then I would shift, change to a different power of 10 and then instead of adding I would, I mean instead of subtracting I would add until it carried through and with this series of numbers I could get the same result as the former one. This is called division, Alice, what is this division… my wife’s got the words. MR. LARSON: Well, never mind, it will come. DR. ATANASOFF: I’ll get it in a minute. Now do you know these two methods are used in computation? I invented them and they are available on those machines today. You can use either system of division that you please. MR. LARSON: Oh yes. That’s actually built into the computer mechanism. DR. ATANASOFF: Yeah, built into the computer mechanism. I could do it with, let’s see, we have some words for that. Words escape me. What are you going to do? Turn it off and I’ll have it in a minute. [Break in Video] DR. ATANASOFF: I’ll repeat that. Where you ordinarily would subtract until it’s carried through and then adding back, that’s called restoring. MR. LARSON: Oh yes. DR. ATANASOFF: That’s called restoring division. And ordinary computers today will do that if you wish. They will also do the other which I invented and consists of, instead of restoring it, you leave everything just as it was, but the next time you add until it carries through. The next time you subtract until it carries through and this is called non-restoring division and it was non-restoring division that I invented. MR. LARSON: Yes, and those are built into the so called read only memory in most computers then. DR. ATANASOFF: That’s right. So in order to save time, I use a second process which saves, well it doesn’t save so much with base 10 because you only save one step in the five steps, but if you’re using base two it saves half your time because you do it so much. It saves half your time if you use restoring division. MR. LARSON: That’s very interesting. With base 10 it’s not so important, but with base two it’s very essential. DR. ATANASOFF: Very effective, yes. and I invented non-restoring division. Alright now I arranged the new computer so that it would have 30 peels. I couldn’t work 30 equations with 30 unknowns. I could only work to 29 equations and 29 unknowns because it had to have some place for the constant term. And in the 30 peels, one place would be for the constant term and the other for all the other coefficients. MR. LARSON: Oh yes. DR. ATANASOFF: So I arranged two drums. Each which had 30 peels and that would, could be used up to 29 equations and 29 unknowns. You put one equation in one drum and another equation in another drum and connect the two drums with logic systems which I have previously given and you could just add, add the, lets say one of the drums is called a keyboard and I needed a noun and I called it a keyboard imackus. The other one is called a counting imackus and those were the two names of them and I’ll just use those names here. And all I had to do was rearrange so that it commenced to divide a pair, say the first coefficient of each equation, divided the two of them, until that coefficient was dropped to zero. When that coefficient dropped to zero, you know the remainder in division after you’ve carried it out long enough to drop to zero. MR. LARSON: Well this is a very good explanation. DR. ATANASOFF: Gosh, that’s wonderful. MR. LARSON: Incidentally I am just amazed such an operation of 29 equations and 29 unknowns. As I visualize it that would take a person a lifetime. DR. ATANASOFF: It would take a major part of a lifetime to work it out and be sure everything was right. MR. LARSON: So that’s an amazing new tool. DR. ATANASOFF: That’s what I was aiming at. I built this machine and I can show the picture here. There is the machine in its last photograph. I will use this photograph to tell something about it. [Break in Video] DR. ATANASOFF: …picture that was taken and I have the negative of this. These pictures came from various sources, but this particular picture, I have the negative. MRS. LARSON: Wait a second. MR. LARSON: We would like to see that. MRS. LARSON: Hold it up slightly. MR. LARSON: That’s great. We’ve got a good picture of it now. Fine. DR. ATANASOFF: Now this is the principal parts of the machine. Here is one of the drums and the other drum was inside this case. Here is a card reader. We put the numbers into the machine in base 10 and it was taken over by the card reader and then the machine converted to base two and I had to build a special system from changing from base 10 to base two. And it worked in either direction. Now it kept the numbers for the given system of equations, it kept the numbers on a scratchpad memory. This is the scratchpad memory. One of these things rights the memory on a card, punching it on a card and the other one reads it. MR. LARSON: Oh yes. DR. ATANASOFF: So that way we keep track of the numbers as we pass through the entire process of solving it. MR. LARSON: That’s very ingenious. DR. ATANASOFF: This, I had to invent everything. I couldn’t buy anything. Nothing was available. So I invented the scratchpad memory and the scratchpad memory consisted of having a card pass below a series of terminals and I put a high spark discharge through paper and this made the record. Clifford was very interested in this and worked on that. We thought we had this scratchpad memory perfected. We tested it, but when we finally came to the final machine, it was correct, all except for one part, one place in 10 to the fourth or 10 to the fifth. It said there was an error in the scratchpad memory. As far as I know this was the only error that existed in the final machine. MR. LARSON: Oh yes. DR. ATANASOFF: Now, in other words, it recorded all the coefficients in base two on one of these and picks it up in the other, put it back in the machine and kept that procedure. This is the way the machine worked. MR. LARSON: That’s remarkable. So you then could feed these things, convert the decimal to binary, perform the calculations… DR. ATANASOFF: After you converted it, it came on a base two card and you just put base two cards in until the equations were solved. MR. LARSON: Oh yes. That’s a very interesting story there. DR. ATANASOFF: Now, let’s see. Here is a picture that may portray some… MR. LARSON: That gives you a very good… DR. ATANASOFF: You might like to take this picture and add it… MR. LARSON: We got that. DR. ATANASOFF: This was made from my memory by an artist also. MR. LARSON: Oh yes. Fine. DR. ATANASOFF: I will say this concludes the mechanical operations of building the machine. MR. LARSON: Fine. Very good. [Break in video] DR. ATANASOFF: …in the course of building this machine, money got to be more and more a problem. So I decided I would write a description of the machine and this was as of that day. This is that description. MR. LARSON: Oh yes. [Break in Video] MR. LARSON: Let’s change the tape here. It will just take a second. MRS. ATANASOFF: Does that mean you have an hour? MR. LARSON: Yes. MRS. LARSON: It’s still going. MR. LARSON: I just wanted to make sure. MRS. LARSON: I looks like it’s still going. MR. LARSON: Let’s look and see. MRS. LARSON: Oh, it’s still on. [Break in Video] MR. LARSON: Go ahead and we just put in the next tape. DR. ATANASOFF: In 1954, Alice, was it ’54 or ’56? MRS. ATANASOFF: ’54. DR. ATANASOFF:’54. I was visiting, oh by this time I had left the Naval Ordinance Laboratory and started a company of my own. MR. LARSON: Oh yes. But in the meantime, let’s say that Dr. [John] Mauchly had left the Naval Ordinance to go to… DR. ATANASOFF: Oh you know he only worked temporarily, once a week for Naval Ordinance. MR. LARSON: Oh I see. He was only down there for a short period of time. DR. ATANASOFF: I’m sorry. MR. LARSON: I wanted to clarify that. DR. ATANASOFF: Yeah, he was working for the university up at Philadelphia, all the time. MR. LARSON: At the University of Pennsylvania. DR. ATANASOFF: I have discussed this matter with Arthur Burks. Do you know who he is? MR. LARSON: Oh yes. I’ve heard of him. DR. ATANASOFF: Well, Dr. Burks was there at Philadelphia with Dr. Mauchly at the University of Pennsylvania. And he can’t quite understand why Mauchly is coming down to see me either. It’s a very tricky spot. But I got a visit one morning, and I had started a company of my own and I got, a patent attorney from IBM came by and wanted to talk to me. MR. LARSON: This was about 1954, you said. DR. ATANASOFF:’54, yes. And he says, he had found out about Mauchly’s patents, and he says, “If you help us, we will destroy the Mauchly patents because they were derived from you.” Those were his direct words. MR. LARSON: So his patents were directly derived from yours. DR. ATANASOFF: That’s what he said. That’s what he said. I wasn’t quite sure how I should react. I had myself formulated that opinion. It was him speaking. I said I can’t. I can help you, but I can’t work full time with you because my staff and I have put all the money we got into this company and we got to make a success of it. MR. LARSON: Oh yes. DR. ATANASOFF: And he wrote me one letter and I answered him. He wanted to get my original notes. I didn’t know where they were. The notes that contained this. I didn’t know where they were at the moment. I had been confused and moved here and there and had stuff… MR. LARSON: I know how that works. When you move, you lose a lot of things. DR. ATANASOFF: That’s right and it was all…okay. So they abandoned me and they proceeded to pay Mauchly and [J. Presper] Eckert a sum of money to secure the rights from them. This is IBM. IBM got its rights in that way. MR. LARSON: Oh yes. DR. ATANASOFF: Then we’ll pass onto the time that I, the Mauchly and Eckert patents had been transferred to… MR. LARSON: Sperry-Rand. DR. ATANASOFF: Sperry-Rand. I’m trying to say Sperry-Rand. MR. LARSON: And well however, the patent did not come out until quite late as I… DR. ATANASOFF: Very late. MR. LARSON:’64, or something like that. DR. ATANASOFF: That’s right. I’m sorry I can’t give you the exact date, but it was very late just as you say. I didn’t memorize those. I have to memorize those. Sperry-Rand, a good deal of water had gone through the dam. MR. LARSON: Oh yes. DR. ATANASOFF: And Sperry-Rand was attempting to collect a billion dollars in royalties. MR. LARSON: A nice round sum. DR. ATANASOFF: Yes. A nice round sum. Well, two of the people, they selected two companies they didn’t think had much power, that they could beat down and collect the royalties. That was Control Data and Honeywell. Now both Control Data and Honeywell were willing to go a little bit higher than the nominal patent fees, but they weren’t willing to go any further and they rebelled. MR. LARSON: Oh yes. DR. ATANASOFF: An attorney named, what is it? MRS. ATANASOFF: Kirkpatrick. DR. ATANASOFF: Kirkpatrick for Control Data had read a book, the first book on the patents of the era by… MRS. ATANASOFF: Richards. DR. ATANASOFF: Who? MRS. ATANASOFF: Richards. DR. ATANASOFF: Oh yes. By Richards. So he found out immediately where I was and came up to see me. So he talked to me and introduced me to the people at Honeywell and the people at Honeywell were talking to me and pretty soon we were shaking the woods. MR. LARSON: Oh yes. DR. ATANASOFF: The thing ended up by my agreeing to act as witness for the two companies. MR. LARSON: So it wasn’t a joint action on the part of Honeywell? DR. ATANASOFF: No, it wasn’t. Separate actions. MR. LARSON: Separate actions. DR. ATANASOFF: During this period, I heard from Mauchly. I hadn’t heard from him for a good while. MR. LARSON: Oh yes. DR. ATANASOFF: I heard from Mauchly. Mauchly says I would like to come and see you and I would like to bring one of my patent attorneys in to see you. Then, I won’t go through the details, but he came down and he brought the patent attorney with him and we talked for a while and we had lunch. Alice prepared us a lunch and then we talked somemore. The name of the patent attorney, his name was Dobbs. Dobbs was a patent attorney for both Honeywell, no, he was an attorney for… MR. LARSON: Sperry-Rand? MRS. ATANASOFF: IBM. DR. ATANASOFF: No, not IBM. MRS. ATANASOFF: Sperry-Rand. DR. ATANASOFF: Sperry-Rand. He was a patent attorney for Sperry-Rand in both of these cases is what I’m trying to say. So, in the course of discussion, by this time I had read a good deal of more material and I had the original patents. I told Dobbs he couldn’t possibly win. I just flatly told Dobbs he couldn’t possibly win. You know it’s very strange. Most people would have been irate and offensive, but Dobbs wasn’t irate or offensive at all and he became a firm friend and is today. MR. LARSON: Well that’s a very interesting point there. DR. ATANASOFF: He, I explained exactly why he couldn’t win and he didn’t win for those exact reasons. MR. LARSON: He must have been a little devastated seeing this billion dollars sort of blow away. DR. ATANASOFF: Oh yes. He was just as genuinely nice to me as he could be and he wasn’t regarded as a very gentle man, but he was in my case and the last time I entered into a court procedure he insisted that Alice and I meet his wife after the court was over. MR. LARSON: Oh yes. DR. ATANASOFF: To just show how much he had become my friend, the result of having this conversation with him. He found out I was exactly right. You could talk about the judge not being smart. He didn’t have to be smart if he could just read ordinary English. MR. LARSON: Oh yes. DR. ATANASOFF: Because it was very clear that Mauchly and Eckert had put together a patent case and it had been accepted. You know why I didn’t have a patent? I didn’t have a patent because Iowa State College forced me to sign a patent contract with them and refused to pursue the patent. MR. LARSON: They neglected a very important thing there. DR. ATANASOFF: Neglect and I shouldn’t of stood for it, but I did. MR. LARSON: Well at any rate, how long did this patent suit go on then? DR. ATANASOFF: It went on for a year. MR. LARSON: Oh yes. DR. ATANASOFF: It actually sat for 10 and a half months. Actually the judge was present in a court procedure for 10 and a half months. I was only there during part of this time. and I can tell you some details of my controversy with opposing council. I don’t know if that is a good thing to add to your… MR. LARSON: Well fine. But essentially the net effect was so that as a result of your notes and documentation… DR. ATANASOFF: By this time I had accumulated all of my documents, including this. One of my original papers. MR. LARSON: That particular one must have been a devastating document. DR. ATANASOFF: The whole thing just put them into a position where they didn’t have a chance. By this time they had gotten rid of Dobbs and had a very aggressive council. MR. LARSON: Oh yes. DR. ATANASOFF: And treated me very badly in the court room, but it didn’t really matter. I withstood it. I don’t know what they, they might have conceived that by putting a hard attorney against me I might breakdown, but of course I don’t breakdown, I just slowly paddle on. When the court procedure was over, I went over to shake hands with the opposing council and the man hated like everything, but he finally got his hand up. He perhaps knew better at that stage, that my testimony had been overwhelming. MR. LARSON: Well both your testimony and your documentation in the form of these reports, could have been very… DR. ATANASOFF: Let’s just name a couple of things that Mauchly attempted to steal away from me. The first place, he actually, what kind of division? MRS. ATANASOFF: Non-restoring? DR. ATANASOFF: What? MRS. ATANASOFF: Restoring? Non-restoring? DR. ATANASOFF: Restoring. I had this restoring and non-storing division and he called it in his patent claims. He got a patent of it. Of course when the judge got through, there was no patent there. MR. LARSON: Oh yes. DR. ATANASOFF: They, his jogging. They used my jogging. MR. LARSON: That’s a very important principal. DR. ATANASOFF: You know what? He used base 10. He didn’t use base two. MR. LARSON: That’s interesting. DR. ATANASOFF: And he used, he didn’t use logic systems. He just counted. He had a machine that would just count. The rest of the story was that I outlined for Mauchly, if I needed a differential analyzer, I would build it according to this principal and Mauchly copied that into his ENIAC. He also copied that into his ENIAC. MR. LARSON: Oh yes. DR. ATANASOFF: By this time, well, they, you know, his wife has now written an article which is in this publication, which shows that he had done digital computing before he saw me, but Mauchly had told me that he had never done digital computing at the time he was at my house. MR. LARSON: There wasn’t any documentation of it. He didn’t make any reports of it. DR. ATANASOFF: She claims to have some, but it hasn’t entered into a court procedure and I don’t think most people, I think she’s innocent. I think Mauchly told her that before he died. And she’s written that since he’s died. MR. LARSON: Oh yes. It was a very interesting historical controversy here. DR. ATANASOFF: Yes. MR. LARSON: Well as they say, your documentation on this is overwhelming. DR. ATANASOFF: You read the judge’s decision. MR. LARSON: Oh yes. [End of Interview]
Click tabs to swap between content that is broken into logical sections.
Rating | |
Title | Pioneers in Science and Technology Series: John V. Atanasoff |
Description | Oral History of John V. Atanasoff, Interviewed by Clarence Larson, March 13, 1984 |
Video Link | http://coroh.oakridgetn.gov/corohfiles/videojs/CL_Atanasoff.htm |
Transcript Link | http://coroh.oakridgetn.gov/corohfiles/Transcripts_and_photos/GMU-Clarence_Larson_Interviews/Atanasoff_Final.doc |
Image Link | http://coroh.oakridgetn.gov/corohfiles/Transcripts_and_photos/GMU-Clarence_Larson_Interviews/photos/Atanasoff.jpg |
Collection Name | Clarence Larson Collection |
Related Collections | COROH |
Interviewee | Atanasoff, John V. |
Interviewer | Larson, Clarence |
Type | video |
Language | English |
Subject | Computers, History of; Mathematics, History of; |
Date of Original | 1984 |
Format | flv, doc, jpg |
Length | 1 hour, 33 minutes |
File Size | 321 MB |
Source | George Mason University, Fairfax, VA |
Citation | Clarence E. Larson Science and Technology Oral History collection, Collection #C0079, Special Collections & Archives, George Mason University Libraries. |
Location of Original | Oak Ridge Public Library |
Rights | Copy Right by the City of Oak Ridge, Oak Ridge, TN 37830 Disclaimer: "This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that process, or service by trade name, trademark, manufacturer, or otherwise do not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof." The materials in this collection are in the public domain and may be reproduced without the written permission of either the Center for Oak Ridge Oral History or the Oak Ridge Public Library. However, anyone using the materials assumes all responsibility for claims arising from use of the materials. Materials may not be used to show by implication or otherwise that the City of Oak Ridge, the Oak Ridge Public Library, or the Center for Oak Ridge Oral History endorses any product or project. When materials are to be used commercially or online, the credit line shall read: “Courtesy of the Center for Oak Ridge Oral History and the Oak Ridge Public Library.” |
Contact Information | For more information or if you are interested in providing an oral history, contact: The Center for Oak Ridge Oral History, Oak Ridge Public Library, 1401 Oak Ridge Turnpike, 865-425-3455. |
Identifier | AJCL |
Creator | Center for Oak Ridge Oral History |
Contributors | McNeilly, Kathy; Stooksbury, Susie; Reed, Jordan; |
Searchable Text | PIONEERS IN SCIENCE AND TECHNOLOGY SERIES ORAL HISTORY OF DR. JOHN V. ATANASOFF Interviewed by Clarence Larson Filmed by Jane Larson March 13, 1984 Transcribed by Jordan Reed DR. ATANASOFF: …computing. I had an interest in calculation from my earliest days. MR. LARSON: Yes, incidentally, where were you born, Doctor? DR. ATANASOFF: I was born in Hamilton, New York. MR. LARSON: Hamilton, New York. DR. ATANASOFF: My father was born in Bulgaria, came to this country at the age of 13 and was left alone here at the age of 15 and managed himself somehow through Colgate University where he met my mother, and I was born in the hill behind the University at Hamilton. Have you ever been to Hamilton? MR. LARSON: No, I never have. I have been in that general area. DR. ATANASOFF: Yes. I, well I’ll start this way. I was five years old when I started school and I would soon be six on October 4th. After about a month, my mother said, “Let me hear you read.” My mother heard me read and she says, the following, “I see I will have to teach you myself.” Those were her exact words. And she did for about a month. She switched me from a Lexile reader to a phonics reader and then told me that you should always use phonic reading. But it’s the best we have, but it’s very poor in English and when you get to spell you will have to memorize every word. MR. LARSON: That’s a very interesting point there. DR. ATANASOFF: She’s a sharp woman and only a month, after a month I took off on my own, following her advice and at the end of the year I could read almost everything. MR. LARSON: Well that’s a remarkable story there. DR. ATANASOFF: It’s a remarkable story because I think if everybody could learn to read equally, rapidly, that we would have a lot more intellectuals in the world. MR. LARSON: Of course reading is the foundation for almost every career. DR. ATANASOFF: Almost everything and I’ll pass in a hurry until, I won’t take up every step of my life until I was nine years old and my father had just got a job as an electrical engineer in a phosphate mining company which was just opening in Polk County, Florida. MR. LARSON: Oh yes. That was very early in the days of phosphate mining. DR. ATANASOFF: It really was. There were a few mines around, but not very many. This mine was just being started. There were no houses there, but they built a few houses and we took one of the houses and when I tell you, for 1913, it just comes back to my memory of that house and whatever I did I could see the surround of that house and I could tell it was during 1913 when I was nine and at the end of the year 10 years old. My dad failed being an electrical engineer because he didn’t know the slide rule. He really didn’t need one because the principal job he had was organizing the repair of the motors which were continually being burned out by the lightning. Lightning in Florida is about 100 times greater than it is in many other places. MR. LARSON: Oh yes. I have noticed that in my trips to Florida. DR. ATANASOFF: I don’t know, I think they have done some things which mitigate the effects now, but at that time it was terrible and Dad just kept his men hard at work rebuilding those motors. That’s what he was doing. So the slide rule was left to me. It was a Deci-Lon slide rule. MR. LARSON: Oh yes, a time honored name. DR. ATANASOFF: Yes it really was. It had a nice booklet that came with it and I could read. At that time I had read everything in sight. I had read, I remember that I read [Edmund] Spenser’s The Faerie Queene at that very early age. MR. LARSON: Yes, well, that’s very advanced. DR. ATANASOFF: I could read it and in a week or two I could do any ordinary problem on the slide rule. But this didn’t satisfy me. I had to understand how that slide rule worked and of course logarithms commenced to play a role. Now here I almost stopped baseball and I spent my whole time, we hadn’t commenced school yet because they didn’t have a school there at the mine. I studied, read, got Mother to help me at the age of nine. My mother was first class in algebra and read on and on and on until I understood logarithms, the meaning of logarithms. I memorized the definition for a logarithm. Then I learned how to use tables of binoms [binomials] by myself during that period and I thought I was content with life for a moment and then I realized that I really didn’t know how to calculate logarithms and I started on the project of calculating logarithms. I didn’t know who to go to and strangely enough Dad didn’t have much of a role in these events. MR. LARSON: Yes. He had other things. DR. ATANASOFF: I got a hold of a book and read about logarithms and I have the book here with me. MR. LARSON: That’s remarkable. DR. ATANASOFF: The very book. The very book. MR. LARSON: What is the title of the book? DR. ATANASOFF: The title of the book is College Algebra. MR. LARSON: Oh yes. DR. ATANASOFF: College Algebra has meant a lot of different things in different places. It’s quite an elaborate book and analysis. And gave the very subjects of algebra. There is a chapter in here on logarithms, then there is a chapter, a further beginning on the calculus in this book. I was, I read the elementary definition of logarithm and I was reaching into calculus and memorizing derivatives. MR. LARSON: That’s remarkable and this was all in a book on algebra. DR. ATANASOFF: I did all this in my ninth and tenth year. The definition of derivative is here and it’s inscribed on my mind. MR. LARSON: Yes. I believe you mentioned, do you have the name of the author of that book by chance. DR. ATANASOFF: The name of the author was Taylor. MR. LARSON: Taylor, yes. DR. ATANASOFF: Taylor was a professor of mathematics at Colgate University and this is a book that my father had used. This is his book. MR. LARSON: That’s remarkable. DR. ATANASOFF: I learned how to do infinite series and to test infinite series for termination and I actually calculated the logarithm of five to the base 10. MR. LARSON: Well that’s remarkable. DR. ATANASOFF: And that’s the time I left the book. MR. LARSON: Fine. That is really a wonderful story there. DR. ATANASOFF: And you know I had only had elementary mathematics up to that time. But my mother helped me with the algebra parts. There wasn’t much to learn, it didn’t seem to me. I had that head start of my mathematics and it’s followed all my years. MR. LARSON: Yes, well with regard to your work on calculating logarithms and so on, you certainly got into the theory of numbers which is so important in your chosen field. DR. ATANASOFF: I thought that if I read anything I could learn it and this has proven to be useful throughout my life. MR. LARSON: Fine. Very good. DR. ATANASOFF: I did a lot of other things. I studied physics and chemistry during this period and one other thing that I should perhaps mention is that Mother had a book. It had something to do with computing, you’ll see. Mother had a book on arithmetic and it had a chapter in it and it said, “Numbers to other bases than 10”. MR. LARSON: Oh yes. DR. ATANASOFF: That was the title of the chapter and I studied that during this year and when I, ever since when I came in contact with the need for numbers with bases other than 10, my memory from that furnished stuff. It was quite a complete treatment of that subject. I haven’t had to go to any book since then in regards to numbers and as well say later when I commence to wondering now will a, should a computer be built to a base of 10 or should it be built to some other base and I think I was the first one to go into that subject, thoroughly. I decided to use a base two. This is where my number theory came from. MR. LARSON: Yes. Well that was very fortunate that you had that base of number theory because without it also then the matter of applying electronics to the base two is certainly a lot better than the base three, or ten or anything else. DR. ATANASOFF: Sure. I, later on when I decided to investigate what numbers, I mean, what base should be used in, for computing. I went into the subject fairly, fairly overtly. You’ll find it in my bulletin there, my manuscript, without any question base two was overwhelmingly more advantageous than any other base. Now another thing that happened way back in 1913, the company had a Monroe in it’s office. MR. LARSON: Oh yes. At that early date? DR. ATANASOFF: Oh yes. I wondered if that could be so. My memory told me so, and it told me exactly how it was. My memory of the machine is in detail. I knew it was a Monroe, but when I got to writing about it, I didn’t want to make a mistake and I got a hold of someone that was connected to the Monroe company and he said they first commenced making Monroes in 1910. MR. LARSON: Well, that’s a very interesting point. That dates the availability that apparently Monroes and Marchants were used tremendously as the only asset that… DR. ATANASOFF: Those machines copied European machines, a similar kind. I didn’t have much contact with European machines, it was mostly Monroes and Marchants mainly that I used and the Monroe served us for many years in calculating and it was a very good machine and now, you could unhook the carriage, unhook it, you pull a couple of levers here and you could slip the carriage back and you could look inside and see how the mechanism worked and I did that almost immediately and I knew exactly how it worked and what was done. I was learning about computing machines clear back to that day. Now I’m not going to be so detailed in my life, I had, however, when I was in the tenth grade I decided, I remember again, I could tell you the shape of the room where I first got this idea. It was a room in the high school where I was attending high school and the, I had plenty of time, I didn’t have to work very hard. I had plenty of time and I was wondering what I would like to do. I decided right then that I would like to be a mathematical, theoretical physicist and that’s what I came to be. MR. LARSON: I think that explains your choice. DR. ATANASOFF: When, I’ll rush ahead to college. My father continued, he didn’t think that electrical engineering was right for me. He thought I ought to have chemistry. I like chemistry very much. I read a book on the subject at that time, a university textbook on the subject and knew it pretty well, but when I got to the University of Florida where I talked to the Dean of Engineering and he said the electrical engineering course is the most theoretical course on the campus. I decided I would take electrical engineering and did so. I had a lot of things that did me a lot of good. The machine work and stuff of that kind, the formulas and things with computing, but I didn’t have any electronics because electronics had started to be used in the universities of the United States, but only MIT and Cal Tech had courses at that time. I didn’t have a bit of work. I knew of vacuum tubes and I knew roughly how they worked at that time from my general reading, but I didn’t have any electronics. I sent out, as I neared the end of my year, 1925 when I graduated from the University of Florida in Electrical Engineering, I sent out some brochures and I got one from Iowa State College in Ames, Iowa. He just wrote faster than the other people. I later got one from Harvard, but I took, I promised the man at the University of Florida that I would take his scholarship and I did so and I went to the University of Florida and went to the Iowa State College in the fall of 1925. During my first year at Iowa State College, we were studying the theory of real variables, I was going into the mathematics department in order to have a theoretical background for my subject. I was studying the subject of real variables and there was a proof in there that had to do with the base two and I was the only man in the class that had ever had experience with base two. I remember that. Everybody else had to worry about that subject, and I knew all about it. MR. LARSON: That’s a fascinating story. Not to the base e, but to the base two. DR. ATANASOFF: Base two, yes. You can’t use e as a base very well because it has to be a whole number. It has to be a whole number as a base for a number system. It has to be a whole number and equal to two or greater, but anyway a whole number will do as a base. Well I can easily work out the proof. I majored in mathematics there and minored in physics and in 1929, I, in the middle of the spring semester, it was the end of a second quarter at Iowa State College. They had a quarter system there then. That was in the middle of the spring semester I went to the University of Wisconsin and took up residence there and started taking three courses. It kind of worried the professors that I was coming in in the middle of the second semester and starting to take three courses. Well, professor March knew I had an aptitude of various things and already knew I had some work in elasticity which he was teaching. One of the other professors was teaching quantum mechanics and I had had, I had just read something about quantum mechanics, but I had had it. Didn’t think I would make it, couldn’t think I would make it. This was Dr. J.H. Van Vleck, and I would ask a question and he would say, “We finished that last semester.” So he would push me down that way. MR. LARSON: Oh yes. Now quantum mechanics was still more or less in its infancy there. But really the big development was really about that time. DR. ATANASOFF: That’s exactly right. We had it direct there during the year. People come around them and say that we think that the professor ought to answer you because we don’t know the answer to that either, all the members of the class. He had 25 members in the class and finally we came up to the final examination and all but five of the members quit. They just resigned from the course. I was left with the other three or four, four or five, which ever it was, and wrote the examination. Professor Van Vleck doesn’t know what to say. He said, “Atanasoff,” he says, “you did pretty well on the examination,” he says, “I suppose I ought to tell you, you did the best than anyone on the exam.” MR. LARSON: Well that’s remarkable. DR. ATANASOFF: Afterwards he became a major professor. MR. LARSON: Oh yes. DR. ATANASOFF: However, I must tell a little about my thesis. The subject was the polarizability of helium where to take the helium atom and apply an electric field and see what electrical moment it had. MR. LARSON: Oh yes. DR. ATANASOFF: I understood the subject all right, but you see this wasn’t a simple atom. It had two electrons and it wasn’t easy to calculate mathematically. We used the wave functions that had been described for helium by Hilderhoff, a German physicist. I was able to convert these so they could be used for computing the polarizability of helium and Van Vleck was in Europe during the next year, but I went ahead with a professor who was from Zurich whose name escapes me for the moment. He was just as much help as Van Vleck would have been and I finished my thesis and got a Ph.D. in the middle of the summer of 1930, got my Ph.D. Afterwards I had left memories at Iowa State College, and I went back there to teach. MR. LARSON: What department? DR. ATANASOFF: I went back in the mathematics department, but after a few months they made me assistant professor of mathematics and physics, in both departments and then in a year or two they made me associate professor of mathematics and physics and I worked in both departments. Of course the mathematics department was an applied department being in a school at that time so that it was a pleasant period of my life. MR. LARSON: Well very good. So you went right into the academic profession. DR. ATANASOFF: I did that yes. I was there only a year or so and I commenced giving, teaching graduate courses in both mathematics and physics, and pretty soon I had post-graduate students who wanted to work with me. It was a wonderful period of life to be in, with just having had a Ph.D. and having students come around and asked. People liked my courses and wanted to work with me. I had as many students as I could handle. During the next six years or so I had eight post-graduate students and all but one got the Ph.D. MR. LARSON: Well there weren’t very many students or professors of theoretical physics at that time. DR. ATANASOFF: No, there weren’t. Now I was thinking about what, one thing I forgot to talk about, you know when I got out of school in 1909, in the summer of 1910, I had a very good teacher and I want to name her because she was extraordinary. Her name was Gertrude Arthur and she drove everybody to the absolute limit, including me. MR. LARSON: Oh yes. That’s the role of a good teacher. DR. ATANASOFF: That’s the role of a good teacher. When I worked on the polarizability of helium of course we had to use approximations on this subject and you couldn’t get the exact answer. You had to use approximations and my first ever; I got an accuracy of five percent, starting with an atom about which we knew very little at that time. I was able to nail it down and get an accuracy of five percent. Later I made a recalculation and got an accuracy of two percent. MR. LARSON: Well considering that you were delving into an absolutely new field there, that was very fine accuracy. DR. ATANASOFF: I was pretty enthusiastic about it. You know, about this time when I got back to Iowa State College I knew what I needed to know, that I hadn’t had. I hadn’t had electronics and I commenced myself into reading electronics. I read about the first book ever published on vacuum tubes by a man in Africa, in Africa. His name was Vanderweil. MR. LARSON: Oh yes. DR. ATANASOFF: Of course he was, he came from Holland. His family came from Holland. MR. LARSON: Oh yes. DR. ATANASOFF: And he wrote a book on which you will find in old libraries today, Vanderweil and I read a number of other books and I also started to work it with my fingers. You would think that I mostly sat and thought, but I had started a lab with my father and he always worked me into doing things with my own hands and I’ve always found it easy. As I stayed around Iowa State College, why, pretty quick I was directing experimental theses. It didn’t mean a thing to me. I could move into theoretical, into experimental work just as quickly as I could theoretical work. I had, I was going to tell you something about my theses were about, because that’s what I had to do. I got it here somewhere. MR. LARSON: Well, you’ve given us a good description of the work. I think that’s perfectly understandable. DR. ATANASOFF: I had one man who was doing graduate work in quantum mechanics. He was working on lithium. Other people were working on the theory of approximating the solutions of partial differential equations because you understand that you had to deal with the partial differential equation in theoretical physics and that’s what I had done with my own thesis and I have given it a lot of attention as to how to proceed with that. I commenced to study new methods, new methods for solving differential equations and I and my students who invented it, a rather new method, which I don’t know where it stands in the whole hierarchy in the methods used for solving partial differential equations today. I’ve got to find out from somebody. I’m going over… MR. LARSON: Yes, well partial differential equations are very essential in quantum theory and also advanced thermodynamics. DR. ATANASOFF: Oh, absolutely. You bet. MR. LARSON: They are absolutely necessary. DR. ATANASOFF: You bet… MR. LARSON: And not an easy discipline. DR. ATANASOFF: Thermodynamics was one of the subjects I taught in those days and I’ll tell you something else about it, but I better shut myself up for a moment. Let’s quit for a few. [Break in Video] DR. ATANASOFF: I was telling you what subjects my graduate students worked in: quantum mechanics, quartz, you know the crystal quartz. MR. LARSON: Oh yes. DR. ATANASOFF: We cut them and analyzed them and you know they have six coefficients of elasticity. It’s a complex crystalline substance and we, I provided the method for computing all these from measurements, and my students calculated, my students actually made these calculations. I believe to this very day the most complete collection of the elastic coefficients of quartz was done by my students and I. And now, you know me, when I did a thesis, I analyzed the partial differential equation by using the Rayleigh-Ritz process. You’ve heard of the Rayleigh-Ritz? MR. LARSON: Oh yes. DR. ATANASOFF: Well I invented a new method to replace Rayleigh-Ritz. And then I made it simpler so there was almost no mathematical work to be done. The only trouble was that as you attempted to solve a partial differential equation you had to solve a large system of linear algebraic equations at the end. And the problem with solving a large system, I’m going to talk a little about that subject, of course, later. Now, I studied whatever computers there were around and this included the IBM tabulator. MR. LARSON: Oh yes. DR. ATANASOFF: It was a great big machine and cost the university a lot of money and I felt that I ought to apply that equipment onto some of my own objectives. I got a man in statistics named Brant, Dr. A.E. Brant to help me and thought we would find a way of, a physics problem that we could solve by using a tabulator. MR. LARSON: Oh yes. It used the IBM punch cards? DR. ATANASOFF: Exactly. It used IBM punch cards. Well I thought maybe the thing to do was to use it to solve complex spectrum. And then I analyzed it further and then I found out that it wouldn’t do it. It was a dumb machine. It wasn’t very smart and then I thought maybe I could add some additional equipment to the IBM tabulator and make it do it and we did so and published a paper. MR. LARSON: Well that’s remarkable. DR. ATANASOFF: And the, strangely enough IBM was not very happy at all. We said nice things about IBM, but they weren’t very happy at all and years later the courts forced IBM to draw every piece of paper that had Atanasoff on it out. I found out that there is a piece of paper in IBM that says, “Keep Atanasoff out of the tabulator.” MR. LARSON: That’s fascinating. DR. ATANASOFF: We didn’t hurt the tabulator at all and in the end it worked as well as it ever did. MR. LARSON: Of course that was a good introduction into your beginning to think about computers then. DR. ATANASOFF: Perhaps I should say that IBM wouldn’t give me a copy of the internal drawings of the computer and I just sat down and figured out how I would have done it and proceeded accordingly and it worked fine. MR. LARSON: Well wonderful. DR. ATANASOFF: Now we’ve got this, my students had all these linear algebraic equations, just ordinary linear algebraic equations, but they have eight or ten or more unknowns in them. MR. LARSON: That means about a week of calculating. DR. ATANASOFF: Oh yes, but pretty quick it was several weeks. My students spent several weeks on those calculations and we knew that what we needed to do was to get the results more accurate, we had to double or triple the number of unknowns in them and that looked as though it would take most of a lifetime to do. MR. LARSON: You didn’t use determinants to solve them. DR. ATANASOFF: It wouldn’t help. MR. LARSON: That would get so complicated. DR. ATANASOFF: No, no, the determinates are fine in doing a theory of such problems, but had nothing to do with solving them. Solving them, you just have to do the numerical work. And my students were making a noise about this subject, and this is the greatest impulse that I ever had towards computing. I commenced to examine all the computers, all the methods of calculating that were available. It took months, but in the end there was nothing. MR. LARSON: Yes, well there were the Marshant calculators and the IBM tabulators, that was about all. DR. ATANASOFF: I tried to us an IBM tabulator, but it didn’t have the capacity, didn’t nearly have the capacity. It was slow too. And Marshant, you know how long that takes and I used that in my own calculations of helium. Oh, I commenced to wondering, to think about analog versus digital computers. I commenced to wondering if I should go into analog computers for solving this. A little analysis showed me that an analog computer, you know [Vannevar] Bush developed an analog computer called a differential analyzer. MR. LARSON: Yes, that’s right. I believe that was the most advanced calculator as of that particular moment. DR. ATANASOFF: That’s right and he spent a $1.5 million on that computer and he got this money from the Rockefeller Foundation. Warren Weaver gave him the money. Do you know Warren Weaver? MR. LARSON: I met him. As a matter of fact, Jane met him at your home. DR. ATANASOFF: Warren Weaver was a major professor of mine in college. MR. LARSON: Oh yes. DR. ATANASOFF: And I’ve had some associations with him since. MR. LARSON: Well that, of course, that work that was concentrated in analog computers, I think probably held MIT back a little bit from going into the digital. DR. ATANASOFF: Oh yes. You know why? I’ve argued with Bush and his assistant, Sam Caldwell. Do you know Sam Caldwell? MR. LARSON: No, I’ve heard that name before. DR. ATANASOFF: Sam Caldwell, I argued with them both. At this time I was getting sharper on the subject. I said, you know analog was my own invention. MR. LARSON: Is that right? DR. ATANASOFF: It was my own invention. MR. LARSON: What did Bush call it? DR. ATANASOFF: He called it differential analyzer. MR. LARSON: A differential analyzer. Analog is a… DR. ATANASOFF: You see he hadn’t sharped up the difference between an analog machine and a digital machine. The word “digital” was not of my making, it wasn’t my word. I called it computing machines proper. MR. LARSON: Oh yes. DR. ATANASOFF: I called a calculator a computing machine proper. MR. LARSON: Oh yes. DR. ATANASOFF: It didn’t use an analog; it went directly at it in a mathematical sense. That was my word and other people called it and I missed my word. That’s somebody else’s word. But the word analog came out of my mouth. It’s kind of amazing isn’t it and I think I can prove it. Now, I’ve argued, I was in contact with Bush and the last time I saw him was in ‘16 and… it was down in Washington. Can’t recall the place. Well, I looked the thing over and I knew that I had to build a computer and I was shaken that that was what my future was. I wasn’t enthusiastic about it. I didn’t want to do it, but I knew I had to and I knew I had the means. Somewhere I felt that I had the internal means to do that. So I started in to examine the whole subject of calculation. Now what did I do? I just sat down at a desk and just started making pictures and thinking. Making pictures and thinking. This went on for months and months, making… and I worked on a number of subjects and that included the subject of memory that you spoke about. I knew I had to have a relatively fast memory. I set my goal as to provide something that was several times as fast as an IMB tabulator. MR. LARSON: Which was all mechanical. DR. ATANASOFF: Which was all mechanical. That’s what I set as my model. I didn’t think about working faster. As a matter of fact I made my computer machine way too slow because all the means I introduced to the computer were very rapid. One of the methods that I induced for computing was in memory. Now I knew that mathematically speaking that depending on what base I was going to use, it would change the method of memory that I would probably want to use. I could see that. So on the subject of base came to the floor. So I decided I would try to find out what base to use first. I decided, you know I’ve already said I decided to use digital machines, what we call digital machines, which I call computing machines proper. Well when I examined the base I have a mathematical theory which is contained in here and I presume it isn’t very good to try to give the theory verbally, but the theory showed me that if you’re doing ordinary arithmetic, the lower the base, the better. The lowest base you can use is two. MR. LARSON: Oh yes. DR. ATANASOFF: One of my calculations showed me I should use the base e, but it turns out that that was a mistake because it turns out… MR. LARSON: It has to be a whole number. DR. ATANASOFF: No. MR. LARSON: No? DR. ATANASOFF: It theoretically gave me the base e, I knew I could use it as base, so I had to use either two or three and I examined which would be better at that stage, but it turns out that there was an error in the theory that gave me the base e. When it comes to multiplication, the base two is best. The base two is really the best, better than e. Better than three. When it comes to division, theoretically from a theoretical point of view, you could use a base of three or a base of four, but there are a lot of other things that enter this field and when it comes to a question of memory, either you punch or you don’t punch, that’s two. If you attempt to apply that to a base e, I mean base three or any other base, it causes a complication of like the base 10 card of IBM’s. It’s something like that. You know how that worked? MR. LARSON: Oh yes. DR. ATANASOFF: Well that’s an illustration of a man using a punch or no punch problem which really is quite suitable for base two. You try to use it for base 10 and it uses up more card than is proper. So, the base, we are driven again the base two. When it came to the logic system, which I will talk about later, the base two is overwhelmingly better. When it came to memory, it was much better and so I was pretty much driven to the base two. When we’re attempting to get a memory for base two, you either punch or you don’t punch, you know, you can, you know if you use a relay computer it automatically works out for base two. MR. LARSON: Relays and diodes… DR. ATANASOFF: That’s right; however, I knew that the cheapest thing I could use would be a condenser. MR. LARSON: Oh yes. DR. ATANASOFF: So I commenced to use condensers and I used .003 micro [inaudible], a very small condenser. I didn’t go anywhere near the bottom. I could have used much smaller condensers. I did use as a matter of fact. The time before, let’s talk about the condensers that I did use. It would take 40, 50 seconds for the signal to die off in a condenser of that size and now… MR. LARSON: What kind, were these coil or electrolytic? DR. ATANASOFF: They were on paper. Any ordinary paper condenser would work. I didn’t have any kind of trouble with any kind of condenser. Everything worked, of course the, I knew that this was really complicated because the condensers would die off and I had to find some way of keeping them from dying off. Now my original purpose would be to use a string or something of the kind that could snap into one of two positions, have a little potential here, then higher and then a little potential here and you could snap it from that position to this position, or that position to that position. And change from zero to one and one to zero. That requires a system with two minimums of potential energy. You see what I’m drawing. MR. LARSON: Oh sure. DR. ATANASOFF: They, that was what I first thought I would do. And then I was driven by the little boy going to the grocery store. The little boy going to the grocery store… MR. LARSON: Your condensers were… DR. ATANASOFF: Like the little boy, it peters out. Runs down. The little boy’s memory runs down and he doesn’t remember what his mother told him to do. So once the little boy goes, as he starts to the grocery store, he repeats what his mother told him. And he repeats it at such frequent intervals that he doesn’t have time to forget between times. MR. LARSON: Oh yes. DR. ATANASOFF: Having had this practical idea in my mind I immediately commenced to put this into electronic shape and I milked condensers into circuits that had this property. At frequent intervals they would test what they were. And put them into the most desirable state, pull themselves up. MR. LARSON: So you have to, you had to read what was… DR. ATANASOFF: In it. MR. LARSON: …on the condenser at the same time… DR. ATANASOFF: Regenerate. MR. LARSON: …regenerate the charge because it went down. DR. ATANASOFF: You got it. MR. LARSON: You had to do that within the 50 seconds. DR. ATANASOFF: Oh yes. As a matter of fact I did it at one second intervals, but our present computers, do exactly the same thing. They have a little condenser in there and it will wash out, but they have refresh cycles which will continue to flip it through the system and are bringing them back to the original state. MR. LARSON: I didn’t know that. DR. ATANASOFF: That is the way they are made. That is true of all memories in modern computers. I’m telling you something out of place a little bit. You know the magnetic memories wouldn’t die off because of time. When you, I could say that if you reach in and take out the charge, why then you have to recharge it in that state in order to have it retain its original value. That’s a complex situation on magnetic, on magnetic cores. Magnetic cores were like that. Well. Now, I worked on a lot of things here, which could possibly be, for months and months, I guess for a year and half to two years, I was working on just little giblets of elements that could be put into a computer. I hadn’t in mind how I could build a computer. I didn’t have in mind how I would ever reach my goal. I went, I’ll tell you what happened. I went out to my office one evening, in the middle of the winter of 1937, ’38, and I sat down at my desk and attempted all my stuff around. Tried to work on this, tried to work on that. I became extremely irate with myself and I was upset to an extreme degree, then I did something that I have only done two or three times in my life. I went out and got into my car and started to drive. It was 20 degrees below zero, but there wasn’t a bit of ice on the pavement. It was all dry and I drove hard for several hours. Driving hard enough so I couldn’t think about computers. I had to think about the road and I was able to mitigate my unhappiness that way and at last I said, “Atanasoff, this has gone far enough. You have got to do something.” And then I glanced over and I was crossing a bridge 189 miles from my starting point. MR. LARSON: Wow, you really traveled. DR. ATANASOFF: I traveled at 80, 90 miles an hour in a Ford V-8 of the era and I had a good heater in there and I was comfortable. It was extremely cold, 20 degrees below zero. Somebody analyzed my situation and they said I was looking for a drink. Well I wasn’t looking for a drink, until I glanced over and saw the water. Then I said, while you’re going to Illinois, of course you could get a drink in Illinois. I went on and there was a honkytonk on the side of the road within a quarter of a mile, there was a honkytonk in Illinois and I drove in. I parked my car and walked in. I had a heavy coat on, very heavy coat. I remember the strength I used to raise it to the hook. It was plenty heavy on my hand as I raised it to the hook and I sat down and ordered a drink. All of the sudden I realized that my mental capacity had improved remarkably, that I could think things through. I knew immediately I could think things through and I knew immediately what I had to do. I sat right down and started working on the computer and I worked on it continuously for three hours. People say did you have any paper? I don’t remember any. I didn’t need any paper. Everything I did I remembered perfectly. And I got up and I drove home at a slower rate. I had come to four conclusions while I sat there in the honkytonk. MR. LARSON: That’s a remarkable story there. What were those conclusions? DR. ATANASOFF: I ought to find it here because you know it’s like me to forget one of them. Yes, in the first place you might not think this is much of a conclusion, but I decided I would not do a mechanical system, but an electrical system. This may not seem as though it was a great step forward. I had some letters from the members of the IBM staff and a member of the IBM staff told me they would never build an electrical computer, never build an electric computer. In spite of custom, I would use base two. I was going to translate the problems that I had with base ten, do the calculations with base two and translate it back. I would use compositors for memory, but I was going to use what I told you about the little boy. There was going to be a mechanism there that would test what it was at frequent intervals and restore it to its original form. I would, this last one came originally from these other things that I had thought about before, but I was finalizing it in my mind as I was sitting there. I would compute by logical action and not by numeration. MR. LARSON: Oh yes. DR. ATANASOFF: I was going to build a logic system into my computer, I didn’t know how to do it, I didn’t have the smallest idea, but here’s a box and it has leads going in and leads coming out and into it you put the initial quantities and out would come the final quantities and the machine was a logic based computer. MR. LARSON: Oh yes. DR. ATANASOFF: And now, when I got home I had these four items and I commenced examining each of them in detail. This went on not merely for a month, it went on for a number of months because I could do parts of them, parts of them I already had answers for, but when it came to the logic system, I had to start from the beginning and build a logic system and that took me months. As I look back over that era, I realize that these four, four steps that entered my mind at the honkytonk are in every computer today. MR. LARSON: That’s right. They are pretty fundamental things. They have to be. DR. ATANASOFF: Pretty fundamental and now I was delighted when I found that out. I didn’t find it out until two or three years ago that every computer, every one of them as long as they have magnetic cores, that doesn’t quite fill the bill, it would be off by one, but as far as all the rest go. The dynamic memory as low capacity and the refresh cycle is exactly according to my principal. MR. LARSON: Oh yes. That’s amazing. DR. ATANASOFF: I then knew when I had gone through these consequences and my trip to the roadhouse had actually, now can you imagine me sitting down and drawing out vacuum tubes without ever putting a vacuum tube to a circuit. I finally got it to where it would work and that was the state it was in. I hadn’t ever tried it mechanically. I hadn’t. I could have very easily. I was good at the arch, but I didn’t feel I needed to because I knew exactly how those vacuum tubes would work with that circuit and so all I had to do was sort of draw out the circuit and plant it. Then I decided I couldn’t go any further by myself. I had to have some help. I commenced approaching at Iowa State University, College at that time. I was walking across the campus and I saw a member of the electrical engineering department named Harold Anderson. I can show you within a foot of where I stood at that time. I said to Harold, “This is what I’m up against. I’ve got to have some help. Can you make any suggestions?” Harold Anderson took a minute. Always a very careful man. He took a minute, maybe a minute and a half, thinking it over and he said, “I’ve got your very man.” He said, “His name is Clifford Berry.” MR. LARSON: Oh yes. And that’s how you got Clifford Berry. DR. ATANASOFF: That’s how I got Clifford Berry. I suppose we had to ask Clifford if he would agree, but there wasn’t any problem. I had Clifford Berry. And that was in the spring of 1939, I decided I could employ Clifford Berry and he couldn’t start work until fall. He started work in September and before the end of the year, October, November, December, he had the circuits working. MR. LARSON: He really put them together fast. DR. ATANASOFF: Not an electrical computing machine, just the rudiments of a computing machine. When he showed that my precepts of how it should go together were right. MR. LARSON: That’s remarkable. So then you actually had a bread board, an arrangement there. DR. ATANASOFF: A bread board. We have a picture of it in here. MR. LARSON: Yes. DR. ATANASOFF: This picture of a bread board in here. MR. LARSON: Fine. I was wondering if you could open it up to that picture and then hold it against your chest and we’ll zero in on it. DR. ATANASOFF: Can you zero in on it? MR. LARSON: Oh yes. DR. ATANASOFF: I can do that. MR. LARSON: Zoom right in. DR. ATANASOFF: Here it is. MR. LARSON: Oh yes. Sure. Oh, very good. DR. ATANASOFF: Can you see it alright? MR. LARSON: We got a very good record of it. MRS. LARSON: Yes. DR. ATANASOFF: Now that was, we didn’t have a picture of it. This is my memory and it was done by an artist who put things together, but he did it according to my memory. Really this is what a computing machine is about. MR. LARSON: Oh yes. And you’ve got all those elements put together there. DR. ATANASOFF: I was able to actually assemble them at that stage. MR. LARSON: Yes and you were certainly able to test the various components of them. DR. ATANASOFF: As far as putting a number in there, you just had to touch a wire to put a number in. it would only do that. I didn’t build any equipment to put numbers in or take numbers out. And then when I wanted to see what was in there I had to test the condensers by means of a volt meter or a vacuum tube volt meter or something. That’s all it took at that stage. MR. LARSON: That’s a remarkable story then. Fine. Then what was your next step after you had done all these and tested and so forth. DR. ATANASOFF: You know my purpose was to make a machine solve system, linear solve system. MR. LARSON: Oh yes. DR. ATANASOFF: And that was to build a machine specifically for that purpose. MR. LARSON: Oh yes. DR. ATANASOFF: And we did that. MR. LARSON: Oh yes. DR. ATANASOFF: And we, I should also, as long as you’re looking at the picture of that machine, why, we ought to, here is the circuits that did the jogging. You ought to take, make a… MR. LARSON: Oh yes. Zero in on that as far as you can. MRS. LARSON: I got it. MR. LARSON: Fine. Good. That’s very readable. DR. ATANASOFF: I had to build a logic circuit to go on that. And here’s the logic circuit. This is exactly the way I drew it before I even had Clifford Berry there, this is exactly, I drew that. MR. LARSON: It’s very complex. DR. ATANASOFF: It’s complex and this is for the base two and if you attempted to build a logic circuit for the base five, it’s much more complicated. MR. LARSON: You’d soon run out of vacuum tubes. DR. ATANASOFF: Right, you’d soon run out of vacuum tubes and this one took 14 trials. Thirteen trials I think it was. I got 14 here in the picture, but my trials came two in a bottle so I had to have 14. Over on the next page is some table which shows how the thing works. I have had to try to put together for a core procedure, a treatment of this subject. Did you know that such circuits are put together today by using complex algebra which is called, what? You know what it is. MR. LARSON: Boolean algebra. DR. ATANASOFF: Boolean algebra, that’s right. You know I already studied Boolean algebra but it didn’t help me in that. MR. LARSON: Oh I see. DR. ATANASOFF: I just didn’t make the connection. MR. LARSON: I guess the use of Boolean algebra came a little later, but it was such a valuable tool, once that is, in fact used today. DR. ATANASOFF: And it’s the very thing that I was perfectly capable of doing, but I just didn’t do it. I… Clifford… MR. LARSON: So Berry worked right with you there. DR. ATANASOFF: Yes, he did. You understand we were on a meager budget. We didn’t have any money and here we are doing it. Nobody else had put a computer together like we did. MR. LARSON: So you then built the machine to use the computer principles for the specific purpose of solving these linear equations. DR. ATANASOFF: The lawyer said that is a special purpose machine and should not be considered with the general purpose machines, with which you are comparing it. My opposing lawyer was saying that to me. I’ve got a story to tell you about that. Maveridge worked a good many years on computers. MR. LARSON: Oh yes. DR. ATANASOFF: He didn’t build much of anything. He just did what I did, that is he made pictures. And after he was dead, the parliament wondered if they hadn’t mistreated him. They knew he was a great man and they wondered if they hadn’t mistreated him. So they got together a committee to decide, at that date if they shouldn’t build a Maveridge Machine. MR. LARSON: Oh yes. DR. ATANASOFF: So that committee droned on and finally decided that the drawings were not sufficiently accurate so that they could build a machine. It wasn’t like my machine which was in vacuum tubes that Clifford could build. It was something more complicated. MR. LARSON: A very complicated gear system. DR. ATANASOFF: Then the following meeting took place. Someone rose and said, “If that machine would only solve linear algebraic equations we ought to build it because those, that subject is the greatest importance to man.” MR. LARSON: Oh yes. DR. ATANASOFF: They agreed on that subject, but they couldn’t see how the Maveridge Machine would ever solve linear algebraic equations. MR. LARSON: Oh yes. DR. ATANASOFF: But I immediately turned my machine the moment I had a grasp of how it would go together. I immediately turned it in that direction. MR. LARSON: Oh yes. DR. ATANASOFF: I didn’t even know this story at the time I did it. I just blundered in. This is what I’m going to tell you next. Let’s cut off. [Break in Video] WOMAN OFF-CAMERA: Did you take the time? DR. ATANASOFF: No, I’m alright. MR. LARSON: Well to continue this, you’ve brought up some very fascinating points and so if you would just expand on that I think it great. DR. ATANASOFF: Do you realize that motivation, my motivation to study computing machines was to solve large systems of linear algebraic equations? This is like X plus two Y equals seven, X minus three Y equals 15. That’s two unknowns in two equations, but when we go to the place where we have 10 equations and 10 unknowns and there on up, we have something that can’t really be done to man’s satisfaction. It takes too long. I, we got to have the use of machinery. I was aiming in that direction, aiming at the desire of the people who investigated Maveridge that we would have a machine to do such processes. This is a kind of complex part of my story. The first place, we will just say a few words about linear algebraic equations. You know what we ordinarily do in solving such systems is to use the equations which we have in pairs and eliminate one unknown between a pair. We can always do that. Now it consists of this, you have a pair of equations by a proper combination of adding or subtracting the coefficients and the constant term in a specific way it will eliminate one of the unknowns. MR. LARSON: Yes. DR. ATANASOFF: That’s the process. MR. LARSON: That’s the standard process. Works very well until you get to three or more, it gets very laborious. DR. ATANASOFF: Very tiresome. Now I guess the main part of it, you know way back I thought I would attempt to solve this problem, I didn’t mention it before, but I thought I would solve it by using, say, 15 or 20 Monroes. They all have a common axis. I geared them all on a common axis. So they all turned together. That looked as though it was a possible approach to this subject, I could have done it, but the problem with getting the numbers into and out of the Monroes was too complex and also had to be recognized before we could get anywhere. Now, I suppose we had a pair of equations and the initial coefficient was two in one case and three on the other. We would multiply the first one by three and the second one by two and subtract them and the first equation, the coefficient on the first term would become zero and it would become eliminated. We would have the problem solved. MR. LARSON: Yes. Well that’s very simple for two equations and two unknowns. DR. ATANASOFF: Yes, it is. However, let me say that you could take any linear combination of any two equations and it would have the same answers as the original system. MR. LARSON: Oh yes. DR. ATANASOFF: Any linear combination that you please. I had to try to bring that to realization. Now, you know in computing, we all know that multiplication requires successive additions and changes of the, changes of the power to which we have to raise 10 for which we raise 10 numbers to get that particular element. I spent a long time thinking how I could do it. Strangely enough I came to the following answer. I would do it by division. MR. LARSON: Oh yes. DR. ATANASOFF: By division. I would take the first equation and divide it by the second one and as I did the division what would also be left over would also be an equation. If you take all the way across the array of coefficients, would also be an equation with the same coefficients, divided properly, the division would go to zero and hence would eliminate one unknown. MR. LARSON: Oh yes. DR. ATANASOFF: You’re understanding it perfectly and I did this yesterday. I applied this over again yesterday as to how I would say it. I’m succeeding with you and that pleases me immensely. MR. LARSON: Well that’s fine. DR. ATANASOFF: Now I ought to tell you this, that ordinarily in computing, you subtract the divisor off of the dividend until the dividend carries through. You know, you’ve gone one step too far. You add it back in until it was just before it was carried through and then you would shift to a different power of 10. Do you see what I’m driving at? MR. LARSON: Yes. DR. ATANASOFF: Then you would subtract and subtract until it would carry through. Then you would add it back in until it would, now this process is what is called ordinary division with computers. Ordinary division with computers. I decided that this process was too complicated. I invented a new process and this was an invention of mine. I would subtract until it carried through and then I would shift, change to a different power of 10 and then instead of adding I would, I mean instead of subtracting I would add until it carried through and with this series of numbers I could get the same result as the former one. This is called division, Alice, what is this division… my wife’s got the words. MR. LARSON: Well, never mind, it will come. DR. ATANASOFF: I’ll get it in a minute. Now do you know these two methods are used in computation? I invented them and they are available on those machines today. You can use either system of division that you please. MR. LARSON: Oh yes. That’s actually built into the computer mechanism. DR. ATANASOFF: Yeah, built into the computer mechanism. I could do it with, let’s see, we have some words for that. Words escape me. What are you going to do? Turn it off and I’ll have it in a minute. [Break in Video] DR. ATANASOFF: I’ll repeat that. Where you ordinarily would subtract until it’s carried through and then adding back, that’s called restoring. MR. LARSON: Oh yes. DR. ATANASOFF: That’s called restoring division. And ordinary computers today will do that if you wish. They will also do the other which I invented and consists of, instead of restoring it, you leave everything just as it was, but the next time you add until it carries through. The next time you subtract until it carries through and this is called non-restoring division and it was non-restoring division that I invented. MR. LARSON: Yes, and those are built into the so called read only memory in most computers then. DR. ATANASOFF: That’s right. So in order to save time, I use a second process which saves, well it doesn’t save so much with base 10 because you only save one step in the five steps, but if you’re using base two it saves half your time because you do it so much. It saves half your time if you use restoring division. MR. LARSON: That’s very interesting. With base 10 it’s not so important, but with base two it’s very essential. DR. ATANASOFF: Very effective, yes. and I invented non-restoring division. Alright now I arranged the new computer so that it would have 30 peels. I couldn’t work 30 equations with 30 unknowns. I could only work to 29 equations and 29 unknowns because it had to have some place for the constant term. And in the 30 peels, one place would be for the constant term and the other for all the other coefficients. MR. LARSON: Oh yes. DR. ATANASOFF: So I arranged two drums. Each which had 30 peels and that would, could be used up to 29 equations and 29 unknowns. You put one equation in one drum and another equation in another drum and connect the two drums with logic systems which I have previously given and you could just add, add the, lets say one of the drums is called a keyboard and I needed a noun and I called it a keyboard imackus. The other one is called a counting imackus and those were the two names of them and I’ll just use those names here. And all I had to do was rearrange so that it commenced to divide a pair, say the first coefficient of each equation, divided the two of them, until that coefficient was dropped to zero. When that coefficient dropped to zero, you know the remainder in division after you’ve carried it out long enough to drop to zero. MR. LARSON: Well this is a very good explanation. DR. ATANASOFF: Gosh, that’s wonderful. MR. LARSON: Incidentally I am just amazed such an operation of 29 equations and 29 unknowns. As I visualize it that would take a person a lifetime. DR. ATANASOFF: It would take a major part of a lifetime to work it out and be sure everything was right. MR. LARSON: So that’s an amazing new tool. DR. ATANASOFF: That’s what I was aiming at. I built this machine and I can show the picture here. There is the machine in its last photograph. I will use this photograph to tell something about it. [Break in Video] DR. ATANASOFF: …picture that was taken and I have the negative of this. These pictures came from various sources, but this particular picture, I have the negative. MRS. LARSON: Wait a second. MR. LARSON: We would like to see that. MRS. LARSON: Hold it up slightly. MR. LARSON: That’s great. We’ve got a good picture of it now. Fine. DR. ATANASOFF: Now this is the principal parts of the machine. Here is one of the drums and the other drum was inside this case. Here is a card reader. We put the numbers into the machine in base 10 and it was taken over by the card reader and then the machine converted to base two and I had to build a special system from changing from base 10 to base two. And it worked in either direction. Now it kept the numbers for the given system of equations, it kept the numbers on a scratchpad memory. This is the scratchpad memory. One of these things rights the memory on a card, punching it on a card and the other one reads it. MR. LARSON: Oh yes. DR. ATANASOFF: So that way we keep track of the numbers as we pass through the entire process of solving it. MR. LARSON: That’s very ingenious. DR. ATANASOFF: This, I had to invent everything. I couldn’t buy anything. Nothing was available. So I invented the scratchpad memory and the scratchpad memory consisted of having a card pass below a series of terminals and I put a high spark discharge through paper and this made the record. Clifford was very interested in this and worked on that. We thought we had this scratchpad memory perfected. We tested it, but when we finally came to the final machine, it was correct, all except for one part, one place in 10 to the fourth or 10 to the fifth. It said there was an error in the scratchpad memory. As far as I know this was the only error that existed in the final machine. MR. LARSON: Oh yes. DR. ATANASOFF: Now, in other words, it recorded all the coefficients in base two on one of these and picks it up in the other, put it back in the machine and kept that procedure. This is the way the machine worked. MR. LARSON: That’s remarkable. So you then could feed these things, convert the decimal to binary, perform the calculations… DR. ATANASOFF: After you converted it, it came on a base two card and you just put base two cards in until the equations were solved. MR. LARSON: Oh yes. That’s a very interesting story there. DR. ATANASOFF: Now, let’s see. Here is a picture that may portray some… MR. LARSON: That gives you a very good… DR. ATANASOFF: You might like to take this picture and add it… MR. LARSON: We got that. DR. ATANASOFF: This was made from my memory by an artist also. MR. LARSON: Oh yes. Fine. DR. ATANASOFF: I will say this concludes the mechanical operations of building the machine. MR. LARSON: Fine. Very good. [Break in video] DR. ATANASOFF: …in the course of building this machine, money got to be more and more a problem. So I decided I would write a description of the machine and this was as of that day. This is that description. MR. LARSON: Oh yes. [Break in Video] MR. LARSON: Let’s change the tape here. It will just take a second. MRS. ATANASOFF: Does that mean you have an hour? MR. LARSON: Yes. MRS. LARSON: It’s still going. MR. LARSON: I just wanted to make sure. MRS. LARSON: I looks like it’s still going. MR. LARSON: Let’s look and see. MRS. LARSON: Oh, it’s still on. [Break in Video] MR. LARSON: Go ahead and we just put in the next tape. DR. ATANASOFF: In 1954, Alice, was it ’54 or ’56? MRS. ATANASOFF: ’54. DR. ATANASOFF:’54. I was visiting, oh by this time I had left the Naval Ordinance Laboratory and started a company of my own. MR. LARSON: Oh yes. But in the meantime, let’s say that Dr. [John] Mauchly had left the Naval Ordinance to go to… DR. ATANASOFF: Oh you know he only worked temporarily, once a week for Naval Ordinance. MR. LARSON: Oh I see. He was only down there for a short period of time. DR. ATANASOFF: I’m sorry. MR. LARSON: I wanted to clarify that. DR. ATANASOFF: Yeah, he was working for the university up at Philadelphia, all the time. MR. LARSON: At the University of Pennsylvania. DR. ATANASOFF: I have discussed this matter with Arthur Burks. Do you know who he is? MR. LARSON: Oh yes. I’ve heard of him. DR. ATANASOFF: Well, Dr. Burks was there at Philadelphia with Dr. Mauchly at the University of Pennsylvania. And he can’t quite understand why Mauchly is coming down to see me either. It’s a very tricky spot. But I got a visit one morning, and I had started a company of my own and I got, a patent attorney from IBM came by and wanted to talk to me. MR. LARSON: This was about 1954, you said. DR. ATANASOFF:’54, yes. And he says, he had found out about Mauchly’s patents, and he says, “If you help us, we will destroy the Mauchly patents because they were derived from you.” Those were his direct words. MR. LARSON: So his patents were directly derived from yours. DR. ATANASOFF: That’s what he said. That’s what he said. I wasn’t quite sure how I should react. I had myself formulated that opinion. It was him speaking. I said I can’t. I can help you, but I can’t work full time with you because my staff and I have put all the money we got into this company and we got to make a success of it. MR. LARSON: Oh yes. DR. ATANASOFF: And he wrote me one letter and I answered him. He wanted to get my original notes. I didn’t know where they were. The notes that contained this. I didn’t know where they were at the moment. I had been confused and moved here and there and had stuff… MR. LARSON: I know how that works. When you move, you lose a lot of things. DR. ATANASOFF: That’s right and it was all…okay. So they abandoned me and they proceeded to pay Mauchly and [J. Presper] Eckert a sum of money to secure the rights from them. This is IBM. IBM got its rights in that way. MR. LARSON: Oh yes. DR. ATANASOFF: Then we’ll pass onto the time that I, the Mauchly and Eckert patents had been transferred to… MR. LARSON: Sperry-Rand. DR. ATANASOFF: Sperry-Rand. I’m trying to say Sperry-Rand. MR. LARSON: And well however, the patent did not come out until quite late as I… DR. ATANASOFF: Very late. MR. LARSON:’64, or something like that. DR. ATANASOFF: That’s right. I’m sorry I can’t give you the exact date, but it was very late just as you say. I didn’t memorize those. I have to memorize those. Sperry-Rand, a good deal of water had gone through the dam. MR. LARSON: Oh yes. DR. ATANASOFF: And Sperry-Rand was attempting to collect a billion dollars in royalties. MR. LARSON: A nice round sum. DR. ATANASOFF: Yes. A nice round sum. Well, two of the people, they selected two companies they didn’t think had much power, that they could beat down and collect the royalties. That was Control Data and Honeywell. Now both Control Data and Honeywell were willing to go a little bit higher than the nominal patent fees, but they weren’t willing to go any further and they rebelled. MR. LARSON: Oh yes. DR. ATANASOFF: An attorney named, what is it? MRS. ATANASOFF: Kirkpatrick. DR. ATANASOFF: Kirkpatrick for Control Data had read a book, the first book on the patents of the era by… MRS. ATANASOFF: Richards. DR. ATANASOFF: Who? MRS. ATANASOFF: Richards. DR. ATANASOFF: Oh yes. By Richards. So he found out immediately where I was and came up to see me. So he talked to me and introduced me to the people at Honeywell and the people at Honeywell were talking to me and pretty soon we were shaking the woods. MR. LARSON: Oh yes. DR. ATANASOFF: The thing ended up by my agreeing to act as witness for the two companies. MR. LARSON: So it wasn’t a joint action on the part of Honeywell? DR. ATANASOFF: No, it wasn’t. Separate actions. MR. LARSON: Separate actions. DR. ATANASOFF: During this period, I heard from Mauchly. I hadn’t heard from him for a good while. MR. LARSON: Oh yes. DR. ATANASOFF: I heard from Mauchly. Mauchly says I would like to come and see you and I would like to bring one of my patent attorneys in to see you. Then, I won’t go through the details, but he came down and he brought the patent attorney with him and we talked for a while and we had lunch. Alice prepared us a lunch and then we talked somemore. The name of the patent attorney, his name was Dobbs. Dobbs was a patent attorney for both Honeywell, no, he was an attorney for… MR. LARSON: Sperry-Rand? MRS. ATANASOFF: IBM. DR. ATANASOFF: No, not IBM. MRS. ATANASOFF: Sperry-Rand. DR. ATANASOFF: Sperry-Rand. He was a patent attorney for Sperry-Rand in both of these cases is what I’m trying to say. So, in the course of discussion, by this time I had read a good deal of more material and I had the original patents. I told Dobbs he couldn’t possibly win. I just flatly told Dobbs he couldn’t possibly win. You know it’s very strange. Most people would have been irate and offensive, but Dobbs wasn’t irate or offensive at all and he became a firm friend and is today. MR. LARSON: Well that’s a very interesting point there. DR. ATANASOFF: He, I explained exactly why he couldn’t win and he didn’t win for those exact reasons. MR. LARSON: He must have been a little devastated seeing this billion dollars sort of blow away. DR. ATANASOFF: Oh yes. He was just as genuinely nice to me as he could be and he wasn’t regarded as a very gentle man, but he was in my case and the last time I entered into a court procedure he insisted that Alice and I meet his wife after the court was over. MR. LARSON: Oh yes. DR. ATANASOFF: To just show how much he had become my friend, the result of having this conversation with him. He found out I was exactly right. You could talk about the judge not being smart. He didn’t have to be smart if he could just read ordinary English. MR. LARSON: Oh yes. DR. ATANASOFF: Because it was very clear that Mauchly and Eckert had put together a patent case and it had been accepted. You know why I didn’t have a patent? I didn’t have a patent because Iowa State College forced me to sign a patent contract with them and refused to pursue the patent. MR. LARSON: They neglected a very important thing there. DR. ATANASOFF: Neglect and I shouldn’t of stood for it, but I did. MR. LARSON: Well at any rate, how long did this patent suit go on then? DR. ATANASOFF: It went on for a year. MR. LARSON: Oh yes. DR. ATANASOFF: It actually sat for 10 and a half months. Actually the judge was present in a court procedure for 10 and a half months. I was only there during part of this time. and I can tell you some details of my controversy with opposing council. I don’t know if that is a good thing to add to your… MR. LARSON: Well fine. But essentially the net effect was so that as a result of your notes and documentation… DR. ATANASOFF: By this time I had accumulated all of my documents, including this. One of my original papers. MR. LARSON: That particular one must have been a devastating document. DR. ATANASOFF: The whole thing just put them into a position where they didn’t have a chance. By this time they had gotten rid of Dobbs and had a very aggressive council. MR. LARSON: Oh yes. DR. ATANASOFF: And treated me very badly in the court room, but it didn’t really matter. I withstood it. I don’t know what they, they might have conceived that by putting a hard attorney against me I might breakdown, but of course I don’t breakdown, I just slowly paddle on. When the court procedure was over, I went over to shake hands with the opposing council and the man hated like everything, but he finally got his hand up. He perhaps knew better at that stage, that my testimony had been overwhelming. MR. LARSON: Well both your testimony and your documentation in the form of these reports, could have been very… DR. ATANASOFF: Let’s just name a couple of things that Mauchly attempted to steal away from me. The first place, he actually, what kind of division? MRS. ATANASOFF: Non-restoring? DR. ATANASOFF: What? MRS. ATANASOFF: Restoring? Non-restoring? DR. ATANASOFF: Restoring. I had this restoring and non-storing division and he called it in his patent claims. He got a patent of it. Of course when the judge got through, there was no patent there. MR. LARSON: Oh yes. DR. ATANASOFF: They, his jogging. They used my jogging. MR. LARSON: That’s a very important principal. DR. ATANASOFF: You know what? He used base 10. He didn’t use base two. MR. LARSON: That’s interesting. DR. ATANASOFF: And he used, he didn’t use logic systems. He just counted. He had a machine that would just count. The rest of the story was that I outlined for Mauchly, if I needed a differential analyzer, I would build it according to this principal and Mauchly copied that into his ENIAC. He also copied that into his ENIAC. MR. LARSON: Oh yes. DR. ATANASOFF: By this time, well, they, you know, his wife has now written an article which is in this publication, which shows that he had done digital computing before he saw me, but Mauchly had told me that he had never done digital computing at the time he was at my house. MR. LARSON: There wasn’t any documentation of it. He didn’t make any reports of it. DR. ATANASOFF: She claims to have some, but it hasn’t entered into a court procedure and I don’t think most people, I think she’s innocent. I think Mauchly told her that before he died. And she’s written that since he’s died. MR. LARSON: Oh yes. It was a very interesting historical controversy here. DR. ATANASOFF: Yes. MR. LARSON: Well as they say, your documentation on this is overwhelming. DR. ATANASOFF: You read the judge’s decision. MR. LARSON: Oh yes. [End of Interview] |
|
|
|
C |
|
E |
|
M |
|
O |
|
R |
|
|
|