# Oral-History:Murray Eden

(Difference between revisions)
 Revision as of 20:10, 21 June 2013 (view source)← Older edit Revision as of 18:32, 4 November 2013 (view source)Newer edit → Line 397: Line 397:

'''Eden:'''

'''Eden:'''

That’s precisely what I was trying to model. A cell divides in two. They remain sitting next to each other. Then one or another or both of them divide. There are all kinds of algorithms you can invent. My algorithm was very straightforward. I said, “One of these two cells shaped as squares will divide, and if they’re still connected each has only three sides that are not covered. It can divide right, left or down. You build more and more. The work on this model and later, on others, reflects the development of computers, because the first use of this model back in Princeton was on what was called the Johnniac. There were three names: John Von Neumann, Herman Goldstine, and Julian Bigelow associated with this computer. Bigelow was the engineer. We made pictures. The way we made pictures prevented us from adding many cells. The most we could do was on the order of twenty or thirty cells—things were too slow and the only output was a Hollerith punch card. We would produce a picture on a punch card. The holes on the punch card are rectangular, not square, in columns. But if you look closely you can see a picture.

+

That’s precisely what I was trying to model. A cell divides in two. They remain sitting next to each other. Then one or another or both of them divide. There are all kinds of algorithms you can invent. My algorithm was very straightforward. I said, “One of these two cells shaped as squares will divide, and if they’re still connected each has only three sides that are not covered. It can divide right, left or down. You build more and more. The work on this model and later, on others, reflects the development of computers, because the first use of this model back in Princeton was on what was called the Johnniac. There were three names: John Von Neumann, Herman Goldstine, and [[Julian Bigelow|Julian Bigelow]] associated with this computer. Bigelow was the engineer. We made pictures. The way we made pictures prevented us from adding many cells. The most we could do was on the order of twenty or thirty cells—things were too slow and the only output was a Hollerith punch card. We would produce a picture on a punch card. The holes on the punch card are rectangular, not square, in columns. But if you look closely you can see a picture.

'''Nebeker:'''

'''Nebeker:'''

## Contents

Murray Eden was born in Brooklyn in 1920. He went to City College of New York, 1935-39, as a chemistry major. He received his Ph.D. in chemistry from the University Maryland in 1951. His intervening jobs included civil service work; helping produce Uranium-235 at the Palmer Physical Laboratory at Princeton (ca. 1941); National Bureau of Standards employment; figuring out how to eliminate boiler scale from naval vessels, then preparing dielectric constant and dipole moments; Aeronautic Instruments work; and appointment to the National Institutes of Health (NIH) Cancer Institute as a biophysicist working on pH and mass spectrometry of biological materials/experimental animals. From 1953 to 1955 he had a full-time fellowship in mathematics at Princeton. He returned to the NIH, working at the Laboratory of Technical Development on a computer to analyze spectra in electrophoresis.

In 1958 Eden went to MIT, first joining Walter Rosenblith’s Communications Biophysics Lab. From the early 1960s to 1976 he was co-founder, and eventually leader, of a lab on Cognitive Information Processing. He worked on pattern recognition, handwriting generation and analysis, and image processing; and he made independent innovations in computerized tomography. Eden returned to the NIH as the Director of the Biomedical Engineering and Instrumentation Program, 1976-1994. The program has faced significant difficulties under the leadership that emerged after Eden's retirement. This interview covers the evolution of Eden's research and career, emphasizing historical and institutional influences on his life and work.

MURRAY EDEN: An Interview Conducted by Frederik Nebeker, IEEE History Center, 10 November 1999

Interview # 382 for the IEEE History Center, The Institute of Electrical and Electronics Engineers, Inc.

This manuscript is being made available for research purposes only. All literary rights in the manuscript, including the right to publish, are reserved to the IEEE History Center. No part of the manuscript may be quoted for publication without the written permission of the Director of IEEE History Center.

Request for permission to quote for publication should be addressed to the IEEE History Center Oral History Program, 39 Union Street, New Brunswick, NJ 08901-8538 USA. It should include identification of the specific passages to be quoted, anticipated use of the passages, and identification of the user.

It is recommended that this oral history be cited as follows:Murray Eden, an oral history conducted in 1999 by Frederik Nebeker, IEEE History Center, New Brunswick, NJ, USA.

## Interview

Interview: Murray Eden

Interviewer: Frederik Nebeker

Date: 10 November 1999

Place: Rutgers, New Jersey

### Childhood, family, and educational background

Nebeker:

You were born in 1920 in New York City?

Eden:

Yes, in Brooklyn, New York.

Nebeker:

Eden:

My parents, both dead now, were Russian Jews. They met in the United States. My mother was a very interesting woman. Had she lived thirty or forty years later, she would have done very well. She told me (and I have no reason to doubt her) that she was George Eastman’s secretary for a while. She was a foreign immigrant who came to the United States when she was about nineteen or twenty with her father and her sister. My grandfather, whom I’ve never met, was a rabbi of some kind. He was never in charge of a synagogue, but he was certainly a ritual circumciser and slaughterer. He was also the sexton in the synagogue. He had a large family. There were seven kids. My mother was the oldest, and she persevered. She went to secretarial school in Rochester, New York, and ultimately met my father.

My father got a degree just before I was born, in civil engineering from Brooklyn Poly. He was an engineer for a relatively short time. By 1921, the U.S. was in the Depression and he lost his job. He used to tell me that he had designed the steelwork for the building at One Wall Street. In any case, for the rest of his life he was in and around Jewish education. He spoke Hebrew very well. He was a Zionist and the President of the Hebrew Teachers Union, and later, Executive Secretary of the Jewish Education Committee in New York City.

My aunts and uncles lived in the same environment and had no more than a couple of kids each, but we added up. We were all living in the same neighborhood in Brooklyn. My parents pushed me so I graduated from grammar school very young. I think I was eleven. I then went to Townsend Harris, a special high school in Manhattan, New York.

Nebeker:

Did you have to take an admissions exam for this school?

Eden:

Yes.

Nebeker:

Did you study science?

Eden:

No, it was a general three-year high school. It was located (if I remember correctly) on 23rd Street in the building where the City College Business School was at that time. The high school occupied one or two floors. In 1929 I was still in grammar school and things went downhill very fast. I grew up in the Depression.

Nebeker:

Was it difficult for your family?

Eden:

Yes, without question. My father had a job, but during the depths of the Depression he was teaching in a Hebrew school in the Brooklyn Jewish Center. It was a big synagogue on Eastern Parkway. The teachers simply took whatever fees the school received and divided them among the faculty. They didn’t have a regular salary, at least for a while. As a kid and even a teenager, it wasn’t that terrible. I don’t recall feeling poor. It’s only in retrospect that I think about it. We didn’t have any rich relatives, and certainly there was nobody in my public school who was doing that much better than my parents. That was just the way of the world. When I started college in 1935, my allowance was a dollar and a quarter a week. The subway fare in New York at that time was a nickel. So fifty cents of that went toward getting to school on 137th Street and Convent Avenue. Then I had a big seventy-five cents to spend. We used to shoot craps on the fifth floor of the main building. Sometimes I’d win, but mostly I lost. I never had roller skates. I sure never had a bicycle. Once I was sick and my father bought me a baseball glove.

Nebeker:

That was a big event?

Eden:

That was a very big event.

Nebeker:

Were you interested in science and engineering as a youngster?

Eden:

I’m not exactly sure. I did a lot of reading; I’m an omnivorous reader. I read every book I could get a hold of at a very early age. When I was graduated from high school in 1935 I wasn’t quite fifteen. I wanted to be a chemist; I was interested in glass. The Alfred University is in upstate New York and has a program in glass technology. I think it still has one that’s supported in some measure by Corning. It might even be located in Corning, New York. I wanted to do that and my parents were opposed, but I persuaded them and they agreed. I wrote to the University and they said, “Well, you’re too young so you can’t apply.” So I forgot about glass. Then I said, “Well, chemistry.”

Nebeker:

Just to have controlled positions?

Eden:

Controlled positions, that’s right. I don’t know whether he ever used it or not.

Nebeker:

So that was biomedical engineering at an early age.

Eden:

Yes, at an early age. His background in mathematics was poor so far as I can recall, but he was an ingenious fellow interested in looking for attributes in the lives of children who had rheumatic fever. That was the big thing at the time. We developed a large matrix, a table in which we had the individual patients as rows and the attributes as columns. Each one in every case was a binary decision. I recall talking about what do you do about a feature that has more than one go/no go condition, like age. We did what everybody does, and I was not a statistician at the time, and said, “Okay, divide the age into intervals. Zero to one year. One to five. Five to ten,” and so on. This is what is now called a sparse binary matrix, but it had a great advantage. We talked about it and decided that if we wanted to look for correlations, it is a big job to look by hand, so why don’t we wire it? So you have two wires coming in at a particular location in the matrix and if the wires make contact then a bulb will light up. This way is positive, the other way negative, so it would light up.

Nebeker:

Sort of like a computer memory?

Eden:

Right. I made a very simple computer memory. Then we started making the calculation of the number of sockets and bulbs and the numbers of wires you have to solder and so on. We got discouraged. I still remember it because I’m sure others had thought of the same thing, but it was nice that we were anticipating what was coming down the road ten or fifteen years later, roughly speaking. Obviously the technology at the time was not much.

My uncle died in 1959 from a heart attack. This was before we knew about good eating and exercise. His influence was very real. I recall that on the day he died I was called by my mother while I was lecturing at Princeton on my growth model.

### Chemistry studies at City College; political activity

Nebeker:

So his influence directed you to science?

Eden:

I was not interested in medicine. I could recall my parents saying, “Why don’t you become a doctor? If you’re a doctor, you could always make a living.” That was their attitude. I said, “No, I think I’m going to be a chemist.” This was just about the time I was graduating from high school. Incidentally, my brother who was six years younger than me, did become a physician, a pediatrician, presumably because of my Uncle Leo. I went to City College, but didn’t do too well.

Nebeker:

What did you major in?

Eden:

I majored in Chemistry. I discovered that I was interested primarily in physical chemistry. To this day I can’t understand organic chemistry. I have no good feel for the structural organization of the field itself, whereas physical chemistry is much more amenable to the usual kind of mathematics. In college I had certain interests. I was on the newspaper, the City College Campus. I was a radical. I’m probably a radical to this day. I was a Zionist at the time.

Nebeker:

Were you concerned with what was going on in Germany in the late ’30s?

Eden:

Yes, we were. I remember going down to Madison Square Garden one time when the German American Bund was having a rally there. There were a lot of people on the outside protesting. The cops were there, and the horses were pushing in on us and we were pushing the horses back, and so on. It was an interesting time. I was not a communist. As we used to say in those days, “We are not Stalinists.” Although I was certainly a radical. The college paper, at the time, was largely controlled by the Young Communist League. I probably would have stayed, but when they asked me to join I said no. That stopped me from becoming the managing editor in my senior year. I quit and joined this student Zionist organization called Avukah. Avukah had a newspaper that I helped run for a couple of years.

I got a degree in chemistry and graduated in 1939. Then I kicked around for a year. I went to art school, Pratt Institute, for a while and took a course in organic chemistry at Brooklyn College just to have something to do. It was a bad period in my life.

### Graduate studies at Univ. of Maryland

Nebeker:

Did you look for a job as a chemist?

Eden:

I looked, but I must admit I didn’t look very hard. I stayed home and fought with my parents. Looking back, there is no question that my relationship with my father was terrible at that time. It was my fault, but I’m not going to try to analyze it. In any case, I then did what a number of people did at the time. I moved to Washington in 1940. I was admitted to the graduate school at the University of Maryland and started going again in Chemistry and worked. The U.S. wasn’t yet in the war but there was a war going on. A lot of young men came down there.

Nebeker:

What was the attraction in Washington?

Eden:

Jobs.

Nebeker:

There were a lot of new jobs?

Eden:

Yes. There was a general feeling that the United States would ultimately get into this war, and some of my colleagues at college told us to come down because, first of all, there are jobs; and secondly, there are loads of women. That was a big inducement in those days. There must have been five women for every man.

### Civil Service Commission employment

Eden:

I lived in what we called a commune. We had about a dozen people, all essentially the same age, and almost all from City College. Most of them I had known before. I got a job with the Civil Service Commission and in a way I enjoyed it. I barely passed the typist test. 70 was passing and I got 74.2. I felt very good about it, but they didn’t offer me a typist job. They offered me a job as Assistant Messenger, or some other low level category. Maybe it was Junior Clerk Typist, but it had nothing to do with typing. The work we were assigned is relevant to one of my later interests. The Civil Service Commission decided to change the way they indexed people who were applying for positions or were in the government. Rather than simply using the last name and first name and so on, they developed a code that ran as follows: the first initial of the last name, then the next three consonants are assigned a number. We converted the consonants to numbers, then added something else like date of birth or some other small set of numbers that related to the individual.

Nebeker:

Why didn’t they simply use Social Security numbers?

Eden:

I don’t know. Certainly Social Security numbers existed, but that’s not what they decided to do. They brought a group of twenty or thirty of us in a big room and some lady gave us a big lecture about how we were supposed to do it and have to be very accurate. They gave us a little card that contained the code because we had to translate the name, decide what the appropriate consonants were (Schmidt and Smith coded out the same), and we were supposed to transpose and then map each consonant into zero to nine. I had my first seminal idea to memorize the code, and then I operated like a printing press because I’d look at a card and do all the conversions in my head, then write down the numbers on the card in pencil and somebody else would transcribe. It was clear by the second day that I was doing way more than anybody else was, so they rewarded me by giving me a different job. Then they gave me a longer appointment. My original appointment was for three weeks or some trivial length like that; then I had an appointment for three or six months. My idea was translatable into other forms. Now I became a typist. We were typing on the first electrical typewriters, the average electrical typewriter in those years were monster big machines. I think they were called Electrowriter. We had to type in octuplicate. The typing paper was in fan fold, so this paper was infinite in length. We were lectured again. They were all lady supervisors then. Again, this was the War. We were not quite in the War, but it was becoming clear that that is where we were going. We were told, “These documents are going to appointing officials of one kind or another and it has to be clean, so when you erase be sure you erase well.” You have to feed your hand in sideways (three on one side and four on the other side) in order to make all the erasures on all the carbons. But the worst thing that you could do was to “x” anything out.

We worked in a very long room. The Civil Service Commission building was two blocks long, of a block size as from Fifth to Sixth Avenues in New York. It’s a long block. The building was between Seventh and Ninth Street Northwest in Washington. It was a very long building. There were a hundred typewriters or so, all in a line, and there I sat at one of them. Virtually everybody else in that room was female. That’s the nature of typing. A lot of my friends were, indeed, Assistant Messengers. A college degree carries you pretty far! I worked the 4:00 p.m. to 12:00 a.m. shift, so I could go to school during the day. I commuted to College Park, which is in those days was a trolley ride of half an hour. I was clearly the slowest in this long line of typists. Once again, I thought through the problem and then one day I said, “Okay, let’s go.” Then I quickly won the contest. I was doing two or three times as much as any other typist on the row.

Nebeker:

Eden:

The night supervisor even asked the day supervisor to stay over one afternoon in order to examine my product. They didn’t understand what was going on because I wasn’t a fast typist. I said, “I don’t make mistakes,” which was a big lie. I recognized that they were very upset about erasure, and once you have to erase eight pieces of paper that killed time something terrible. So I never erased. That’s the whole story. I never erased. If I made a mistake, I would usually justify it this way. If you make a single character mistake on somebody’s name or address, unless it’s the first letter of the last name, it’s not an easily detectable error. If you spell Texas, T-e-c-a-s, it’s going to get there because everything else is right. So that was my secret. I had learned the value of redundancy. I never – or rarely --corrected anything.

So in typical bureaucratic fashion, they made me a supervisor. So I had this harem of a hundred women, and my big job was to bring the cards to them so they could type the names. I was going to school at the time in chemistry. I decided that physical chemistry was my real interest. After about six months or eight months, I was offered a day position.

Nebeker:

That was more desirable than 4:00 to 12:00?

Eden:

A large increase in pay and everything else. I turned it down on the grounds that I wanted to be able to continue in school. So that’s what I did. Then they got rid of me and told me those were the conditions. By this time I had saved enough money that I could work for about one semester worth, and my father lent me some money, so I managed a semester with no income coming in.

### Employment at Princeton; mass spectrometry for Uranium-235

Eden:

Another interesting thing happened at that juncture. I had a friend named Herbert Bridge, who ultimately became a professor of physics at MIT. As far as I know, he and I are the only University of Maryland graduates who ever made MIT faculty. But we were classmates. I had trouble—there was a good deal of anti-Semitism in the University of Maryland. I applied for a fellowship and they turned me down. I remember talking to Herb and I said, “I don’t understand.” He said, “Well, you know, it’s very simple. You’re Jewish.” Anyway, Herb went off to Princeton. He got a job there. He wrote me a letter telling me, “Read an article in Coronet.” Coronet was a little magazine being published.

Then I read the article, and it was about splitting the atom and its potential use as a weapon. At the end of the semester, I went up to Princeton, interviewed, and got a job there. This was one of the corners of the Manhattan District. It was run by a fellow named Robert R. Wilson who became very well known later on. Among other things, he was the director of Argonne National Laboratory for a while. He was a physicist. I started working there, and met some very interesting people. The room next to mine was occupied by Dick Feynman and a fellow named Paul Olum. They were both mathematicians. Olum spent his academic career primarily as a mathematician, although Feynman was primarily a physicist.

Nebeker:

You lived in Princeton?

Eden:

I lived in Princeton. I rented a room and there I was. I wasn’t there very long; about six months. I had a very big disagreement with my immediate supervisor, a young man named Robert W. Thompson. I don’t know what happened to him. He had trained at a physicist, and to my way of thinking he was a remarkably unpleasant fellow. Looking back it’s kind of funny. I was a kid of twenty-one or twenty-two, and I didn’t like to take orders. My boss wasn’t that much older than me. He might have been a year or two older than me, but then Dick Feynman wasn’t that much older than me either, and we were pals.

I have a little story about Feynman. I told you I was a chemist. There is an operation in chemistry called a titration. One component is added gradually to a fixed amount of another. Essentially you are trying to determine when a certain amount of one material combines with a certain amount of another material, and if the ratio is one-to-one or one-to-two or whatever. There is usually an integer relation between these quantities. The curve itself is plotted like you’re measuring pH. It has a point of inflection that is considered to be the end-point. It occurred to me that the point of inflection is not necessarily the same as the point of equal mass balance. I’ve forgotten the precise term. It was assumed in every chemistry course that I ever took that they were the same. You always look for the point of inflection. I asked, “Is it mathematically correct?” My knowledge of calculus, although I had taken calculus in college, was not that great. I remember talking to Dick one time and he didn’t know what a titration was, so I explained it to him. He said, “Oh, that’s an interesting question.” He got to the board and started writing in the upper left hand corner. It wasn’t a complicated problem, but it took a couple of blackboards worth. He worked it out and, indeed, my intuition was correct. For certain classes of reactions there is a difference between the point of inflection and the point of mass balance. The difference is very small under almost any condition, but it is not identical. I felt very good about it, and then he started erasing. I said, “Dick, don’t do it.” He said, “You can reconstruct.” Well, he did it off the top of his head. It took me two weeks.

Nebeker:

Did you get a paper out of that result?

Eden:

No. In those days I never got to the point of worrying about papers. The world was a different place. I had no real knowledge of science, and popular books about science were relatively rare. I really knew very little about the dialogue among scientists, or how one goes about carrying on an experiment in such a way that there is a result that is publishable. That came later. I did learn some things in Princeton. I wasn’t there that long, but aside from that, the goal was so clear where we were going . Security in those days was sort of a joke. We were in the Palmer Physical Laboratory at Princeton. I don’t know what the building looks like now, but our labs were in the basement. They had windows and there was a railing and chicken wire fence, and in order to get into Palmer you had to show your badge.

Nebeker:

Because the whole area was secured.

Eden:

The whole area was secured, but we didn’t use it. It was too much trouble. We jumped through the window. We went over the barrier and in. The fellow who ran it was named Smythe.

Nebeker:

Is that the famous professor?

Eden:

Yes, H. D. Smythe.

Nebeker:

Didn’t he go to Cal Tech?

Eden:

I don’t know what happened to him. He was older. He must have been 35.

Nebeker:

An old guy.

Eden:

I don’t know. Maybe he was even older. He had an office and a big filing cabinet with a combination lock on it. I said, “I’d like to find out about something or another.” I don’t even remember what it was. His secretary said, “Okay, go help yourself.” Now she knew I belonged, but I was as low as they came in the pecking order. I pulled out the drawer and I looked in the different bins. I knew where we were going, but I wasn’t very clear on it. I’d pull out and read a sheet of this, put it in and go on. I never did get to Los Alamos. They told me later that once General Rhodes came inside he was much tighter.

Nebeker:

What were you actually doing?

Eden:

Our problem was to produce Uranium 235. We used a mass spectrometric technique, essentially. Bob Thompson, who was my superior, had worked for A. O. Nier, a physicist at Minnesota who built the first mass spectrometer. We built a device called the isotron. My particular job was to prepare the raw uranium tetrabromide in a capsule glass which we then put into the mass spectrometer, broke the capsule, and evaporated the uranium tetrabromide and separated U235. We used to produce a milligram or so per day. It was a very slow process. But that was my job basically as a chemist.

I had certain side interests. I had to do titrations, and I wanted to do potentiometric titrations. But rather than use a galvanometer, which is very slow, Herb Bridge and I were going to build a vacuum tube amplifier.

Herb and I started building the amplifier, and Bob Thompson walked in and said, “What are you doing?” I said, “We’re building an amplifier.” He got very upset and he said, “Oh no, you can’t do that. Go back to doing what you were doing before.” About the same time there was a decision made. There had been a fair success elsewhere, at the Fermi Lab, of course, so they decided to just stop operations in Princeton. I would have been offered a choice. You could go either to some strange place in Tennessee, or you could go to this even stranger place in New Mexico. I wasn’t offered, because as soon as I heard about it I said, “Well, this is not for me. I’m a city boy.” I had connections in Washington. I said, “Well, I think I’ll go back to Washington.” So that’s what happened. In retrospect, it probably was a big mistake. Bridge went off to Los Alamos, and he’d write me a card every now and then about how things were going there. I went back to Washington and got a job at the National Bureau of Standards. Now we’re up to early 1943.

### National Bureau of Standards employment

Nebeker:

Eden:

That’s an interesting question. I was 1A when I left Princeton. Up until that time, I had no problem. I was protected; that was clear. I must have had a draft card by that time. In any case, I decided to go to Washington. I interviewed at the Bureau of Standards and they said they’d let me know. Bureaucracy being what it is, and the typist still typing while we were talking, I went back to New York and I didn’t hear anything. Every now and then I’d call up and say, “What’s happening?” Then I got my notice, “Greetings from the President,” to be inducted. My mother prevailed on me to call up the Bureau of Standards. Again they said, “Oh, yes. We’re going to send you a telegram.” Within a week I was supposed to report for induction. Finally, I got the telegram from NBS but I figured, “It’s too late.” My mother said, “No, go down to the draft board.” I went down to the draft board and the woman didn’t even read the damn telegram. She said, “Oh, we were waiting for this.” She scratched my name out, and I became 2B. Later, I also became 2BF because I can’t see anything, my correction was about 20-400, so I probably could have avoided military, at least at that time in the War. A little later it got tighter.

But I was essential to the War effort, and I remained there. It was kind of sad. I used to go down to the induction center every six months and we’d all stand naked in a single line. Then I would be called out and sent home and all these other boys would have to go into the army. I felt guilty about it, but I didn’t worry about it too much in those days.

Nebeker:

What were you doing for NBS?

Eden:

I had a variety of jobs. The first job we had was working on eliminating boiler scale for naval vessels. I was in what was called the pH standards section.

### Ph.D. research

Eden:

I made use of the time to start working on my Ph.D. I was able to use the Bureau of Standards facilities to do my research.

Nebeker:

Did you have an advisor at Maryland?

Eden:

I had an advisor at Maryland, but it was pro forma because my basic advisor was my boss, Roger G. Bates. Ultimately, my Ph.D. thesis was on the dissociation constants of malic acid. The interest from a scientific or academic point of view was that the two carboxyl groups of malic acid are not quite in symmetric positions. It’s not like succinic acid. In malic acid there’s an OH group on one side. But the ratio between the dissociation constants of its two acids is very close to the least that it can be. It turns out that there’s a mathematical result about the minimum ration between acid groups on the same molecule. In a way, it continued to turn me toward mathematics. The minimal ratio between the dissociation constants of a single dibasic acid is sixteen. In other molecules there are certain regions of symmetry. For malic acid it is probably at the order of thirty. But the small ratio makes it very difficult to get highly precise values for these dissociation constants. For me it involved the development of recursive procedure for calculating. We introduced a variety of correction factors. The Debyl-Hueckel theory. So it wasn’t really physical chemistry. It had no real relation to what they were paying me to do, which was to work on the boiler plate, but it was where I went. NBS, like a few other agencies, let their researchers – especially students – work on their own projects; maybe not during the working day, but after hours. It took a long time because I couldn’t spend too much time on my thesis; I had other things to do.

### NBS working environment and research

Eden:

They shifted me around. I got into a fight with another chap (not Robert Bates) who was my direct supervisor, and worked in the electrochemistry section where we were plating the inside of fifty caliber machine gun barrels with chromium. My clothes were destroyed within two weeks.

Nebeker:

Why is that?

Eden:

Because the bath that you use is at ninety degrees centigrade and made of dichromate dissolved in sulfuric acid. That’s a terribly corrosive substance. You couldn’t help getting it on your clothing. You tried to avoid getting it on your hands. We had these big tanks to hold the fifty caliber machine gun barrels. We had a hoist and dropped the barrel down into the tank, connected it to the power supply, and then plated the inside. We’d have to put an electrode through the barrel. I did that for a while, then got back to pH standards. Then they shifted me over to the mechanical instruments section.

During this period of my life is I had interests in two different areas. I like mathematics. I didn’t know very much mathematics—just what I learned as an undergraduate. And I liked gadgets, so I then went to work in what was called the Aeronautic Instruments Section and later was called the Mechanical Instruments Section. By now we are later in the War, about ’44 or ’45, but before the end of the War. I worked on a method for converting liquid oxygen into gaseous oxygen for use by aircraft personnel on bombers. I was the only person in the section who had a background in chemistry. They were all physicists and mechanical engineers. One of my good friends, Jack Rabinow, was very highly regarded by the Bureau; he was a prolific inventor. In those days he was famous for his work at the Diamond Ordnance Lab on the proximity fuse.

At the end of the War I still hadn’t finished my thesis. They put me on a library research problem that had to do with a set of tables the Bureau was preparing on the dielectric constants and dipole moments of everything on God’s green earth. I spent two years in the library. I actually read every language in the world. Papers in the Latin alphabet letters are easy, primarily because I knew what the subject matter was and I could always interpret the graphs with the help of a dictionary.

Nebeker:

So you were collecting experimental results from publications from all over the world?

Eden:

Yes. The guy who was my boss was a very pleasant fellow, but at the end of two years I said, “Look, I’ve got to get off.” He said, “You can’t because we’re still not done. Don’t you want your name under publication?” I said, “I just can’t stand it anymore.”

### National Institute of Health, Cancer Institute; mass spectrometry

Eden:

So I interviewed and got a job at the National Institute of Health in the Cancer Institute. Then I was converted to another category. I had been called chemist (physical) and then physicist (mechanical). But while in the mechanical instrument section I was changed to physical scientist. I didn’t mind the change. When I came to the NIH they made me a biophysicist, so that became my title. The laboratory that I was in was called the Laboratory of Biophysics. The director, Egon Lorenz, had been involved in the development of radioactive materials for medicine. I had nothing to do with that. Since I had a background in pH, we worked on the measurement of pH under different circumstances in experimental animals, mostly rats. We developed very fine glass electrodes. You have to slit the skin of the rat, and then you can stick the electrode in various places, mostly in muscle or tumor tissue. It turns out that if you inject the peritoneum of the rat with a large amount of glucose, you can almost instantaneously (within 30 seconds) get a significant drop in the pH of tumor with small changes in that of muscle. So here you have the electrodes sticking into the muscle and the glucose does something. I think we understood why, but I’ve forgotten the details now. I believe it reflected the fact that tumors have an anaerobic metabolism. I worked on that a while.

There was a mass spectrometer in that laboratory, one of the first in a biology lab; I worked on that. We were working on a variety of things. The biological relevance was that if you fed a rat a high fat diet, which they would eat, then they would develop acetone in their breath, which you could then measure in a mass spectrometer. You could do the dynamics of how long it took to rise and fall and things like that. That got me back both into pH and into mass spectrometry.

One of the novel things we did was develop a technique for measuring the molecular weight of compounds, intact molecules. Now the way a mass spectrometer works is that it ionizes and breaks up the molecule into little pieces. We knew that the concentrations are so low in the mass spectrometer that the particles, whether they’re ionized or unionized, are basically perfect gases. They obey the perfect gas laws. The orifice in the inlet is a little pinhole.

Physical chemistry provides a proof that for a perfect gas, the velocity or the rate of the diffusion through a pinhole is as the square root of the molecular weight, or the mass number, whether you’re measuring the intact molecule or any of its fragments. That’s true of ionized particles or unionized particles. So all you do is you measure a particular peak in the mass spectrum as a function of time. You don’t have to look at the whole mass spectrum. You just look at the peak and how it decays as a function of time. The slope of that curve in a semi-law plot will then tell you what the square root of the molecular weight is, and it works very, very well. It had some particular advantage, and it made the annual review of mass spectrometry. For example, we used the technique to measure the area a water molecule on a glass surface. We could readily estimate the surface area of the manifold. Our technique gave us an estimate of the number of molecules that could stick to the glass surface. We used both regular water and heavy water. If you go through the procedure with water as the sample, what you’ll find is that as the surface becomes covered with water molecules, this curve that I’m generating no longer is straight, semi-log split. It curves over, and then it becomes asymptotic to zero and then you can estimate how many molecules you have permitted through the pinhole and the number can then be divided by the surface area of the manifold. This agreed well with what physicists had discovered by using other methods. That was kind of nice.

I worked at the Cancer Institute on a variety of problems.

### Public Health Service fellowship in mathematics, Princeton

Nebeker:

Eden:

Well, almost. I’m trying to remember now what happened next. My Ph.D. was in ’51. I was still at NIH. I had a friend at the Cancer Institute, Arnold W. Pratt, called “Scotty”, and we used to discuss mathematics. He ultimately became the Director of the Division of Computer Research and Technology at NIH. He was a physician, but had a strong interest in mathematics. At this time the whole issue of coding and computer storage was becoming important. We were talking one day and he said, “Why don’t you get yourself a public health service fellowship and study mathematics?” Which is what I did. I quit the NIH once I got the Public Health Service fellowship and went to Princeton in the math department. I was in the math department for two years.

Nebeker:

So this is additional? You have a Ph.D.?

Eden:

Yes, I have a Ph.D. Coming back to the issue of mathematics, the training in mathematics in those days (and I don’t know that it has changed that much) was primarily in calculus, whether it is for physics or chemistry. More recently there’s a good deal more stress put on differential equations and matrices and things like that. Those subjects have relatively little to do with information directly. The kinds of things that we were doing then in the NIH seemed, to me at least, to have more to do with discrete information than it did with the kinds of things that a physicist ordinarily did. As a trivial example, the interest in titration is not in the point of inflection. You want to know whether this is equal to that, and is in the proportion one-to-two or two-to -three or whatever. So it is a discrete relation. While I was still at NIH (I think I published; I’ve been looking for it and can’t find it), I worked out a set of theorems about how from the mass number you could tell how many nitrogens a molecule contained; modulo four, it turns out. It’s just a little number game. Oxygen is 16, so oxygen always divisible by four; nitrogen is fourteen and carbon is twelve, and so on. You can also tell how many halogen atoms are associated because there will be a change, again, modulo four. There will be a change of one in the value because, except for isotopes, the predominant halogen mass number is odd. So you’re counting odd things as opposed to even things. The valence of nitrogen is three. If you take the three hydrogens attached to the nitrogen that number is also odd. You can play some games like that, and you actually prove what are really theorems.

Anyway, I went a second time to Princeton to study mathematics. My mentor was Willie Feller, who in those days was the major probabilist in the United States. He had just finished writing what became the classic text in probability and as far as I know is still important. I studied that, and logic with Alonzo Church. Those were the basic things I studied for two years. I found out relatively early on my studies in Princeton that I couldn't become a pure mathematician. I’m not comfortable with abstract notions, or my head doesn’t work well in proving theorems even though I’ve proved a few. But I enjoy mathematics, and I felt I was a reasonably good investigator in the application of mathematics, in particular the mathematics of probability theory and, to some extent, logic.

Eden:

After the two years I went back to NIH in 1955. I was interviewed in several places, but I didn’t get along with an interviewer at IBM. He made an interesting comment to me based on my background. I told him I wanted to apply my skills toward biology and medicine. He said, “Well, we would be interested in hiring you. There are two reasons to hire you. One, for the PR of it. If what you’re doing is relevant in some sense so we can get publicity.” I’ve even forgotten what the second reason was, but that frosted me and I decided I wasn’t going to go to IBM. At that time my wife would not have wanted to move to New York. Then I interviewed at National Security Agency (NSA), and they didn’t want to tell me what they wanted from me. Obviously I knew by that time that they were doing coding and cryptography and what not. I didn’t like the fact that as a visitor I had to have a Marine go with me to the men’s room.

Ultimately I went back to NIH. By this time I had begun to publish in what ultimately became biomedical engineering. I did not join the Institute of Radio Engineers (IRE) until later, about 1960. That was actually after I came to MIT. Just to continue the chronology, I was in a laboratory at NIH called the Laboratory of Technical Development. The director was Bob Bowman. Bowman had no interest, for reasons which are unclear to me, in IRE. He was then a member of the ISA, and an very much an inventor and a skillful gadgeteer. He was a physician by training, but he wasn’t practicing medicine. We used to go around NIH looking for problems in which technology might make a contribution. I worked there on a variety of different things.

I worked on two important pieces of work at the time. We developed a special purpose analog computer to analyze the kinds of spectra that you might get in electrophoresis, or anywhere the curves of the experimental data had peaks, under circumstances where you know that these peaks and valleys represent concentrations of some entity, molecules or ions usually, but with a lot of overlap of the peaks. This is the kind of problem that is amenable to mathematical description, usually, and it’s surprising what you can do. We could take the circumstance where you had a peak and just a little teeny bit of shoulder on one side and you could estimate the concentrations of the two components that are overlapping with remarkable sensitivity, so long as you knew the distribution function that the peaks followed. So it could be used in a variety of applications, including optical spectra where there is pressure broadening. We built this device, and it worked fairly well. It was ultimately manufactured by Dupont for a period ten or fifteen or twenty years, but by that time I was getting ready to leave NIH.

Another thing that I worked on was understanding the physiology of how urine concentrates in going through the kidney and into the bladder. There is a theory called the counter current theory because there are channels in the kidney that are hairpin-turn loops in which the fluid goes through first the center of the kidney and then up, while water and chemicals dissolved in the fluid can pass from one side of the loop to the other through their cell walls. There is also a plexus of capillaries going in one direction, and there is an exchange between the capillaries and the substance being removed from the fluid that becomes urine. The liquid, which is captured in the kidney by the glomeruli, ultimately goes to the bladder. The man who was then the director of the lab and later became the scientific director of the Heart Institute as well as Scientific Director of NIH, was Robert Berliner. I provided the mathematics for it as theory at the time. He later became Dean of Yale Medical School. He directed the preparation of a major review paper. Another instrumental development I worked on had to do with separating out groups of red blood cells based on their density. Human red blood cells have an age of about sixty days. The young are lighter in terms of density than the old ones, so you want to separate them out by tagging the newly formed cells with some radioactive material to identify a particular fraction. I also continued to work on what I had done at Princeton, basically a growth model. As far as I can tell, and nobody seems to have found a precursor, it is the first algorithmic, computer-worked-out model of two-dimensional growth. The original model is very easy to describe.

Nebeker:

Microbial growth, for example?

Eden:

That’s precisely what I was trying to model. A cell divides in two. They remain sitting next to each other. Then one or another or both of them divide. There are all kinds of algorithms you can invent. My algorithm was very straightforward. I said, “One of these two cells shaped as squares will divide, and if they’re still connected each has only three sides that are not covered. It can divide right, left or down. You build more and more. The work on this model and later, on others, reflects the development of computers, because the first use of this model back in Princeton was on what was called the Johnniac. There were three names: John Von Neumann, Herman Goldstine, and Julian Bigelow associated with this computer. Bigelow was the engineer. We made pictures. The way we made pictures prevented us from adding many cells. The most we could do was on the order of twenty or thirty cells—things were too slow and the only output was a Hollerith punch card. We would produce a picture on a punch card. The holes on the punch card are rectangular, not square, in columns. But if you look closely you can see a picture.

Nebeker:

An early graphical output.

Eden:

Exactly. Really a pictorial output. As far as I know, that is the first pictorial output. My first ones were, of course, done by hand. I did it with little red stickers that I put on a big grid. Later when I came to MIT, working on TX2, we could grow these easily and in a few minutes we could do a couple of hundred thousand. They’re parameterized in different ways: you can change what happens at the edges, you can make them grow in one or another direction.

Nebeker:

Did other people pick up on this model?

Eden:

For a while nothing happened. It probably helped me get to MIT, but that’s a separate issue. Practically no one was interested. Rather recently, on the order of ten or fifteen years ago, I was back at NIH and I had a French post-doc working for me. He said, “I saw your name in La Recherche.” I said, “Oh really? That’s interesting.” He showed me the copy, and my French isn’t too bad, and plainly it says “Eden.” The author was talking about developments in crystallography. It took me a little time, but I finally realized that they had rediscovered my 60s paper in the late ’70s or ’80s. I actually looked it up in the Science Citation Index, and twenty years after this paper was published in 1961 there were 20 citations yearly. The major paper (there were two papers) appeared in the fourth Berkeley Symposium on Statistics and Applied Probability. Jerzy Newman, an eminent mathematical statistician at Berkeley, organized and ran these symposium, every five years. They’d always have a session on oddball contributors of our time. That’s what I had a major paper on the subject of biological applications. It was never in a journal, strictly speaking, but in any case was rediscovered by the mid ’80s and still gets a lot of citations. Now, it’s a very simple model and it is easy to complicate. Even I have complicated it. I gave a lecture at a meeting I was invited to in Vienna two years ago and talked about the model. It’s so easy now that we ran a set of experiments and made some fancy ones of one kind or another in a couple of weeks.

### MIT filtering electroencephalograms

Eden:

How I got to MIT is a funny story. I was in the Heart Institute (this was before it became the Heart, Lung, and Blood Institute), and everything was going fine. There was a big symposium on biophysics held in Boulder in 1958. The principal organizer was Frank Schmitt, a biologist professor at MIT. I had met him before, as well as his brother Otto Schmitt, one of the foremost biomedical engineers of the day. I had also met Walter Rosenblith. Rosenblith was also at MIT. He has an interesting background himself, and when I came to MIT, I joined his Communication Biophysics Lab. From that conference in Boulder sprang a very large growth in biophysics. Biophysics is a little older in the sense that in about 1955 the Biophysical Society was created. I recall Frank Noble and I gave a paper there. I went to Boulder as a member of the NIH delegation and gave my paper on this growth model. Frank Schmidt threw a big cocktail party. In those days I probably drank more than I should. I drank a lot at this cocktail party, and I had to go to the bathroom. I was standing at the urinal next to Walter Rosenblith, and he said, “What’s a smart guy like you doing at NIH?” I said, “I don’t know.” He said, “Why don’t you come to MIT?” I said, “Make me an offer.” I’m not kidding—that’s precisely what happened! I went back to Washington and forgot the whole thing. About six weeks later I got this letter of invitation to come to MIT. I went originally as a research associate, but I talked to Jerome Wiesner, whom I had also met before, and he said, “We can’t hire you as a faculty member immediately, but we’ll make you an associate professor as soon as a proper package can be assembled.” About six months later they did, and I stayed.

Nebeker:

What was Rosenblith’s interest? Was it this work on the growth model?

Eden:

No. In those days, if you talked about the growth model it related to what we referred to as pattern recognition. If you think about what I said about this model that dealt with cursive script, it’s also pattern recognition. So there’s a connection between in a variety of things that I’ve worked on. In a way, physical and biological scientists are trying to make sense out of patterns of one kind or another.

Nebeker:

What was Rosenblith’s interest in you?

Eden:

I don’t really know. He had heard me speak. He came down to NIH and gave a lecture. I don’t remember the topic. While I was still at Princeton I had met Claude Shannon and we hit it off. I mentioned Bigelow. I became involved with the people in information theory. There is a very well-known and important paper by Shannon and E. F. Moore with kind of a cute title, “Reliable Circuits from Crummy Relays.” It is a beautiful paper because in a straightforward way it deals with the problem of correcting errors in an error-prone network. That’s a problem that exists to this day—how much redundancy can you build into the network so that the functions that come out of the end are what you want?

Rosenblith certainly had an interest in biology. The name of his laboratory was Communications Biophysics, and there were a number of people who associated with him, including Norbert Wiener. Wiener had developed, perhaps with Rosenblith, what he called the Computer of Average Transients, or CAT. If you were looking at electroencephalograms which were one of his major interests, or if you looked at what are called evoked responses, each one of these repeating curves have a family resemblance, but they look a little different. If you average them in the right way, what does it look like and what does it tell you? A lot of work has been done in other places, but it was done early in Rosenblith’s group, which started about ’56 or ‘57 before I came. Much of the equipment was still not working when I got there.

Wiesner had another idea with regard to filtering, especially filtering electroencephalograms. His argument was that if one were to put electrodes on the wall and look at the noise, the predominant frequency would be sixty hertz. He said, “Let’s do it with the brain and see what we get.” We know of alpha and beta and others in encephalograms. He wanted to find out what affected these frequencies in the brain if they were filtered properly. A device was built for him which basically did autocorrelation functions on the brain waves. You record something, and then on one channel you delay it a little bit, multiply the two together, and if there is no correlation it should be zero; if they are perfectly correlated it should be one. You can draw this correlogram, done originally mechanically by using a drum that would index over after each rotation, while on that magnetic surface you had laid out a certain portion of the signal, and then you multiply that by the tape that was going.

Nebeker:

Is that how some of these analog devices worked?

Eden:

Yes, absolutely. They worked; there’s nothing wrong with analogs. They’re kind of slow by modern standards. Nobody is likely to resurrect the analog, but I’d hate to try to predict technology’s future.

Nebeker:

So that was Wiener’s idea?

Eden:

Yes. The experiments were done largely in Rosenblith’s lab by various people, including Larry Frischkopf, Bill Peake, Tom Weiss, Moise Goldstein, and Bill Siebert a little later. All were MIT professors.

### MIT working environment, funding; transition to electrical engineering

Nebeker:

What was the atmosphere at RLE in general?

Eden:

It was very friendly. I came just after the semester had started in the fall. I was met by Peter Elias and Claude Shannon at the Kendall Square Stop or at the faculty dining room, which was in the building that the Sloan school occupied on Memorial Drive, and we walked into the campus. Peter Elias, whom I had known for some time, had been a contributor to very early coding in information theory called block coding. One of these two guys asked, “I’ve got to talk to somebody who knows about electronics. Who should I talk to?” The other guy said, “Well, Jerry Lettvin, of course. Who else?” He wasn’t kidding. Jerry Lettvin was an incredible circuit designer. As you know, I was coming from a biological/biomedical background, and I was going to be associated with what was called the Center for Communications Science (which didn’t live very long; MIT didn’t raise much money for it so they folded it after about a year or so). I knew that’s where I would have an office. I said, “With what department would I be affiliated?” They looked at me and said, “What else? It’s Electrical Engineering.”

Nebeker:

So that’s how you got into electrical engineering?

Eden:

That’s how I got into electrical engineering. But I will admit that by that time I was a pretty good country inventor, if that’s the right word, tinkerer. I had a lot of experience at the Bureau of Standards and then a fair amount at NIH, and I like instrumentation. I was not a really good electronics designer in the sense of being able to look at a circuit and know what it’s going to do without some studying. But, I ultimately taught all but one of the core courses in the undergraduate Electrical Engineering Department. There was no computer science as such that at the time, but obviously I knew how a computer worked and we used what we had available.

Nevertheless, I tell you the story about my experience because, due primarily to Jerry Wiesner, there were a lot of strange people in electrical engineering, and especially in RLE. There was Lino Feretti, a jazz musician who wanted to use the computer in order to make new music and new instruments. It took him about a week to make one note, but he would be supported by RLE for a number of years. There was a Mexican mathematician and physicist whose English was poor named Manuel Cerillo. He wanted to understand style in art and also music. Cerillo took a photograph of Warren McCullough and made it like Michelangelo’s Moses. You could see that it was the same individual, the facial expressions and everything, but it looked like a painting. It was just amazing. We never did figure out how he did it. The first serious papers trying to relate neurology to formal logic were by McCullough and Pitts. McCullough came to MIT in the early ’50s. He was a neurophysiologist or psychiatrist, and he was interested in formal descriptions of logic in the brain. Walter Pitts was a mathematician and a person who clearly could not live by himself—he was very strange. But he was protected. He was very difficult to understand him when he talked, and most of the time he wouldn’t talk. Pitts and Lettvin gave a series of lectures trying to explain what Cerillo was doing. I was asked by Jerry Wiesner to listen and see if I could figure out what’s going on. I never could. Later, I talked to Manuel. He showed lots and lots of illustrations.

Nebeker:

Eden:

No, it wasn’t that he was secretive, but we couldn’t understand him. He had a notebook full of numbers and matrices and he’d say something about each page, but I couldn’t understand what he was talking about. The point to this anecdote is that MIT in general, and certainly RLE, was very receptive to interdisciplinary directions, largely due to Wiesner. The times were such that money was coming in mostly from the federal government, mostly from the military –the Army Research Office, and it was done with almost no strings attached. With a little ingenuity almost every basic research project can be associated with a project listed by the military. We did what we wanted and that was great fun.

Nebeker:

There was more discretion, evidently, on the part of the people doing the hiring there then would be the case now.

Eden:

Certainly. Nowadays it’s quite different, to refer to what I see at MIT now. People are very friendly to me, I have no problem on a collegial level. They will not give me an office, or even a desk, probably because space is so tight. I’m not bringing in any money. I’ve been away a long time. I’m not costing them anything; I get a little expense account for my course. That suggests to me that in some measure resources are in short supply. On the other hand, there is a tremendous amount of money, potential and actual, coming into MIT from industry. I don’t exactly know what the strings are, but there are strings, clearly. Big sums of money. A former student who I had not seen in thirty or more years, an Indian named Kenan Sahin, was in the Times or the Boston Globe yesterday. On the spur of the moment, he gave MIT one hundred million, just like that. Now, he still has \$1.35 billion dollars left, so that was not quite a tithe. MIT is now campaigning to raise one and a half billion dollars for a new building. The new building will be largely for computer science and electrical engineering. The architect, Frank Gehri, famous for this fantastic museum in Bilbao is doing the design. I’ve seen the sketches and lots of people in the department hate them. It’s avant-garde, and in recent months he’s toned it down a little bit. It’s going to cost a lot of money, and they raised most of the money. The money is coming in, especially for computer science and media arts, which was what Jerry Wiesner was interested in the latter part of his life. He had always been interested in it because his first professional job, as far as I know, was as the engineer going around with Lomax and tape-recording the folk songs in the south and Appalachia for the Library of Congress. I’m not even sure how he got to MIT. So he had this rather catholic interest in things that are quite different from simply engineering. The building in his name, the Wiesner building, is like that, with a lot of art work and especially sculpture. Near the library nowadays there is a sculpture court in his memory.

The style for the place has not changed, but in other ways it has changed dramatically, in that the first thing that people think about when they’re planning a project is what the relation is going to be with the money source. Of course, in some cases you still look for a money source after you decide on a project, but my impression is there is less of that going on.

### Symbol Subcommittee of the Universal Grocery Product Code Council

Eden:

I had a particular association on what you might call low technology. Low technology has had a profound effect on commerce. For instance, bar coding revolutionized commerce in the United States and the world. I was the principle technical consultant to the committee, called the Symbol Subcommittee of the Universal Grocery Product Code Council. That’s the fancy name. I was the technical consultant. These were grocers, food chain CEOs, food producers, etc.

Nebeker:

What years were these?

Eden:

Nebeker:

ISBN maybe?

Eden:

Yes. Then they have the numbers, and it’s an ugly typeface. I was told that seventy percent of all packaged merchandise in the world, especially foods, carries the bar code. Essentially the same one is used in Europe and the rest of the world. It’s remarkably low tech, but it’s everywhere.

### Handwriting analysis by synthesis

Nebeker:

Maybe you can tell me how you got going on the handwriting analysis.

Eden:

It started almost the first day I came on the MIT campus. I went to a seminar on speech recognition. I was sitting next to a chap, Morris Halle, who became a preeminent linguist and phonologist. I leaned over and whispered that it would be a lot easier to study the written words the speaker was putting on the blackboard, because they would stay put for study while spoken words vanished right away. We started almost immediately and had worked out the main parts of the theory in a few weeks.

The original motivation was simply to observe handwriting as a sequence of strokes. That’s what you learn when you’re studying handwriting. When I studied handwriting we learned Palmer method—you make loops and such and you practice. Handwriting consists of these particular sequences. Can you abstract some primitives and then describe how you put the primitives together or modify them in some way to give a description, and then take an English word and translate it into that description? Well, if you can do that then the rest is easy. We had a finite set of symbols in which each symbol with names like bar, hook, loop has a very small number of particular parameters, really only two; also certain common parameters for the whole word, and you just plug it in your computer and let the computer write.

Nebeker:

So your idea was synthesis first, and then do that understanding of the handwriting.

Eden:

We did synthesis first, and then we said okay, now let’s look at a written text—can we use the same system to analyze it? Well, to do it by eye is not too difficult. The technique was called analysis-by-synthesis.

Nebeker:

Your primitives were certain types of strokes?

Eden:

Nebeker:

So from the beginning this program worked as far as the handwriting analysis?

Eden:

We went through two stages. We had the primitives and the rules for adjoining the symbols and how they’re connected. It turns out, for example, that you only have to deal with the down strokes. The up strokes are thrown away; they carry no information. So you go down, and then your hand drifts over some to the right (left in Hebrew and Arabic). We can predict how your hand has to go based on the stroke that you just finished and the one that’s going to start. That’s nice because you want to make a parsimonious theory.

The first writing we did was essentially stochastic writing. You’d go point by point. When we produced this writing, it looked too much like what a kid does when he’s just learning how to write—it’s struggling, very angular. Then we came up with the idea that the motion of the hand is essentially ballistic, and it’s true in all kinds of muscular events. Take your eye movement, for example. Your eyes are moving all the time with so called saccadic movements. During the time that your eye moves, you don’t see. You don’t know that you don’t see, but your eye is doing a lot of moving, and it’s moving the order of a degree or less. That’s not a big movement, but your eye moves. It’s not the same as the movement in which you continue to see; so-called pursuit movement. So if you’re following something and if the speed is right, you can follow it. But ordinarily your eye skips around from place to place. You can tell by analyzing the signal that it’s ballistic. When you start you aim your eyes. You don’t know exactly where they’ll stop, but then when you get close to the goal you can modify where it finally ends up.

That’s what we did with handwriting. Your hand begins to move. How much energy are you going to expend? That tells you how far it’s going to go. How much curve is it going to have? You have to apply an orthogonal force to move it, and then let it go. It gets to a certain point and then you do the next. Well, suddenly all the strokes were smooth and beautiful.

By that time I had a couple of graduate students and we published some papers. They did most of the work. You can do the analysis by trying to fit: take these strokes which now are the mathematical description and correlate with what you think is a stroke in the handwriting of the individual. You modify the parameters accordingly so you can get the parameters for that stroke, go on to the next, and so on. It works very well. Nowadays, especially with regard to generation, you can buy a program. You can buy it for \$50 and it will write perfectly good handwriting. Whether they use our method or not, I have no way of knowing. Going back thirty or forty years ago, things were a lot harder to do, but intellectually they haven’t changed. They were harder to do because the computers were not what they are now.

### Cognitive Information Processing Group; pattern recognition

Eden:

By that time or shortly thereafter I left Rosenblith’s group and set up a group with William Schreiber and Samuel Mason. Mason’s claim to fame was that he invented flow graphs. He was an ingenious engineer with a wry sense of humor. His research was on sensory aids for the blind. One such, with Schreiber and Don Troxel, was a machine that read printed books. Bill Schreiber came to MIT from Technicolor. He has worked mostly on picture processing. In recent years he has been one of the leading experts on high definition TV.

So we set up this lab. We had a number of graduate students who later became faculty including Troxel, Tom Huang, Oleh Tretiak, Jon Allen, Ted Young, and Barry Blesser. After a while we were joined by Paul Kolers, a psychologist who had been at Harvard and apparently got in a big battle with Jerry Bruner or some other tenured psychologist there. We couldn’t get him an appointment at the MIT Psychology Department, but he stayed with us a while and he chose our name. We became the Cognitive Information Processing Group, and it was that from about the early ’60s until long after I left in ’76. We had some strange people. We had a psychiatrist for a while who tried to model very disturbed children.

Nebeker:

Can you give some examples of things you and the others in this group worked on in those years?

Eden:

Yes. Most of the things that I worked on had to do with biology. That’s how I came back to biology; I had sort of drifted away. Most of the pattern recognition problems that we proposed for students to work on came from medical applications, diagnostic applications, and we really worked on a whole range of problems.

Nebeker:

So because of the handwriting analysis, you thought some of the same things could work for--

Eden:

Well, basically. We look at patterns, we know they are patterns because human beings can look at them and say, “Ah ha, this is so-and-so; a dog, a hat, etc.” In some cases you could make up a theory, but there’s a difficulty here. For example, we don’t know why the chromosomes look the way they do. They’re not a formal system, whereas handwriting is a formal system—long before I got involved there were teachers who taught you how to write handwriting and there were people analyzing handwriting. One of my students could read sheet music. It had to be simple, but again, the only reason it had to be simple is because our computers weren’t that complicated. Music has a definite structure. It’s been invented by people, and we can understand that. If we understand it, and if we can cast it in mathematical terms, then we can read or write or play music, which we did.

In the case of most other things, like biology, you can’t really start with a full-fledged theory. You have to first find out from the physicians or the biologists what they know about the structures. Chromosome karyotyping is a very good example. When we started, it was very difficult. We knew very little other than each chromosomes size and shape.

Then a biologist by the name Caspersson, if I remember correctly, discovered that you could differentially stain the genes in the chromosome so you’ve got a bunch of transverse stripes. Suddenly the problem changed because we now had additional information related to structure. We looked for debris in urine. We looked at cancer cells in papanicolau stains. We looked at x-rays to tell the difference between tuberculosis or not.

We were not the only ones. Jack Sklansky at Irvine was one of the people who did a lot of work on biological patterns, especially x-ray images. Another person who had something to do with it but was not deeply involved in biological objects was Herb Freeman, who was at RPI. Freeman worked on a whole bunch of pattern recognition problems. He worked, for example, on how to optimize patterns on a plane or on a sheet in order to cut material for tailors without leaving too much material.

Nebeker:

These pattern recognition problems you said were mainly in the medical or biological area. How did they come to you? Was it a case of graduate student needed a project?

Eden:

Most of the time. There were relatively few things that I did on my own. Sam Mason died in the late ‘60s and I became the group leader. I’m not a good boss in the sense that I don’t know how to make people do what I want them to do. I am conscientious, so I tried to talk to the graduate students about what their interests were and try to get them to work on a problem. Like most professors, I’d read the topics to them and they would make some decision one way or the other.

A lot of engineers have been fascinated by the notion that what they are doing is for the good of humanity. There was a period of time (and this was true in the ’50s, and probably in the ‘60s and most of the ‘70s) when a lot of engineers were interested in biomedical engineering for that reason. Nowadays it’s different. You talk to the kids nowadays and they want to start a company and have an IPO and then make a million dollars and that’s it. More power to them! But they have no particular awareness that what they’re producing is clearly irrelevant to the good of society.

Another person who worked in part on biological problems was Azriel Rosenfeld in Maryland. He was not an IEEE member, but he certainly had worked in pattern recognition, in part on biological objects. Most of his money came from the military, at the time when the military was free with their money. Most of his applications involved recognizing tanks from pictures and aerial surveillance and such. There were a dozen places where people were doing pattern recognition. Primarily in biology.

At the same time, I had an appointment at Harvard Medical School. I taught a course called Biomathematics to second year students, and worked on the logic of medical diagnosis at the medical school. Again, diagnosis is a task where patterns are to be found if you know how to look for them. We spent a lot of time interviewing good diagnosticians and recording what they said, then tried to make sense of what they said in terms of what came to be known as expert systems. Our interest was less in the fact that we could do diagnosis, which we could do by computer about as well as the average doctor, but rather what is the structure. There’s a decision tree to be worked out. How does it work? How can we find them?

Unfortunately, we never published. We had only one small related paper that came out in the New England Journal of Medicine. The big problem was that every physician is different, and so they invent their own set of algorithms whereby they come to a diagnosis. It was obvious that it couldn’t be a big study. We may have interviewed altogether a dozen physicians, each one diagnosing about two or three standard cases. This is very labor intensive, and it was done on a very small budget. What I got out of it mostly is that I could sit with the output and figure out why this doctor was doing this or something else most of the time. But they were all different, so there didn’t seem to be any general structure.

Again, that’s a pattern recognition problem. Expert systems are now almost standard, not so much in the actual practice of medicine but in the training of physicians.

### Commercial applications

Nebeker:

Did any of these pattern recognition studies that you or your students did in this period result in a program that was either commercialized or used in hospitals or by researchers?

Eden:

Not to my knowledge.

Nebeker:

So these were contributions to the science of pattern recognition?

Eden:

I never have been terribly interested in knowing the relation of my work to the commercial world. We started a company, for example, for doing differential white cell counting using pattern recognition. In a way, we were beaten by the fact that by using flow techniques you can discriminate the common classes of white cells, so you didn’t have the trouble of making slides. But our system worked, and it was developed originally within my group. Lester Smith, an engineer entrepreneur approached us and he wanted to set up a company, so we set up a company. We were very naïve. He got a ridiculously small share of founder’s stock. He ultimately sold out to Wallace Coulter, who was the inventor of the Coulter Counter, which was a competing device.

So I had very few ventures into the commercial world. One time I had a student who invented a very good way of detecting forgeries, again based on the handwriting. It worked remarkably well on forgeries in which you could get the time sequence. If you’re writing with a light pen or a mouse, you can follow the time course as well as the spatial movement. In consequence, it becomes relatively easy to tell when somebody’s forging, even though when you look at it looks exactly like an authentic signature. I had done some consulting for venture capitalists in Providence, and I said, “To build the first model, it would cost about \$500,000.” We didn’t persuade them to invest. We didn’t do anything.

Nowadays there are a dozen companies that have methods for determining security, authenticating handwriting, or eye grounds or fingerprints. We worked on fingerprints and a whole bunch of things. But one of the issues in almost any of these patterns that have commercial value is that you have to be able to demonstrate in a practical way that it will work. I had no great interest in it. Most of the students at the time didn’t have any great interest in it either.

Nebeker:

It was a long way from a prototype system to product.

### Computerized tomography, image processing

Eden:

Right. We independently invented computerized tomography. I think it’s fair to say we did it independently, although [Godfrey] Hounsfield at EMI may have begun a year or two earlier. We published the first section of a human tissue in the 1969 international conference on engineering in medicine and biology.

Bill Simon called me. I had known him before. One of the things that I didn’t mention before was that I was involved with a group from Lincoln Lab who proposed what was called a LINC computer. It was supported by NIH, and was really the first small computer designed for laboratories. We rented space in one of the buildings in Kendall Square, and these famous biologists came to our site and were given the modules, and taught how to put them all together. The principal designer in that development was Wesley Clark, who had been in Lincoln Lab and had worked on Rosenblith’s CAT that I mentioned earlier. As a matter of fact, most of the staff were from Lincoln Lab.

So that was a place where there was an intersection between my interests and Rosenblith’s and biology. We pointed it that way. We were interested in seeing how you get computers into the hands of the biologists. Of course, they had been using a lot of amplifiers, recorders, and some primitive analog devices. Many years earlier (in 1950) I had an analog system for doing division. We had a mouse living in a little bell jar and we could measure his oxygen consumption and CO2 production, and compute by division his respiratory quotient. These are primitive kinds of computation devices. By the time we talk about the Linc computer, which was at the end of the ’60s or ’70s [1962-69], computers were coming along at a great rate.

Nebeker:

Yes, earlier you had maybe one big computer and you had to share that, and then we went to having your own mini-computer in a laboratory.

Eden:

We had a schism between the people from Lincoln Laboratory and people at MIT. The Lincoln Lab group wanted a number of professorships. I really can’t speak for MIT, but there was this kind of bickering going on. Their salaries were substantially larger at Lincoln than ours. In any event, they agreed to stay for the pilot program and leave. After they left, the project folded. At the time it was the largest grant that NIH had ever given. But Charlie Townes, who was the provost, said, “Well, we’ve got to give them back the money.” So we gave up on that after the couple of years in which the Linc Computer were being built.

Most of them went to Washington University in St. Louis. It was a sensible move because Clark, one of the principals, his associate, Bill Papian, and almost all the others were interested in biology. Wes had been working even earlier on this computer of average transients. But they went off to Washington University and got appointments there. Charlie Molnar, who was a graduate student in Rosenblith’s lab also went to Washington University where he was involved in developing modules that could be put together to make a computer. He died probably ten years so, a relatively young man. So that was part of my background.

I mentioned the sectioning, tomography. Bill Simon called me. He had been in the group at Lincoln Lab, but he decided not to go to Washington. He had developed a little magnetic tape memory for Linc. He went to the University of Rochester in Bioengineering. He called me up one day at MIT and said, “If you were to take a bunch of pictures of a biological object, can you reconstruct a section?” I said, “Well, let me think about it.” I thought about it for two minutes. There’s a well known theorem in mathematics called the Radon Theorem or the Radon-Nikodym Theorem, which tells you that if you can obtain a sufficient number of two-dimensional or more generally n-dimensional projections, you can, up to a certain level of accuracy, reconstruct a section in any lower dimension, or reconstruct the n plus one dimensional object. A mathematical theorem may tell you that something can be done. The theorem doesn’t say how you’re going to do it. I started thinking about how to do it, and I wasn’t getting anywhere. I talked to Oleh Tretiak, and within ten minutes or less he said, “Okay, this is easy. This is Fourier transform theory.” Well then, he’s right. It’s really easy to reconstruct sections using Fourier transforms. Curiously enough, the Fourier technique, even though we used it in 1968 and 1969, was not used in computerized tomography until much later. Somebody rediscovered it.

To finish off this sad story, Tretiak worked it out and we put it in our conference abstract. I leaned on him and said, “This is a golden opportunity. Let’s do something with this.” We talked to the people at Peter Bent Brigham Hospital Radiology Department. I have connections with all these people because I think at the time I was still on the faculty in the medical school as a lecturer. I knew the head of radiology, Herb Abrams, because I knew his brother as a grammar school student. (His brother, incidentally, is well known in the movies in Hollywood. His name is Mason Adams.) We went over and talked to Herb Abrams and his staff. They were interested in angiography. They wanted to look at skinny little blood vessels in the hand or the heart or maybe elsewhere. Our pictures were very crude. We could only digitize in a matrix of 32 x 32 or 16 x 16—very small numbers. We could have done better, but it would have required more computer power than we had readily available to us. They looked at our pictures, which were full of quantizing noise, and said, “Well, we have no interest.” Now it’s quite clear that at the same time the people in England were pursuing the same kind of thing, Hounsfield at EMI and his hospital associates. But what he knew that we didn’t know is the requirement is to study the brain. We took a picture of some bones, and we could demonstrate the difference between the marrow of the bone and the hard material, and we could do it anywhere up and down the bone.

The Brigham radiologists discouraged us, but I still didn’t want to give up. We had some connection with one of the x-ray companies, and I talked to their director of research and said, “Would you be interested?” He said, “Well, yes, but you really need a better object. Will you make other pictures?” I went back to Oleh and said, “Oleh, let’s develop this.” He showed no interest. Maybe he was thinking of moving to Drexel. I’m not sure. So we abandoned that project.

But I mention it as the kind of thing that moved my interest in some measure from what you might call pattern recognition to image processing. There is a relation between them even though they’re ordinarily separated in engineering journals. In usual circumstances, the image processing has to precede the pattern recognition. But even then there is a problem. If you don’t know what you want to process in the image, then what are you going to do with the image? You have to know what you’re looking for, and basically that depends on your intuition about the pattern that you’re looking for.

Eden:

Among other appointments I held, I was on the BEIR Committee, the Biological Effects of Ionizing Radiation, which is run out of the National Research Council. The NRC is the working arm of the National Academy of Science. I’m not a member of either academy, but that’s neither here nor there. While sitting on this committee I got to know these people. In 1975, I got a call from a physiologist named Ed Rall, who I’d known for many, many years in NIH. He is still around, though long retired. He said, “Would you be interested in coming to run our bioengineering program?” There was a connection made. I was looking for a sabbatical. I said, “Well, I might be able to come on a leave of absence. How do you feel about that?” He said, “If you can come for two years that would be alright.”

I came down and interviewed with him. I talked to the Scientific Director, Hans Stetten, who again was a good friend of mine at NIH (I spent a lot of time at that institution!) and I said, “What are you looking at me for? I’m not a horny-handed engineer. Sure I come from MIT, but I’m mostly a paper and pencil engineer.” He said, “Well, we’re trying to improve the intellectual quality of our engineering group.”

The engineering group, up until about that time, actually he left a little earlier, was run by Lester Goodman. Les Goodman’s name is important, especially with regard to bioengineering development. Among other functions he was the first chair of the AEMB (Alliance for Engineering in Medicine and Biology). He died, sadly, from a brain lesion or other neurological problem. He was a mechanical engineer, and very definitely a hands-on engineer and very proud of it. He had been the head of bioengineering at Case Western, so he was no slouch, academically speaking. But he really wasn’t that interested in supporting theory. He was gone from NIH to Medtronics, so I came in and ran the program. My title was Director of the Biomedical Engineering and Instrumentation Program.

Nebeker:

This was not within a single institute, but for NIH as a whole?

Eden:

No, it was in an organization called the Division of Research Services.

Nebeker:

Which serves all the institutes?

Eden:

Well yes it serves all the institutes, but in the table of organization it stands as an institute, so it’s at the same level. But we were engineers, and we worked on problems that derived from the intramural biological community. The intramural program at NIH is a very large program, but it still represents only about eleven percent of the money of NIH. It represents probably eighty or ninety percent of the bodies there. It’s a big program of about ten or eleven thousand people, about twenty-five hundred or three thousand professionals. These were our clients.

But in any case, we changed the character of engineering at NIH somewhat. We brought in many more Ph.D.s. I think at the high end and probably now there are about forty or so people at the Ph.D. level. There are a few who are not Ph.D.s, but they are independent engineering investigators. The total body count was about one hundred and thirty-five.

Nebeker:

Eden:

Yes, that’s right. I basically became an administrator. I made a mistake almost from the very beginning. I could have very well established my own group and spent a fair amount of time with my own group, but at the beginning I said, “This place needs some stimulation and I cannot go do my own work, essentially in competition with some of the people in my program.”

Nebeker:

When did you start there?

Eden:

I think it was in the summer of ’76. I arranged a two year leave of absence with MIT, and when the two years were almost up I came back to MIT. Pragmatism has its own place. I knew that the government pensions were calculated on the basis of the average salary of the highest three consecutive years of service in the government. By the time I came back to NIH, my salary was literally about twelve times the highest salary when I was there twenty years earlier. So if I could stay for three years, and indeed I did, it would make a tremendous difference in my pension. So I went back to MIT and talked to the vice president for research, and he said, “Look, if you want to you can take a leave of absence for two years, we will be willing to give you another leave of absence for a year, and then you can decide.” So that’s what I did.

But by the end of the three years, again for reasons that are not clear to me, I decided, “Oh, I love it here. I’m going to stay.” And I did. But things deteriorated. The world changed and I got a new director in the division, a woman who had been Director of Boston City Hospital’s Clinical Research Center. She and I battled a good deal and I just didn’t like it. By this time I was well over seventy and I said, “Okay, the hell with it. I’m going to retire.” I retired in the spring of ’94. It’s a very sad thing; I don’t like to talk about it very much. My program was essentially destroyed. A few years thereafter NIH administration took my program out of her control. They are currently trying to build the program up again, but most of the good people left. I did ultimately establish my own group in image processing. Of that group, two are now professors in Switzerland. One is a professor in the Hong Kong University of Technology. The other is Brescia in Italy. So they’ve done very well. I had a Russian, Leonid Yarolavski, an older man with a considerable reputation, but since he was a foreigner he had no tenured appointment. I wanted to keep him, but my director refused to reappoint him, and that was the straw that broke the camel’s back. He was an incredibly important figure in imaging processing. Had he been an American, he would be on top; there’s no question of it. He is not that experienced in bioengineering, but in image processing he is a world figure. He is now a professor at Tel Aviv University. He couldn’t stay, so he went to where he could go.

So that was basically what happened to my program, and it was sad.

Nebeker:

So ’93, ’94 you left?

Eden:

I left in ’94. We stayed in Bethesda and life was very pleasant. But there were several reasons why we returned to Boston. One, because we had no family living in our neighborhood. My kids are scattered all over America from Juneau, Alaska to Austin, Texas. I have seven grandchildren. We have grandchildren old enough to have kids of their own, but so far they haven’t done anything. I still had an office as a scientist emeritus at NIH, a little, teeny office. The atmosphere was not very conducive for work and I hated to go there. I still had, and to this day still have, a lot of friends in Cambridge and in environs, not only in electrical engineering, but in all the departments: in Economics, in Linguistics, in Marine Engineering, Aeronautics, Physics. I knew a lot of people there. You mentioned Jule Charney. Sure, I knew meteorology. I knew Frank Press, who was in Earth Sciences. I had a lot of good friends and many of them were still there. Very few people left at MIT once they became professors.

So we decided to go back. It was a little traumatic because we had to sell the house, move. At our age it’s not easy, but I don’t regret it. I do feel very comfortable. It is true that it’s very hard to get any work done because I’m now entirely by myself. I have no real facilities at MIT. Whenever I have an idea, like optical illusions-- it occurred to me that nothing really serious has been published on optical illusions and its relation to perception since Richard Gregory in the late ’70s or the early ’80s. Gregory is a philosopher psychologist and a very talented writer, an Englishman. There’s nothing like going to the library and reading literature. There are still people publishing papers on illusions.

Nebeker:

Someone you could imagine doing some work for?

Eden:

No. I’m taking voluminous notes. I mentioned I’m going back to Gombrich’s Art and Illusion, reading it. I started with the old literature—I have books at home—and then I decided what has happened in the last twenty-five years or so since I looked at that literature. I go to the library and read the papers and, sure, there are little things that they’re discovering, but nothing fantastic. Most of the interest is not in the psychology of perception, but in neurology. Now it’s FMRI and PET and so on, you can map what’s happening in the head of a subject as he thinks or as he looks or whatever mental activity you ask him to perform.

Well, that’s relevant, but I don’t think it is as relevant as the current modish thinking. For example, people have studied the difference in giving an individual a hard mathematical problem compared with the brain pattern if you give an individual an easy mathematical problem. The regions in the brain which are active are quite different. What does it tell you? You can localize these regions, so ultimately you may be able to know more about the anatomical detail. But I don’t know that it tells you very much at the behavioral level. This is hard and this is easy. So what else is new?

Nebeker:

I’m afraid we have to stop to get to your training.

Eden:

I meant to talk about IEEE.

Nebeker:

We’ll get that information, but I don’t want you to have to race.