Jan. 17, 1973: pages 44-86

Jan. 17, 1973: pages 44-86

Jan. 17th, tape 1, side 1 pages 1 through 43
Jan. 17th, tape 1, side 1, pages 44 through 86
Jan. 17th, tape 1, side 2, pages 87 through 166

Jan. 18, tape 1, side 1, pages 1 through 43

Jan. 18, tape 1, side 1, pages 44 through 55


In 1972, Joseph Desch, Robert Mumma and Donald Eckdahl of NCR were approached by the Smithsonian Institution and asked to donate their recollections to its History of Computing project, now part of the National Museum of American History. Desch and Mumma were interviewed on the 17th and 18th of January, 1973 by Henry Tropp.

[Page 44]

the optimum base is e.

HT: Yes, well you know, it’s rather interesting that that number e keeps coming up over and over again. And, when I was talking to John Atanasoff about the work he was doing at Iowa State in the same time period, he said he had done an analysis and this would have been about ’38 or ’39 – about the

JD: That was before I did it.

HT: yea – the optimal number base. And, to his surprise — maybe not really.- but it also turned out to be e, which then you had your choice of two or three as your working base

JD: That’s right.

HT: and he chose two and you could have chosen three just as easily.

JD: That’s right.

HT: It’s just that two is easier to work with; the binary .. the binary mode was easier to handle.

JD: That’s right.

Well, you know the funny thing is that as late as a year ago someone sent me photostatic copies of three articles written – no, it was Bill McDermott got it for me — He got — he got

[Page 45]

copies of certain sections of books or magazines where the professors who wrote the articles arrived at the same thing.

HT: Aha.

JD: Just a few years ago they were working on it.

HT: Right. Just. It’s an independent kind of thing because I doubt that anybody published it, so everybody was doing it pretty much on their own.

Just to identify this document more carefully, I mentioned while you were talking that it was dated October 24, 1945 to Mr. H. M. Williams, who was then Vice president of Engineering and Research here at NCR. And it’s labeled, because of the current litigation, .. NCR Deposition Exhibit Number 69, in the .. suit, I guess, between Honeywell and Sperry Rand.

JD: That’s right.

HT: Is that right? That gets an identification on it. .. You know you’ve raised so many interesting questions while you skimmed through. For example, .. and I’m not going to be chronological, I’ll just rattle these off as they come to me.

JD: You see, I was trying to be chronological in this thing.

[Page 46]

HT: Yeah. Like this work you were doing for the Unversity of Chicago. Today we know all about the project, Fermi’s work at the University of Chicago, during that period. But, we know about it as kind of a result rather than the stages of development. But what were they using electronic counters for in their work?

JD: Measuring radioactivity. Now, they used Geiger Counters to .. feed the device that I brought there and to count the impulses. And they were dealing with pretty high levels of radiation and they needed very high speed counters. I didn’t know what I was building; you see, the contract I had was with the Bureau of Standards.

HT: [Laugh]

JD: But when they told me to deliver it, they told me to deliver it to the University of Chicago.

HT: Excuse me.

JD: Now, this was separate from this business of the megacycle or the recycling counter, where I had been to Chicago before with Dr. Wilson, but not on this project, on another project. And when I got there, the Marines were so thick around the place, it made me suspicious.

[Page 47]

HT: [Laugh]

JD: So I delivered the thing. Prior to delivering it, Wilson came – not Wilson but another — another doctor of physics, and I forget his name now, came down to test it. And he comes in with a Geiger Counter tube about that long [laughter].

HT: [Laughing] About — about a yard — about a yard long and about four inches in diameter is what

JD: [Laughing] And he hooked that thing up and started counting cosmic rays.

HT: .. In other areas of — of interest, did people discuss or talk about the need for faster and better, and more volume of computation?

JD: . .

HT: Just — just general scientific computation. Was that in the air at the time you were doing this work?

JD: You see, that was the motive for MIT’s work.

HT: Right.

JD: They — they wanted a rapid arithmetical – they didn’t want a machine that would do everything that I wanted it to do in — in a commercial field. Their use would have been much simpler, but speed was of the essence.

[Page 48]

HT: Mhm.

JD: Now, they would add and subtract, but whether they would multiply or not, I don’t know. They probably would have to multiply, but I’m not — I’m not too sure.

HT: I would say, if they got a high enough speed then you could do your addition as – I mean your multiplication as just repeated addition.

JD: You’ll find in that same document that we just finished reading, you’ll find . . . what we sumitted to Warren Weaver of Section D4. Well, I’ll tell you what happened to make it totally clear on the record. I saw it a moment ago as I went through. I’m again referring to this document of October 24, 1945. And, by the way, there’s a list of all the documents here.

HT: Hhm.

JD: I gave them a list. Here’s [shuffling of papers].

HT: I didn’t see that

JD: Now, this is a — this is a multiplier – add, subtract and multiply device – and here is the principle by which it operates. It operates by — by addition, . multiple addition and then shift, multiple addition and shift

[Page 49]

HT: Mhm.

JD: under control of the multiplier.

HT: Yes, much like one of the early mechanical

JD: Yes.

HT: machines where you

JD: That’s right.

HT: put the number in and turn the crank.

RM: But this operated very fast.

HT: Yes.

JD: And, there’s a description of it here.

HT: I saw your name on this, Bob. Was this one that you were working on?

RM: Yes. This is the one – see, this is the one that was, in fact, in interference with the ENIAC and, in fact, they knocked out several claims of ENIAC because of the earlier date. It is covered by U.S. Patents Nos. 2,442,428; 2,495,075; 2,503,127; 2,524,115 and 2,533,739.

JD: That’s right. That is a museum piece right there.

RM: Then they reused this machine and ran the machine in Minneapolis. It still runs –

HT: Aha.

JD: And we multiplied, added, and subtracted, for the judge up there just a few years ago,

[Page 50]

HT: Aha.

JD: or last year.

HT: I gather you can do division then by continuous subtraction if you wanted.

RM: Well, you could.

HT: No?

RM: This was — this was designed for division. The tubes were in there. But, we – there again, bcause of war work, we never finished it.

HT: Aha.

JD: It’s a tougher job to do than multiply.

RM: Yes, but it had the quotient accumulator, and it, had the control section built into the system, but not checked out.

JD: That’s a picture of it there.

HT: Now, are these — are these all thyratrons?

JD: Yes.

RM: They’re all thyratrons. Everything we’ve done was with miniature thyratrons.

JD: Now, here is a description of how this works.

HT: So, you were working up to six digit numbers on this?

RM: .. Well, we had an eight digit, an eight digit accumulator.

[Page 51]

HT: I’m sorry. That one is covering it up. I’m sorry. So, it does run to ten to the seventh, right, so you did get into, into the eighth.

JD: Now here is the reason for this thing .. .. being in here the way it is, giving it to my boss. My boss saw this and its operation and all that, but he couldn’t — what happened was this. Warren Weaver set up a committee, because one of his projects .. .. was to develop an artillery, a field artillery computer, electronic, to be right out alongside the guns.

HT: Mhm.

JD: And, and this was heavy artillery. So he called upon the Arma Corporation, MIT, ourselves, Eastman Kodak, Bell Telephone, RCA, and – I think there was one more company – and we were all to present our present position,

HT: Mhm.

JD: our present technological position, and in return we would get the position of all the other people.

HT: Mhm.

JD: We’d all just play musical chairs with this thing. So this is what I turned in to Warren Weaver. . With it, .. you notice that there are no – quantities

[Page 52]

are mentioned; I mean there’s no parts values specified.

RM: It’s a system.

JD: It’s just a system description

HT: Mhm.

JD: rather than a “how to build it”

HT: Right, right.

JD: sort of thing because it was, at that time it was, not only classified, but it was patentable. So, here is a description of how it works if you want to go over it. .

HT: Right. These are the half dozen or so pages just preceding this [inaudible]

JD: That’s — that’s all.

HT: . document. Now this was —

JD: This came about as a request from Warren Weaver, Section D4 of NDRC.

HT: Mhm.

JD: I think it’s D4.

HT: Well, you mentioned the ENIAC .. clash, the — the patent clash. And I guess an appropriate question at this time, although we’re chronologically jumping ahead, .. because you were so involved in — in classified work during the war, when did you

[Page 53].

first know about ENIAC in its plans or construction stages, or at — at what point of your development did you find out that the Moore School was doing this, making an electronic device?

JD: It’s going to be hard to tell you a date.

HT: Well, just approximately.

JD: It was after we had finished our war work. And were you along on that trip to Philadelphia, Bob?

RM: Yes, I was along on a trip to Philadelphia, yes.

JD: We were invited by Eckert and Mauchly to visit them in Philadelphia to determine our interest in acquiring their corporation. We assembled a quite large delegation to make the visit including engineers, salesmen, financial experts, patent lawyers and management personnel. Eckert-Mauchly was interested in selling their operation to us, their product being the UNIVAC I. Our observation of ENIAC was a mere pass thru without any explanation.

HT: Oh, so it had already spun off from the Moore School?

RM: Yes.

JD: And we were invited to come down to see if we were interested.

HT: I see.

[Page 54]

JD: And while we were down there, Eckert and Mauchly took — took us all over. That is, not only myself but out entire visiting group. They just marched us through with no explanation or anything else, but I saw racks and panels galore and what I took to be registers. They had at least 24 tubes in ’em or more. And no explanation, no nothing. We just saw it. I wasn’t – due to its size and the complexity of it and I had something simpler myself, I — I wasn’t very much interested in the Eckert-Mauchly machine which — which machine operates very similar to that multiplier really. Excepting that it is internally programmed; whereas ours – you turn a switch to go from add, subtract to multiply.

HT: Mhm.

JD: Now we were – I was interested and tried to get the corporation to buy Eckert-Mauchly, because Eckert-Mauchly was using a mercury delay line memory and .. and it was a much more sophisticated machine – UNIVAC I – much more sophisticated machine, I thought, than ENIAC. And it — it — and it set the pattern more nearly for what we now have. And that is, we do have memories that are not mercury

[Page 55]

delay lines, but we have disks and tapes, and so forth, that can be used, whereas I don’t even know what — what the memory was on ENIAC.

HT: ENIAC essentially had 20 words of tube memory.

RM: Were there just tubes?

HT: Yeah.

RM: Just — just actual — actual tubes, right?

HT: Mhm. They had 18,000 tubes in –

RM: Sure a power house.

HT: maybe — maybe even more.

JD: See, I never knew that.

HT: Yeah, I was just curious because of your high level of security classification and the things that you were on. I was curious as to whether or not during the war you knew of this,

JD: No.

HT: this development because it was being done for Aberdeen .. as well as your own work.

JD: Right now .. there was really – there’s been a word said to me – and I can’t tell you by whom anymore now because I don’t – didn’t pay much attention to him – we gave monthly reports on all our work and all of my reports are here, if you want to read ’em,

[Page 56]

HT: Mhm.

JD: covering all of this equipment that’s listed in this — in this thing we were just talking about, this October 25th document. Those reports contained the detailed operation of the machines, including the one that went to Aberdeen.

HT: Mhm.

JD: And we said – and if I could get my hand on those right now I could tell you how many copies we made. We used to make up to 35 copies, and send it down to them. [Recorder off]

HT: I was going to say in any event we’ve looked at some of the reports and they run from 13 to 36 copies for each of them.

JD: Yes. Yes. Now, we submitted those, whatever the number they requested. And some things were more important for them to disseminate than others And, someone had – has said to me that the chance existed, whether it was taken or not I don’t know; that since — since .. people at ENIAC, at the Moore School were doing NDRC work, they had access to these copies.

HT: Mhm.

JD: And I didn’t even know they were doing — doing any

[Page 57]

work, so I didn’t get any of their copies,

HT: Yeah.

JD: but — but if ENIAC went to Aberdeen, they got an awful lot of information if they got the report.

HT: Well, ENIAC didn’t go to Aberdeen, of course, until long after it was completed.

JD: Yes. But, they could have been getting this information from me because the report contains a lot of the problem areas that had to be solved.

HT: Mhm.

JD: So – but, I’m not saying that that happened.

HT: Yeah.

JD: I’m saying it could have happened;

HT: Mhm.

JD: whether it did, I don’t know.

HT: But your first actual contact with John Mauchly and Pres Eckert occurred after the war, during this .. period when they were interested — when they were looking for people to buy the company?

JD: That’s right. Now I can’t tell you what year that is because I –

HT: That would be the late –

JD: don’t even have any correspondence to, to –

HT: Yeah. It would be the late forties or even possibly into 19– well.

[Page 58]

JD: It would be between ’45 and ’50.

HT: Yeah, yeah or very near the end of that period .. near the tail end.

JD: Well, I’ll tell you why it would be near the tail end; because we were out looking for companies right around ’50, because we did buy one in ’52.

HT: Right.

JD: CRC.

HT: Right. Let’s get back to — to NCR and this early period, in terms of this research that you were doing, and your staff with — with Bob and other people, .. beginning to do this developmental work. .. In terms of just general vision and attitude within the company before you got all involved in wartime work, .. how did you see this material, or this development, fitting into your postwar developments?

JD: [Pause] Kind of .. that’s a very good question. [Laugh]

HT: I guess I’m trying to ask an impossible question of both of you and that’s why this is, you know, an informal thing, because here you are, a war is about to break out, or has broken out. You

[Page 59]

are all involved in these things. You have started a development, a research development within the company, and for a while, as you said, they were kind of going along parallel until you get totally involved in the — the wartime needs. .. Was there, in that early period, any thoughts about “gee, we’ve really got something going here that as soon as this war is over, here are ways in which we can implement it, or here’s directions we ought to go?”

JD: Yes. During the — during the period when we were .. locked up by the Navy, we certainly were doing an awful lot of thinking of what we were going to do after the war and how we were going to use it.

HT: Mhm.

JD: And,.. .. the – this machine that Bob .. .. redid after the war – that is, he modernized it

HT: Mhm.

JD: by putting in new tubes and, and correcting a lot of things that should have been done, but we didn’t have time to do before the — before the.. .. all Hell broke loose on December 7th. .. He – during that period, one of the things that we wanted to do was, was get on the market a computing machine that would – that would do all of these functions

[Page 60]

plus whatever else we’d add to it in the way of paper handling and so forth which we could do easily. .. It would enable us to apply to our accounting machines the process of multiplication and division which we didn’t have

HT: Mhm.

JD: in mechanical machines

HT: Mhm.

JD: and also a higher speed machine that would — that would do everything faster and, also, we could do analysis work by capturing information from cash registers on tape

HT: Mhm.

JD: because we didn’t have cards

HT: Mhm.

JD: and feed that into this machine and do a sorting and tabulating .. .. operation automatically and at high speed.

HT: Mhm. I guess what I’m getting at is that really we have two different environments going, if we think in terms of the ENIAC environment and we think in terms of the environment here, both during the peacetime, pre-war and wartime period because you were working primarily in terms of improving

[Page 61]

a product line that involved cash registers and business

RM: Business machines.

HT: Business machines.

RM: Special purpose.

HT: Whereas in the Moore School environment, or at MIT, .. other places where they were thinking about computation, they were thinking

JD: Scientific.

HT: in terms of scientific computations which automatcally got them into what we now call the stored programs

RM: And general purpose.

HT: and function tables that they could call out information they could get, which meant a, a totally different approach

JD: That’s right.

HT: in terms of ideas. And yet the guts, the basic core of the machine, is still fundamentally an attempt to speed up this process of doing arithmetic.

JD: Now that was true at one point in time. .. However, we’ve — we’ve changed. That’s when we bought CRC.

HT: Right. So, your change occurred about –

JD: Our motive for buying CRC was to get an internally

[Page 62]

programmed machine which we could use for business

HT: Mhm.

JD: as well as scientific work and which was an entirely different approach than our mechanical machine could be.

HT: That’s right. But that’s a whole new thing.

JD: But that came a little bit later than the period you’re talking about.

HT: Right. Right. That came in the early fifties.

JD: Yea.

RM: Actually though, this multiplier came into being as our Computronic which was tied into a 3000 bookkeeping machine which furnished the adjunct for multiplication, automatic extensions and all – that sort of thing.

HT: Mhm.

RM: And this was on the market. Then we didn’t – of course, we didn’t put it on the market until 1959, but put it on the record here.

JD: How many did we sell?

RM: I sup– produced a few — Well, 4,246 machines were built.

JD: I knew there was a big number.

RM: Selling price of $18,000.

[Page 63]

JD: Now that was based on — on the technology that we’ve been talking about.

HT: Mhm.

JD: That’s the only machine we marketed using that technology

RM: Which has those little gas tubes in it.

HT: Aha.

JD: That’s right. It was the only — it was the only machine that we used our own technology, because it lent itself to that. It did a mult- a set of computations, getting its signals from a mechanical machine.

RM: It had a printer on it.

HT: I see.

RM: A printer and a keyboard.

HT: Okay. I’m still, you know, not being chronological, and I’m thinking back as these pictures you went through come to mind, and one of the things I wanted to ask about was the conversion. .. When you first described your prototype it was a decimal machine. Now at some point in time then you decided to do the arithmetic, I – I’m gathering, in a binary mode, which meant then you had to have a decimal input and a conversion and the same with output. Or am I wrong?

[Page 64]

JD: That’s – you’re wrong.

RM: Yeah, we

JD: It didn’t ever happen until we got CRC, in which case, the internal operation of that machine was binary.

HT: Okay. Now.. in this ’45 document, .. you have – let’s see if I can find them

JD: Yes. I — I know what you mean.

HT: quickly, some converters ..

JD: Yes.

HT: binary to decimal converters.

JD: We wanted to build that converter because basically I had in the back of my head that I could build these multipliers and adders and devices cheaper using binary arithmetic

HT: Mhm.

JD: if I could convert it back to decimal, so it could be read by an operator.

HT: Mhm.

JD: So I was trying to develop a — a, an easy, commercial way. You see, even our present-day computers have to do that

HT: Mhm.

JD: because if they’re running in a binary system here

[Page 65]

it’s – and they give you answers in a decimal system, they’re – they’re, they’re conv- they’re converting inside,

HT: Mhm.

JD: but by a different method than I used,

HT: Mhm.

JD: of course. So

HT: But that — that was just thinking ahead then. That was — that was not connected with the operation

RM: Well-

HT: of the arithmetic units themselves.

RM: I think there is a little misinterpretation of terms. Now you’ll find that what they call a recycling binary converter in there – and that is simply a binary counter that would count up to 16. But, when the counter reached the count of 10, we’d automatically set it back to zero and that name was applied – I think you’ll see that —

HT: Oh, I see. So that’s not — that’s not really what I’m talking about then.

RM: No.

HT: That’s not really a conversion unit from

RM: From binary to decimal.

HT: binary to decimal as I was thinking of it. I see

[Page 66]

RM: That — that may be part of it. I’m not sure that’s – but that’s one of the items we had in there was what they called a recycling binary counter

HT: Mhm.

RM: which would count in binary up to 10 and then recycle back to zero.

HT: Mhm.

RM: That’s what Joe

HT: I see.

RM: referred to as the Chicago – the University of Chicago

HT: Right. Right. That was

RM: machine.

HT: the one that I was referring to.

RM: Yeah.

HT: And so, that was actually then in use in the .. cosmic ray counter that you

RM: Oh, yes.

HT: delivered to the University of Chicago?

RM: It counted to a million. We had to use that.

HT: Right. Right. So you did have a conversion unit.

JD: And we also used it in Aberdeen.

RM: At Aberdeen we had to use it. Yeah, wherever we

[Page 67]

encountered a higher speed, we had to go to that because you could not get the high speed out of gas tubes.

HT: What — what kinds of – what were the purposes for the equipment that you delivered to Aberdeen? Were these connected with the firing tables or with the trajectory calculations?

JD: Yes. The ballistic tables that they were using all had defects. What they had, there were two screens. And.. and the shell would go through the first screen it would start the counter and when it would go through the second screen, it would stop the counter.

HT: Mhm.

JD: Knowing the distance between the screens, we could get the velocity of the shell,

HT: Mhm.

JD: and then we could also get the trajectory from that. So.. they had equipment, but it was obsolete.

HT: Mhm.

JD: It was made by General Electric, but it was – must have been 20 years old. And it was working on an entirely different principle. When they saw what we brought down, they were pleased because we

[Page 68]

demonstrated this one megacycle control.

HT: Mhm.

JD: I was negotiating a contract with them when the Navy took me over or we would have built a system,

HT: Mhm.

JD: and delivered it to ’em, but the Navy and the Army had a big argument and the Navy won

HT: Mhm.

JD: because the Navy had a hotter job.

HT: That’s when you went underground essentially? [Laugh]

JD: That’s when we went underground; but they had a hotter job. So,.. I — I didn’t know who was going to get that Aberdeen project. I’d lost contact with it. I think we left our equipment there, didn’t we?

RM: Yes, we did. I’m sure we did.

JD: And — .. .. and so they could play with it, and we never got any calls for maintenance; the thing stayed with it. And, by the way, that was given a good shock test on the way down –

RM: [Laugh] We threw it off the train.

HT: You threw it off deliberately?

RM: Well, we —

[Page 69]

JD: We had to throw it off or it would have ended up in Washington.

RM: We got off the train right along the tracks and had to get down to the ground. I mean they slowed the train down before we got at Aberdeen.

JD: I’ll tell you how it happened.

HT: [Laughter]

JD: Mr. Mumma and I were told to deliver the Aberdeen equipment immediately. We left Dayton in a hurry hand carrying two boxes of equipment. We had train tickets to Philadelphia but none from Philadelphia to Aberdeen. In Philadelphia we transferred the hand carried equipment to a train in the railyard ready to leave for Washington. Bob and I took the equipment aboard the last coach on the train and enroute we paid our fares to the conductor. He accepted the fares to Aberdeen but he told us that the train would not stop at Aberdeen. After some argument he agreed to ask the engineer to make a stop at Aberdeen. The only agreement the engineer would make was to slow the train down to 5 MPH and we could jump. We agreed but when we approached Aberdeen the train speed was not reduced to 5 MPH but to what seemed like 20 MPH. After

[Page 70]

passing the small station we threw the two boxes off the rear platform and then Bob and I jumped. We landed in a cinder embankment and had to carry the equipment back to a station wagon waiting for us. They took us to the Proving Ground. We were messed up some but not injured. The equipment operated perfectly, something I did not expect after the shock it got when hitting the ground.

HT: Well, that – let’s see you got involved in this, you began in ’38 and, of course, by ’41 then, you had – if I’m recalling correctly, you had that first prototype, and then a second model,

RM: Yeah,

HT: an improved model –

RM: and had — had the multiplier done by that time. Finished that –

HT: Mhm. Okay. We might – might talk a bit about the .. development of these small thyratrons.

JD: And a lot of the other OSRD jobs were finished also by that date. .. The ..

HT: Mhm. By the time Pearl Harbor had occurred?

JD: Yeah. Many of those pictures in there are pieces of equipment that we built before we entered the war.

[Page 71]

HT: Well, on this — this .. problem that you had of building small thyratron tubes, who was involved and — and what kinds of characteristics were you shooting for and what kind of success were you getting?

JD: Well, we went through a long series of, of empircal testing really. .. I used Cone for quite a long while. .. Who else did I use?

RM: Oh, I think we had somebody else, but I can’t think who it was. Of course, you did a lot of the work yourself, too. But – of course, we were — we were fighting to get the gas pressure down, to get the speed up, short deionization time was what we were fighting for, to get that deionization time down, to make these things operate fast and to experiment with various gas pressures, and, of course, we used Argon gas.

JD: A lot of physical configuration and various gas mixtures, too were tried.

HT: To know how to make this tube, you first had to evacuate it.

JD: Oh, yes.

HT: And then you had to

JD: And bake it.

[Page 72]

HT: yeah. I noticed the little – what looked like a small furnace sitting off in – on one side, right next to the sink, in that laboratory.

JD: That was a little furnace; we baked every one until we degassed it, really.

RM: Didn’t we spray our own cathodes, too?

JD: Yeah, we sprayed our own cathodes with barium strontium carbonate.

HT: Mhm.

RM: And then we had to activate those cathodes.

HT: Well, actually then you were making each tube, you know, kind of one at a time.

RM: That’s right. There’s no big manifold. You make ’em up.

HT: Right. You were blowing each one. .. The glass blowing machine is no longer there, but I saw where it was sitting.

JD: Mhm.

HT: And, what kind of success were you getting in terms of uniformity or did each tube have to be tested for its range of characteristics?

JD: We finally reached a — a very — very good yield.

RM: Very successful.

JD: We finally got so we could turn ’em out pretty fast.

[Page 73]

HT: Mhm.

D: But in the beginning, we were doing a lot of research work trying to find the best combination, the best characteristics. Some of the problems that we had involved .. not only the speed of ionization but of deionization. ..

RM: We had this leak —

JD: At one time we had trouble with the thyratrons having a short life. We finally discovered that if we reduced the value of the negative grid potential that life improved. We deduced that the higher negative grid was causing positive ion bombardment of the oxide coated cathode.

RM: Yes. Maybe that’s where you’d get positive ion bombardment of the cathode. Of course, another thing that happened, the cathode evaporated off till you’d get a dark deposit of barium

JD: Oh yes.

RM: on mica insulators and the next thing you know, we’d short out a control grid. And then so the tube’s life was rather limited, and, of course, that wouldn’t do, you know. We wanted more than a few hundred hours of life.

HT: Mhm.

[Page 74]

RM: but – so there was lots of tests. We made different configurations of insulators to break up this path.

JD: In fact we ended up with a double insulator.

RM: Yes, we did.

JD: Yes. And — and to keep this .. evaporation of solid material from depositing and conducting a coating between the grid and the cathode.

HT: Now on this. I think this area of the tubes themselves is worth exploring .. because you felt the need for them and clearly there must have been other people who had a need for tubes of this sort, knowing the kind of work that was going on during the war and immediately afterward. Were you supplying tubes to any — any other groups working?

JD: No.

HT: Because I – again I’m stretching my own memory, because when they were looking for counters for ENIAC, somebody mentioned to me that they thought they tested some of the thyratrons that you were using.

JD: They may have because we might have given them to NDRC.

[Page 75]
RM: Well, of course, one thing happened. By the time we went into the war for the Navy, we got a – we had Sylvania build tubes to our specifications .. that were these miniature thyratrons, and they were the 2C4 and the 6D4.[ed. note: for more information on the 6D4, see Radio Museum or the National Valve Museum.; for the 2C4, see specs 2C4]. By the time we went into the war for the Navy, we got a – we had Sylvania build tubes to our specifications

HT: So

RM: … one was a six volt and one was a 2 1/2 volt tube.

HT: So these were then becoming available to other people through Sylvania?

RM: They were on the market then.

JD: No, I had to pay the bill. I bought 50,000 tubes

RM: Yeah, I know.

JD: during the war and .. .. I put no hold on them. I allowed Sylvania to sell them to other people, too,

HT: Mhm.

JD: so they could have bought them from Sylvania.

HT: So other people could have — could have had these available?

RM: They had standard sockets on them so they could be used by anybody then, you see.

JD: Well, they had a glass base

RM: Oh yeah, that’s right. They had the — they had

[Page 76]

the button base. That’s right. Seven-pin button base. Yeah, okay.

HT: So they couldn’t – they had to be used in a socket that was designed?

JD: The same as the miniature tubes are now that you buy on the market.

RM: And you can still buy those tubes today. At least the six volt type, the 6D4.

JD: Yeah, they’re still on the market.

HT: Check to see how we are doing.

In terms of the – your experimental data with these tubes, how widely was that information dseminated among different research groups during the war?

JD: It was disseminated only to the Office of Scientific Research or NDRC.

HT: So that they were in a position then to tell other projects what you were doing and give them the information.

JD: We — we kept them informed of our position and -and that’s what’s in those reports.

HT: Let’s get into the OSRD and NDRC period. .. How did you, you know, first – well, you actually began before the war, doing research for .. ?

[Page 77]

JD: Well, we were doing research of a — of a nature that was complementary to them, or vice versa.

HT: Mhm.

JD: We were both working on, oh — oh building electronic arithmetic counters. Now they — they never did build out a whole series of them because, being a research oriented group at MIT, .. they they tried all kinds of combinations of tubes from scratch.

HT: Mhm.

JD: I chose a thyratron which existed,

HT: Mhm.

JD: that is, the principle existed but they were -they were experimenting with very exotic electronic . . devices which are adequately described there in their reports in working on altogether different principles.

And, we got to know, during that period, we got to know men like Mr. Overbeck. I don’t know if it’s Dr. Overbeck or not, but Overbeck appeared. .. We .. Coleman heard about us, and his superior, Dr. Harrison, came out here and paid Mr. Williams a visit and asked us to take contracts with them, which we gladly did. And that was the beginning

[Page 78]

of our association with NDRC. At that time it was NDRC.

HT: Right. Let me — let me back off again so that I don’t lose the-questions that keep popping into my head. During this period there were a number of computational projects going on. The ENIAC came along later, but there was the work at Harvard under Dr. Aiken. And there was George Stibitz’s work at Bell Labs. .. I guess those are the two most prominent ones. Did you have any contact with either one of those? ..

JD: Yes. I did not know about Aiken’s work at all. That was another Navy job that I didn’t know anything about. But, with Stibitz .. at Bell Lab, I had access. I saw relay computers.

HT: Mhm.

JD: That’s what I saw.

HT: Right. That’s all they were building during the war.

JD: But, I saw those computers. .. Sam Williams

HT: Right.

JD: was the head of that.

HT: Right.

JD: and he was a good friend of my boss, Harry Williams.

[Page 79]

HT: Mhm.

JD: Two Williamses. And every now and then, we’d drop in on West Street and, and — and have a chat with them and see the latest. And Sam Williams used to come out here, too, and see what we were doing.

RM: Yeah. I met him several times.

HT: Did you have any contacts with Stibitz?

JD: No, very little contact with Stibitz. We never talked — we never talked technical matters. They would demonstrate the operation of the computer and so forth, but they were — they were so relay-oriented, and I wasn’t, that we didn’t have much to really talk about. It’s amusing to see how people have a fixation about a certain thing.

HT: Right.

JD: I had a fixation on thyratrons and he had a fixation on relays.

HT: Aha. .. How about the wartime computational work that was going on at MIT. Who were some of the people you talked to then, after the war began? I know about the earlier associations.

JD: Well, essentially the same group.

HT: Same group.

[Page 80]

JD: Because I worked under the section under Dr. Harrison,

HT: Mhm.

JD: who, I think, was Dean of Physics at the time. .. Dr. Coleman, is now I have learned, is with the National Academy of Science. .. Coleman was a very nice man to work with;

HT: Mhm.

JD: extremely competent.

HT: Mhm.

JD: And then, during the war, I had .. all of the work that was done on those vacuum tubes was under Dr. Caldwell.

HT: Right, you mentioned that.

JD: And, of course, he had also the commercial part of it that was occurring before the war

HT : Mhm.

JD: and then during the war, he was also involved in it, and I was associated with him; and some of the – it, it finally ended though because he took another direction.

HT: Mhm.

JD: He had another program and we quit supporting him at some point in there – I don’t know exactly

[Page 81]

when it was – because he was getting more money than he knew what to do with from the government.

HT: Did you have many contacts with Bush during the wartime period?

JD: Well, I had several nice contacts with him. I knew him very well and I spent a lot of time .. talking about computers and so forth, .. counters and so forth. He himself was extremely interested in the subject and, of course, it was from his analysis viewpoint mostly. He — he was trying desperately to improve the differential analyzers.

HT: Did you have any contact with Norbert Wiener?

JD: No, I didn’t. I saw a lot of him because his office was right next to Bush’s, I think. But — I knew him when I’d see him but I didn’t know him personally.

HT: You didn’t have computational oriented.

JD: No.

HT: conversations

JD: No.

HT: with the

JD: No.

HT: famous professor?

JD: No.

[Page 82]

HT: When did you first meet Von Neumann?

JD: I couldn’t tell you. .. I don’t believe I ever did meet him.

HT: Oh.

JD: . No, it was Norbert Weiner, that –

HT: It was just Weiner that you

JD: Yeah. I didn’t know Von Neumann.

HT: You didn’t cross paths with Von Neumann?

HT: There were – it’s interesting during the war, when so many people were operating on classified projects that were closely related, but they tended to be fairly compartmentalized, and so apparently there was very little interchange, except at the level where people read each other’s reports at — at the higher echelons I would guess.

JD: The only time that happened to us was that case I showed you there on the – Warren Weaver’s .. development of an artillery computer. That’s the only time when we were all in one room together.

HT: Aha. When all the group, RCA and all the others were –

JD: By the way, RCA was represented in that meeting by Dr. Vladimir Zworykin. Zworykin. I knew Zworykin well.

[Page 83]

RM: .. Wasn’t Rajchman in that thing, too?

JD: No, he wasn’t in that meeting.

RM: I see.

JD: I knew Rajchman.

HT: Rajchman was doing a lot of research on tubes in that period of time.

JD: Yes, I know he was. Mostly memory tubes.

RM: I went down to RCA to see Rajchman about the multiplier tube we used … for the Navy. That was during the war, of course.

HT: Aha.

RM: Get more information on it and use it more effectively.

HT: Aha. Well, during the war, and I guess maybe this is worth identifying here before out tape ends. .. Were the two of you off in different directions at various times?

RM: We were together.

HT: You were together during the – during that whole period. .. Do you have any comments you want to make, Bob, about that early adder-multiplier in terms of how it operated and — and what it’s reliability was and — and what your thinking was about it?

[Page 83]

RM: .. Wasn’t Rajchman in that thing, too?

JD: No, he wasn’t in that meeting.

RM: I see.

JD: I knew Rajchman.

HT: Rajchman was doing a lot of research on tubes in that period of time.

JD: Yes, I know he was. Mostly memory tubes.

RM: I went down to RCA to see Rajchman about thabout the multiplier tube we used … for the Navy. That was during the war, of course.

HT: Aha.

RM: Get more information on it and use it more effectively.

HT: Aha. Well, during the war, and I guess maybe this is worth identifying here before out tape ends. .. Were the two of you off in different directions at various times?

RM: We were together.

HT: You were together during the – during that whole period. .. Do you have any comments you want to make, Bob, about that early adder-multiplier in terms of how it operated and — and what it’s reliability was and — and what your thinking was about it?

[Page 84]

All of it was to demonstrate a principle because it was very huge by any standard…the whole program was to prove the reliability…and I think we finally did get more reliability out of gas tubes than you’d ever expect to get.

RM: Well, I think we pretty well covered it. It was, all of it was to demonstrate a principle because it was very huge by any standard. You never proposed to sell anything like that, but the whole program was to prove the reliability and make the thing – make it operate. And, I think we finally did get more reliability out of gas tubes than you’d ever expect to get. Of course, we were always ridiculed pretty much by these other people about the reliability of gas tubes, and they thought we were on the wrong track. I don’t know. Maybe we were. You know, there’s always that rivalry.

JD: Well, everybody was on the wrong track. It’s all integrated circuits now. [Laughter]

HT: Well, not — not really. You know the basic principles were the key thing that have remained fairly constant even though the technologies got more sophisticated. But, if you look at the operation, it’s still a system that you can describe in much the terms you describe your early work. .. It’s just there is more of it, it’s faster. Instead of 20 words of storage, you may have 100,000 or 200,000. .. But .. how

RM: I might mention one more thing. We – during this

[Page 85]

period of time, we did file a patent on a binary computer that – for addition and multiplication and that sort of thing. In fact, I have a patent on it. But I don’t think we built a model of it. I – we did build a – in fact I’ve got my patent up here, the patent on it.

JD: Who did the work?

RM: Well, I think I worked with Justin Compton on it. And I think that you’ll find –

JD: Oh. That is why I don’t know anything about it.

RM: You’ll find that Justin Compton and I were .. .. co-inventors of the thing.

HT: Aha.

RM: But,.. now it’s very easy to see because one gas tube was one binary counter,

HT: Right.

RM: as you well know.

HT: Right.

And it’s fantastic how simple a thing – how it fell out, y’know, and how simple it became.

RM: And it’s fantastic how simple a thing – how it fell out, y’know, and how simple it became.

HT: Mhm. Yeah, that’s why, you know, it’s curious. I — I noticed your analysis that you mentioned earlier on the optimization of the base of the arithm- arithmetic operation, and these early

[Page 86]

machine were decimal. When were you starting to think about a binary operation?

[Sounds of RM walking across room]

JD: I suppose it coincides with the time span of that binary to decimal converter. That’s probably when it first started to tick.

HT: That’s about 1942, ’43, if I remember —

JD: No, it’s earlier than that because it — it occurred before the war.

RM: Yea, this’ll – [sound of pages turning]

JD: You’ll have to get it out of the reports. You see, the reports are all dated and that — that would give it to you.

HT: Mhm. I’m going to turn this off and flip the tape over.

END OF SIDE I

Go to Third Session of the interview | Return to first session