The Manhattan Project

Nicholas Metropolis' Interview

Printer-friendly version
Nicholas Metropolis arrived in Los Alamos in 1943. Shortly after receiving his PhD in physics from the University of Chicago, Metropolis was recruited by J. Robert Oppenheimer to lead efforts in computational research for the bomb. Working under Metropolis’ supervision were John von Neumann and Stanislaw Ulam. Metropolis recalls collaborating with von Neumann and Ulam and developing the Monte Carlo method. The Monte Carlo method is a statistical approach to solve many-body problems. Metropolis also recalls contributing to the development of the MANIAC I computer. Metropolis shares many stories regarding his research and his personal relationships with his colleagues.
Manhattan Project Location(s): 
Date of Interview: 
1993
Location of the Interview: 
Unkown
Collections: 
Transcript: 

Richard Rhodes: There are two particular themes that I am interested in that I know you were involved with very much. Anything else that you remember that you would want to talk about would be wonderful. One is the developing of computing. Los Alamos made a major contribution to the development of computing in the world. The other has to do with the period around the invention of the two-stage thermonuclear weapon. Could you talk about your experience with those things?

Nicholas Metropolis: Well, let’s see.

Rhodes: It probably would help to know how you came here and when.

Metropolis: Oh, well, I came here at the beginning.

Rhodes: In ’43?

Metropolis: Yes.

Rhodes: Oh, I didn't realize.

Metropolis: I think it was [John] von Neumann who suggested, because he was also a consultant to Ballistics Research Laboratory, which had already given a contract to Mauchly and Eckert [Computer Corporation] for the ENIAC [Electronic Numerical Integrator And Computer]. And that was going on at the [University of Pennsylvania’s] Moore School of Electrical Engineering. This is probably old hat to you.

Rhodes:  No, no, not at all.

Metropolis: I see. Okay, well, von Neumann came here, I think the beginning, the very, very beginning, and maybe it was already Christmas of ’45––of ’44. He suggested that Stan Frankel and I look after the ENIAC and see if we can set up the problems there. And the reason for it was, that Johnny had a very good argument for letting us do the shakedown cruise, as it was called, because we could test out about 95% of the control units, whereas the problems that the Ballistics Research Laboratory had were about 25%.

Rhodes: You were calculating the hundred dynamics of implosion, yes?

Metropolis: Yes, that’s right. But it was a very simple problem in the sense that––relatively, it was very complicated. But all in all, it was rather a simplified problem because it represented really a whole list of successive approximations to the reality of the implosion. So later on we brought in [Anthony L.] Turkevich, who was that chap––who was at the University of Chicago for a long time.

Rhodes: He was also a mathematician?

Metropolis: He was a chemist.

Rhodes: Oh, he was a chemist. You were a mathematician, yes?

Metropolis: Well, I had become a mathematician, I think, pseudo mathematician because my training was actually in physics. That’s how I began to know Edward Teller, he and I go back a long ways because prior to that, he was a very good friend of my thesis advisor, who was [Robert S.] Mulliken.

Rhodes: At Cal Tech?

Metropolis: No, no, not [Robert A.] Millikan.

Rhodes: Right. I’ve confused the two.

Metropolis: Mullikan actually got the Nobel Prize [for Chemistry] in ’66, long after I had contributed to it.

At any rate, that’s how it all got started. And then Feynman and I had utilized the––in the old days—I say the old days because we were looking at Marchants and [inaudible] that were tabletop computers, mechanical computers. We set up a shop complete with [inaudible], with repairing these things, because it was very essential that Feynman, who was aware of reliability, felt that instead of taking several weeks to ship these things off to San Diego, we could just set up. And some of the repairs were just very simpleminded.

Then, later on, when the IBMs came along we used them. This was a suggestion of this guy at Columbia, I forget his name.

Rhodes: The IBMs were card-sorting machines?

Metropolis: Yes, they were all individual machines and they were business machines really. But we converted them into doing a lot of problems that we could do. Feynman's idea was to try to get a list of gals to do the comparison with the machines. This was, in a sense, his continuation of the reliability theory.

Rhodes: In other words, they would use the Marchants and so forth?

Metropolis: That’s right.

Rhodes: In comparison to the card-sorting system?

Metropolis: That’s right. That’s right, exactly right. At first, the gals were winning. But then they began to tire a little bit as the day wore on. Then the machines caught up with them and passed them. Of course when they stopped the machines then were able to make a progress in a great way. That’s how I think I got started in computing.

Then, after the war, we used the ENIAC for several problems from Chicago, because Frankel and I went back to Chicago. I went to Chicago. He went there for the first time because he was a Californian.

Rhodes: Sorry, the ENIAC was at Princeton?

Metropolis: No, the ENIAC was at Philadelphia. And then they shipped it down to Aberdeen, “Aber-Daber,” as it was called. There’s an interesting story there. When we saw the whole machine—and it probably would not fit into this apartment, you know, the whole kit and caboodle, because there was lots of air conditioning as well as big accumulators. By the way, one of the twenty accumulators is at the museum here. It’s available, I guess.

Rhodes: To see?

Metropolis: To see, yeah. Well, then they moved to Aberdeen and things worked out okay. Although it had been the prediction of many that it never would do a single addition again, after the war and after the move especially, because it was about 150 to 200 miles to move from Philadelphia to Aberdeen.

Rhodes: And it had all sorts of wire connections.

Metropolis: Wire connections, and things were very carefully done. But it actually worked. It’s really amazing. There was a chap down there––I’ve forgotten his name. That’s a sign of old age.

He was a chap that worked at Aberdeen and was on their staff there. Johnny von Neumann who was very, very quick—I mean, you have no idea how quickly he would infer things and extrapolate them. Well, he was fantastic. Stan Ulam was here during the war, but he didn't do very much during the war. I mean, mathematicians were not as prevalent as they are today. Although Johnny von Neumann was very, very good and very quick and very sharp. He just was a universalist. He was not a mathematician.

[Talking over each other.]

Metropolis: That was just a side issue as far as he was concerned, because he knew so much about physics and philosophy and even things like history. He was very, very sharp. He worked all the time.

Then things went along and the ENIAC was moved to Aberdeen. There, somebody––this chap had a good idea of utilizing some of the function tables, as they were called, in the setting up of problems, because in the old days each problem had to be set up individually. We used jack plugs to connect from the things that ran around the trays, as they were called, that were available at that time. The idea was to try to utilize the computer. It made a lot of sense to do that, because if one could affect this kind of operation, then there would be a background control of the problems that would be set up by the function tables, as they were called then. So, we could just go along and set these switches and thereby effectuate the whole setup of the problem instead of utilizing the jack plugs.

Rhodes: So they're having to rewire?

Metropolis: That’s exactly right, because we had to rewire. That was the old way of running the machine. But then it turned out that Johnny had asked Mrs. [Adele] Goldstine, of Herman Goldstine fame, if she could look into the problem. She concluded that it would be impossible to do it because there was so much background control that she would run out of opportunity to set up the problem.

So I, with Mrs. von Neumann, went down to Aberdeen to look at this. We were going down to do a set of problems at Aberdeen to try to do these problems. But then we also looked in to why Mrs. Goldstine––Johnny was very excited about this method. We looked into the question of setting up the problems on this background control. We ran into a chap that was down there that we knew. He was working on augmenting the use of the machine, of that computer. This was just exactly what was needed to set up the background control of this computer. We asked him about it and then mentioned it to Johnny. Johnny was a consultant both to Aberdeen and to various other places. He arranged so that the possibility of our using that unit would be available to us to do the background control. That was just what the doctor ordered. So, he arranged to make the exchange.

So, we worked on it and we were fortunate enough to have a long weekend. It was, I guess, April or May, Memorial Day. I guess Indianapolis was running that time.

Rhodes: This was what, ’46?

Metropolis: No, this was already in ’48.

Rhodes: Oh, oh, oh.

Metropolis: Right after the war, the machine was left there in Philadelphia, and then it was moved late in ’47. That’s why [Enrico] Fermi, among others, decided that he would build a machine, a small computer, to handle the problems that he was interested in.

At any rate, this was done, and we had this long weekend. There were some troubles with it. We found out many things that we understood better. Finally, we got it running. The machine then was utilizing this background control. This had the advantage that––I mean, I don't know whether much of this makes sense to you. It would take several hours to really discuss it in detail. So that, at any rate, we––

Rhodes: Well, it makes sense in that it sounds as if you were moving toward a much more flexible programming system.

Metropolis: That’s right. That’s exactly right. It had the advantage that people could––if the function tables—that meant there was a limited finite capacity to them. But if, let’s say, a person doing a problem would use half of that function table’s capacity, then we could utilize the rest of it for another problem. And therefore, the back and forth was available to them.

This played an important role because we did not have to wait until the whole background control that was utilized in running that machine. So, we could actually shift from one problem to another. This made for much greater efficiency because we didn't have to––I mean, just naturally we would have access to the machines just by a flip of a switch, from one problem to another.

We ran those problems in ’48. Then, one of the main problems was to convert electronic engineers into computer engineers.

Rhodes: It’s still a problem.

Metropolis: That’s still a problem, even today. As you were saying, it helps to––with this problem. And this is a basis of one of the very first mistakes that Enrico Fermi [made]. You know, he was a very conservative type. He never made mistakes at all. He just never made mistakes. He said, “How many tubes do you have there?”

I said, “Well, we have about 18,000 tubes.”

He said, “Well, it’ll never run.”

And that was his first mistake. Why did he make such a statement? He made such a statement because he had compared the electronics of his cosmic ray counting or whatever particles that he wanted to count. He said that the problems he was involved with, and extrapolated from his knowledge of the circuitry based on electronic engineering, whereas the computers had to be much more reliably built than the electrical engineers would effectuate.

Rhodes: So he was thinking in terms of the breakdown.

Metropolis: That’s right. And, that’s why he was saying, based on his 20 tubes, 20 or 25 tubes, there was a factor of four times 72. I mean, it was a factor of 70, close to 100 in supporting that. So he said it would never run. I was a little bit surprised at that, because I didn't think that all of that money would be going into the construction of that machine if it were going to not run at all. So, it turned out that he was wrong, unfortunately.

He had a very keen sense that he would never make any kind of mistake at all. And whenever he said something, it was always fact. I mean, I mention this because of its uniqueness.

Rhodes: I had a sense in reading about him and looking at people’s stories about him that he really tried to quantify everything.

Metropolis: That’s right. He was really tremendous, tremendous. He had a great sense of choosing the right problems, the right type of problems for each type of computer. For instance, when we had the IBMs, he was very much interested in the running of the problems on the IBMs. So, when he had a very simple problem, which turned out to be quite nice and publishable. Then later on when the MANIAC [Mathematical Analyzer, Numerical Integrator, and Computer] was running here, he had just the right size of problem for it. So it was that. But then, we continued here with running a problem––I mean, machine development as well as continuing with IBMs right on through and commercial machines, so that we had the 1620s or something like that.

Rhodes: Those were card-sorting 1620s?

Metropolis: Yeah. They were electromechanical.

Rhodes: But then for your computer, you would go to Aberdeen?

Metropolis: Well, we would go to Aberdeen, and we utilized a lot of computing everywhere we could find it. The CIAC [Computer Incident Advisory Capability] did a certain amount of computing. That was a Washington development, a CIAC there. And then we also used the ENIAC, now located in Aberdeen. Then we also used the electromechanical machine of Harvard, the Mark 1, was then available. Then we had our own development here. We also had the commercial applications.

I think that one of the incidents that is worthwhile to mention, and that is that we had very good relations with the IBM people.  

[Background comments.]

Metropolis: Mostly they were Monte Carlo problems that we were very much involved in. Then we were also very much interested in hydrogen bomb problems.

Rhodes: This is during the late ‘40s?

Metropolis: Yeah well, and then it was in the late ‘40s and early ‘50s.

Rhodes: The problem that Stan Ulam discusses in his memoirs of trying to hand calculate the hydrodynamics of what must have been the Runaway Super, at the same time it was running on a computer—

Metropolis: Yeah, but that was not a very good piece of work.

Rhodes: How so?

Metropolis: Well, it was being done by throwing dice, random numbers.

Rhodes: This is the calculation that Fermi and Ulam were working on together?

Metropolis: Well, no. That was the oscillator problem.

Rhodes: I’ve not seen it called that.

Metropolis: Oh, okay. There were three people working it. But that was already in the ‘50s.

Rhodes: It was just after 1950.

Metropolis: Yeah. There were several problems––well, two problems at least. One of them was run on the MANIAC. That was what Fermi––I’ve since learned that Fermi called it his great contribution to physics. But I don't know. I’ve not heard that described at the time he was doing that. It was the very last problem that he did because he died in ’54. Stan and I saw him at the Billings Hospital at the University of Chicago.

Rhodes: I tried to tell you something you did, and I shouldn’t do that. What I’d like to try to understand is what went on to try to figure out whether or not the Runaway Super would work. I think that’s what was involved at that period in time, just before the Teller/Ulam discovery and breakthrough to the two-stage idea.

Metropolis: Stan immediately went, after they had done this problem with throwing dice and utilizing the slide rule and things like that, along with [Robert R.] Everett, he immediately went to Princeton.  I think that he relayed the results of that computation to Oppenheimer. That was the, I think, one of the reasons why Oppie decided that he did not want the Super to be effectuated. That was why he took the position that he did.

Stan––one of the things that bothered me a great deal was that I had gone to, at a time shortly after the war, to Los Angeles when I learned he was having this [inaudible] malaise. I went to see him in the hospital and he was still in a coma—post-operational things. I wandered in with Francoise, his wife, and saw him. She was observing the situation and noticed that Stan awakened and recognized. So, his faculties were not all gone. This was a critical moment for her. She was very delighted. But, the point of the story is that I had suggested to him, in the condition that he was in, that he ought to come to Los Alamos to recuperate and give up his position at USC. So he did. I never heard that mentioned. He was always very polite and very kind to me. I appreciated that. While he never said that he was brought to Los Alamos by me, he always felt a deep regard for that. But, he never said it and nor did Francoise.

Let me get back on track, because that was a deviation. Let me recommend that you help me in trying to figure out what it is that you want.

Rhodes: I have the impression from Stanislaw [Ulam]’s memoirs that the work that was done to try to understand whether the classical Super, the Runaway Super would work, turned crucially on the period of calculation when there were hand calculations being done against the slower process of programming and running computer calculations. I had the impression that had to do with calculating the hydrodynamics of the––well, calculating something relating to the Super that was critically dependent on the amount of tritium. Because, I guess he kept at––in the calculations, working with more and more tritium.

Then, as a result of the discouraging results of these calculations, Teller was really thrown into a kind of crisis about, “This isn’t going to work. What next?” It became a very creative process in terms of coming up with the right idea. That’s the sort of version that’s—I’d love to understand it in more detail or to know in more detail what went on. You were a part of all this, were you not?

Metropolis: Yeah, right.

Rhodes: Yes.

Metropolis:  But I think that his calculations were too crude, although they were a little more sophisticated than the earlier work that had been done on the ENIAC. That was in ’45. But on the other hand, the ’45 results, there was a meeting held here in ’46, here in Los Alamos, and it was there.

Rhodes: The Super Conference of April?

Metropolis: Yes. Right.

Rhodes: But there were indeed Super calculations in ’45?

Metropolis: That’s what we had done on the ENIAC.

Rhodes: Oh! I didn't understand that. I see. They had to do with the Super?

Metropolis: Yes, but it was a very simplified program, although it was very, very complicated. It was nonetheless as simple as it could be made.

Rhodes: What conclusion did you reach in ’45?

Metropolis: Well, in ’46, then when we had the conference, we concluded with Frankel and I that the results were encouraging. Stan looked at them a little more carefully. The two of them were working together. They were able to include some effects that we had neglected in order to simplify. But still in all, his calculations were shown to be somewhat of an approximation.

Rhodes: They were throwing dice to get random numbers because they didn't know what numbers to plug in.

Metropolis: That’s right. Well, they were trying to throw dice in order to simulate the Monte Carlo approach.

Rhodes: You mean literally throwing dice?

Metropolis: Yes, literally throwing dice. I mean, in fact that would be Everett throwing his dice and using his slide rule to calculate the effects based on the numbers that would show up on the faces of the cube.

But then Stan went immediately and I said, “Why do you do that, Stan? Why do you do that? Because, it’s not right, I mean especially if you're a mathematician, to actually say ‘Never.’” He thought he had the result that then claimed that Oppie was basing his––now I don't know this. I don't know this.

Rhodes: Well, I speculated the same thing. I think he just didn't want to build it.

Metropolis: That’s right.

Rhodes: He thought it couldn’t be built and probably was glad. But, that’s another––

Metropolis: Exactly right. I said to Stan, “Why do you do that? It’s unlike a mathematician to use the word ‘Never.’” He always used the word, “It will never work.” But, in 1970 it was shown that with the onset of more complicated machines and smaller packages and things like that, they were able to include more and more effects.

 Finally, the Super, the Runaway Super, was actually shown to work. But, nonetheless, it had the desired effect, mainly to stimulate other directions of trying to effectuate the method of two dimensional––

Rhodes: Two dimensional? Two-stage, you mean?

Metropolis: Yes, two stages and also two dimensional, because then we were able to utilize this cylindrical properties of the gadgets. But now that’s a later development of––a present development of parallel processing, enabling us to really think about problems in three dimensions, because of a number of computations that we can make each second—runs up, builds up, as wells as enabling one to use parallel processing. That was one of the features of it, and that’s what helps motivate the fact that we can do problems in three dimensions. Because then we are living in the world of nature, if you will.

Rhodes: Why did Stan Ulam do that?

Metropolis: I don't know why he did that. I don't know why.

Rhodes: Was it a personality conflict with Teller?

Metropolis: Well, that could have been. That could have been.

Rhodes: What was your participation in these events?

Metropolis: I was in charge of the research on computers and designed a computer. We thought we would try to develop computer engineering as opposed to electronic engineering, thereby training people who were electronic engineers to learn about computer engineering.

I spent a year at Princeton, at the Institute [for Advanced Study], thanks to Johnny, to help with that project, and at the same time to think about organizing the plans that we had in line for Los Alamos. The time scale was 1948, I would spend there. 1948 I would spend the year at Princeton. And then ’46 to ’48 I had spent in Chicago teaching in the physics department.

Rhodes: Oh, I see. With the whole operation that Fermi––

Metropolis: Yes, that’s right. And then, there were delays in Chicago, but I don't know whether you're interested in any of that. There were delays in Chicago in getting the buildings built. In fact, at one point Fermi told Sam Allison, who was then the official director of the––I mean, it was called the Institute for Nuclear Studies. Then later when Fermi died, they decided they’d call it the Fermi Institute, and that’s what it has been ever since.

I had returned to Las Alamos to try to do some of the organizing of the projects here, and spent the year in Princeton learning about the machine there that Johnny was building. That was the project that was going on at Princeton.

Rhodes: Did that machine eventually get done and have a name?

Metropolis: It was called the IAS computer because Johnny––and therein lies a tale.

Rhodes: Yes, so I’ve heard.

Metropolis: Right. I had proposed it be called the MANIAC, and Johnny thought that was too frivolous. But I had in mind calling it the MANIAC because it would put an end to the naming of machines. It would not be allowed to pass, because it had just the opposite effect.

Rhodes: Oh, I see.

Metropolis: Because it was called the MANIAC––

Metropolis: So George Gamow, the astrophysicist, said that “Metropolis and Neumann invent awful contraptions.” That was his interpretation. We had a bona fide. But actually, the word MANIAC came ahead of what it stood for.

Rhodes: Yes, as is often true with acronyms.

Metropolis: Right, exactly. I don't know whether you would like to hear about the details of the fact that the Selectron, which was going to be the memory system that von Neumann was going to use that was being developed at the research laboratories of RCA at Princeton Junction. They were going to build the Selectron. The Selectron was delayed and delayed and delayed. And finally they were saved by the Williams tube that was being developed by Professor Williams at Manchester, England. It’s an electrostatic device. They were going to utilize all of the equivalent on the outside of cathode ray tubes instead of putting the material on the inside. That’s what all of the problems were connected with, was putting it on the inside.

Then, we ran into a bit of luck here because we had somebody coming here from Toronto who––and they had heard about the Williams tube in advance of us. Richardson was the name of a person who came from Toronto and had some computer electronics under his belt. So, when he came he said––he proposed that as long as there were a delay of some kind, because we had followed in the footsteps of by way of training the people here to become computer electronics engineers instead of electronic engineers.

He proposed that we spend a little time looking after these things. He recommended that we take the two-inch tubes instead of the big, five-inch tubes because that’s what they were using in Manchester. Von Neumann decided that they would go along with the stuff that Manchester was trying to do, whereas we were concentrating on the two-inch tubes. We continued, and there was a schism there. The director, [Norris] Bradbury, who is here, said that “Let us go our separate ways.”

So, we had developed the two-inch tube and then we went on to develop the rest of the controls. We even had a line printer from the Analex people that was being used. What’s the name of that outfit that––cryptographic analysis in Washington?

Rhodes: The National Security Agency?

Metropolis: Yes, the National Security Agency, bravo. That’s good.

Rhodes: You were building a computer here, at that point?

Metropolis: That’s right.

Rhodes: A MANIAC?

Metropolis: The MANIAC. That’s the story of Fermi, being—just the right problem.

Rhodes: Yes.

Metropolis: For the right machine. He wanted to do the problems that he was working on at Chicago in experimental physics. He wanted to get the so-called, as they’re called, phase-shift analysis. That’s what it was called. He did all of the menial things, typing of instructions and things like that. He wanted to do everything by himself. He wanted to know––he wanted to know. Just having the experience of putting on these—doing these menial tasks. We had people there especially trained for that kind of thing.

Rhodes: I don't know if you know this story. When they were working on the CP-1 [Chicago Pile-1] and he was so busy, he was really working out of his office rather than over their building the reactor. He told Emilio Segre at one point, “Emilio, I’m doing physics by telephone.” He didn't like it at all.

Metropolis: Yes, that’s correct. I knew Emilio very well. He had been here in this house when he would come here. But he died two years ago, I think.

Rhodes: Yeah, we had been out to see him a couple of times. It was so abrupt, so sad.

Metropolis: Yeah, right.

Rhodes: So, the MANIAC was used, was it not, for some of the calculations for [Ivy] Mike [nuclear test]?

Metropolis: Oh yes.

Rhodes: It was ready by then?

Metropolis: It was ready in March of ’52.

Rhodes: Yes, just in time, really.

Metropolis: Just in time, exactly. We had problems ready to go on and things like that. Marshall Rosenbluth was handling that.

Rhodes: Bob Serber told me that the ‘46 conference was less optimistic about the Super design than the report.

Metropolis: Well, some people were of that opinion.

Rhodes: Yeah, that sounds as if its––

Metropolis: That’s right. I mean, the report itself was fairly optimistic. But then there were others that were there. Bob Serber, I knew very, very well. I knew Charlotte very well. Charlotte was his first wife, who died. Then he remarried and is happily married.

Rhodes:  One of the things that I’m trying to understand, and Dr. Teller fortunately talked a great deal about it the other morning and I’ve seen Ulam’s version in his book, and that is: what happened around the discovery of the two-stage idea? Teller particularly mentioned the other morning something I hadn’t really heard from him before, which was, the ideas of the compression of the thermonuclear fuel had been floated from time to time.

He said, “You know, I would have thought of it––I would have paid more attention to it.” He gave it a physics explanation for why it wasn’t clear that this was important. He said that the three-body problem didn't clear the four. But he also emphasized something that puzzles me, which is: having focused so much on the Runaway Super idea and believing that it could be made to work, people really didn't think about other possibilities. Is that clear in terms of your memories of those times?

Metropolis: You have to know that Edward was not very sympathetic to the controlled thermonuclear reaction. The controlled thermonuclear reaction––in other words, the analog of the pile––there are four 2x2 matrix and there are four elements. One is the controlled fission bomb and the other is the fission bomb itself, which is uncontrolled. And then there is the controlled thermonuclear reaction, and then there is also the uncontrolled thermonuclear.

Three of those projects have been very successful and very easy really, in retrospect. But the fourth one remains very, very difficult. That’s a good way of looking at the situation and looking at nature, if you will. That is, that the runaway thermonuclear reaction is very effective, but the controlled thermonuclear—which is what the tokamaks and things like that have been able to do—have been. And Edward, who is an optimist of the first water, has said admittedly that he wants to stay away from the controlled thermonuclear. That is a fact. I mean, he thought that the controlled thermonuclear would be after his time and therefore not of interest to him.

Rhodes: I want to cut down on the tape because the chimes overrode your voice. You said, “There was the sun out there making a mockery of it all.”

Metropolis: Yes.

Rhodes: You mean, there was the sun––

Metropolis: Thermonuclear.

Rhodes: Burning away.

Metropolis:  That’s right. I mean, every day we’d see the sun coming up and every day we’d see thermonuclear reactions in terms of the sun.

Rhodes: Proving that it is possible.

Metropolis: That it is possible.

Rhodes: Like a big fireball.

Metropolis: That’s right. And it may be just that the size of it––nature tries to utilize the more difficult ways in very simple structures, and thereby expanding it to the size of the sun because it is very big, much bigger than the earth is. So, there it is making a mockery of this. And maybe the controlled thermonuclear––and that’s what I was saying, Edward, who is normally very, very optimistic, is very pessimistic about the controlled thermonuclear reaction.

Rhodes: So optimism in the context of the Runaway Super has to do with thinking it will work and therefore neglecting other ways of igniting it? Because that, after all, was the problem, right, getting it started?

Metropolis: That’s right. Johnny’s great contribution was his role in seeing that the implosion methods would work very well. And, there was an added [inaudible] in increasing the density of the material as a consequence. He was the first to realize that there was something to be said for the––but then we became a little conservative, because instead of just having a lining which was imploded, they had the solid material and the shockwave was going into it.

Rhodes:  I understood that we didn't go with the shells that were originally being discussed, because the calculation was too difficult to make. Is that correct?

Metropolis:  Not too difficult, but it showed instabilities.

Rhodes:  Oh. Yes, but we’ve gone to shell since then?

Metropolis:  No, not really. I mean most of the things have been––

Rhodes: Solid course. I see, but a different kind of squeeze. I’m understanding that you can get densities—greater densities with greater squeeze.

Metropolis: That’s right. And thereby improving the efficiency of the bomb.

Rhodes: Still, it puzzles me that Teller, who was intimately involved in the development of implosion and in fact says that he’s the one who made it clear to von Neumann that you've got compression—increased density I should say, of the solid material.

Metropolis: I don't know. I don't know that part of the story.

Rhodes: It surprises me that he didn't connect that with the necessity to increase the density of the thermonuclear fuel.

Metropolis: As a consequence of the implosion aspect.

Rhodes: Or having realized that you got a better reaction out of solid material, out of metal, if you made it denser. It might make sense to think about the thermonuclear fuel, increasing its density too.

Metropolis:  No, no. I mean, if you just have a shell, which comes in, then there is a free run and enables the material to achieve a higher density than if it were solid. Okay? Because what it depends on is—the rate of the increase is much more effective if it has a free run in, rather than with a solid. Okay? I mean, the solid––at first it was just the fact that it could be imploded out as opposed to explosion. Well, you get a much bigger increase if you have a shell and have it collapse to the center. Then, around the center there is a much higher degree of density than if you had just—moving it into a solid. Of course, you have the same effect, but not quantitatively. Quantitatively, you have a much bigger effect if you have a shell than if you have a solid material.

Rhodes: Oh. But the shell, as you said, wasn’t something that would work.

Metropolis: That’s right, because of the instabilities.

Rhodes: I’m just trying to see if there’s any parallel in terms of the basic ideas between increasing the density of the solid ball of the pit, which is what happens with implosion.

Metropolis: Yeah.

Rhodes: Metal itself is increased in density, bringing the atoms closer together.

Metropolis: Right, right.

Rhodes: That on the one hand, and the problem of thermonuclear ignition, which was ultimately solved by increasing the density, compressing the thermonuclear fuel. Are those parallel ideas or am I missing––am I wrong here?

Metropolis: Let’s look at the gun method of ignition that was used in Hiroshima. You brought in two halves of the gun and then you had a cylindrical shape. So you brought in two pieces, neither of which was supercritical. But as a consequence of being juxtaposed, then you achieve supercriticality and the bomb goes off. Now, it was the spontaneous fission that drove going to a spherical shell or a solid material. It was the increase in density that would be achieved there that would help, and you avoided pre-detonation by having at first the shell study. If you imagine a spherical shell coming in and then squeezing to the center and increasing the density over and above just the mere fact that you now are filling––

Rhodes: Changing the geometry.

Metropolis: That’s right.

Rhodes: You also increase the density.

Metropolis: You also increase the density.

Rhodes: And with the solid pit, it’s subcritical in amount at its density.

Metropolis: Yes.

Rhodes: But when you increase the density, then it becomes supercritical.

Metropolis: That is correct. That is correct. And what do you achieve by the solid particle? You achieve the fact that you have less instability, or more stability. Okay?

Rhodes: Uh huh.

Metropolis: Is that all right?

Rhodes: Yes. I think I understand that part. Is there no analogy in terms of––oh, I see what you're saying.

Metropolis: There is the fact the shockwave helps to bring in a much more dense material and thereby––and with the less instability, you'll make it a solid implosion.

Rhodes: Uh huh. So the advantage of the method, you're saying, was that it reduced the instabilities.

Metropolis: That’s right.

Rhodes: Still, the process of making this amount of material supercritical, when it was a solid ball to start with, had to do with increasing its density.

Metropolis: That is correct.

Rhodes: Because, at higher densities, a smaller amount of plutonium will chain-react.

Metropolis: That’s right.

Rhodes: And it’s that increase––it’s that compression of the solid material that I’m trying to find if there’s an analogy with the compression of the thermonuclear material, which made it possible for it to burn at a lower temperature.

Metropolis: Yes. That is correct.

Rhodes: Those two ideas are sufficiently analogous. I’m surprised that Dr. Teller didn't put them together in his mind and see that compression of the fuel was an important aspect of it. Because isn’t it compression that the two-stage was successful at achieving?

Metropolis: The two-stage phenomenon is radiation coming in and causing it to squeeze, whereas in the original Super it was lenses that were focusing the part. And it was the fact that there was a solid ball instead of having the spherical shells coming in.

Rhodes: The advantage of radiation compression or radiation implosion was again getting the process started before everything blew apart.

Metropolis: That’s right. That is correct. That is correct.

Rhodes: I have a feeling this discussion is impeded by the security rules that make it impossible for you to explain more.

Metropolis: That’s right. I’m trying to recognize I’m talking to both uncleared people.

Rhodes: Well, this is probably a question I should have asked Dr. Teller anyway, and I will ask him.

Metropolis: Okay.

Rhodes: Because he did clarify one of the important physical reasons why he didn't see the two-stage idea for a long time. That had to do with the fact that the various parameters he thought all increased together.

Metropolis: Yes.

Rhodes: And so, increasing the compression wouldn’t change anything.

Metropolis: Yes. Yes.

Rhodes: So, all right.

Metropolis: Okay. But the whole history of the Teller/Ulam interaction is a very complicated one. You have to realize that both of them were individuals.

Rhodes: That’s very clear. You must have heard about that as it was happening.

Metropolis: Oh yes.

Rhodes: Could you talk about that a bit?

Metropolis: The trouble is that I don't know what the declassification things are about, and Edward is quite right in being opposed to the secrecy aspects. I mean, he allows certain things to be kept secret, but only for a limited time. And then, you only hurt your own compatriots.

Rhodes: Oh yes, not to mention historians.

Metropolis: Yes.

Rhodes: Well, what has been made public is that Ulam’s idea had to do with a two-stage design that worked in terms of mechanical shock, of the shockwave, coming from the primary to be used to compress a secondary capsule. And Teller’s immediate reaction to that was to see the radiation would be much more effective.

That much at least has been made public. And in fact, radiation compression was indeed the breakthrough idea. In addition, then there’s––but I don't even know how much of this has been officially declassified and how much my friends have ferreted it out of documents—the notion that then there was also a spark plug, which was a secondary fission device inserted in the middle that would start a second fission explosion going on. That much at least is in the clear.

Metropolis: Yes.

Rhodes: Does that help in terms of talking about what these two people were doing?

Metropolis: Well, it does. It does help. But, the thing is that Stan thought that it could be done with neutrons. I don't know whether that part has been ferreted out. But Stan thought it could be done with neutrons. While it was so, it was much easier to do it with radiation.

Rhodes: That helps a lot. Yeah. Is it possible to say at all, to say who should receive credit for this discovery? I mean, let me give you an analogy. I’ve been working for the last year on the Russian program. I’ve learned a great deal about the early years of their work. No one over there hesitates to say Andrei Sakharov thought of the idea of the two-stage design using radiation to transport it and so forth. They're very clear about that.

Metropolis: Yes, but on the other hand, what isn’t so clear is to what extent were they helped by a person known as Klaus Fuchs.

Rhodes: I think it’s fairly clear that they weren’t helped very much. He was able to report to them on the Super Conference of April ’46. But after that he was in England. Whatever he knew he learned from that program rather than from our program.

Metropolis: You know, Fuchs had his office next to mine. Whenever I walked in, and I would walk in early, like 8:00, he was always there. When I left at night, at 5:00, 5:30, he was still in his office working away. In other words, he worked long, long hours. He was so careful about pointing out all kinds of things. And yet, he was not a spy in the usual sense of the word. He was an idealist, I think. That’s the way I would describe him, that he was an idealist and that he just wanted more than one nation to know the secrets of the atomic bomb. Therefore, he felt that giving these secrets, for which he never received a penny.

Rhodes: That’s right. He didn't want money.

Metropolis: He didn't want money.

Rhodes: I have seen what he gave. One important document that he gave to the KGB and it is a very detailed description of the Fat Man bomb.

Metropolis: Yes.

Rhodes: Including something I’d never seen over here because it’s still classified, a very detailed description of the Urchin.

Metropolis: Yes.

Rhodes: Which I thought was an extraordinarily ingenious device.

Metropolis: Yes.

Rhodes: Just as ingenious as I’d always heard.

Metropolis: Right.

Rhodes: With its little shaped charge and all those things, right. He passed hundreds of pages of documents. When you saw him in his office early he may have been writing out all these things.

Metropolis: That he was eventually to give away.

Rhodes: But he was effectively, I think, cut off from knowledge of the further developments of the thermonuclear here. And you know there’s one interesting test that how much they knew from Fuchs, and that is Joe-4. Joe-4, their first thermonuclear design, was an alarm clock. It was a series of layer cake design, they called it, using lithium deuteride interspersed with uranium.

Metropolis: Right.

Rhodes: And they claimed it was about 15% to 20% fusion and that it was about 400 kilotons and that it was about the size of Fat Man.

Metropolis: Yes.

Rhodes: And that would be really quite remarkable, but it ain’t the Super.

Metropolis: That’s right.

Rhodes: They went off in a different direction, which suggests that they really didn't get the later information from here.

Metropolis: Right, right.

Rhodes: Is that consistent with what you understand?

Metropolis: That’s right.

Rhodes:  Then, in addition to that, they also claimed that they were not able to pick up the fallout from the Mike shot.

I just talked with George Cowan yesterday and he said, “Well, that’s probably true. They probably didn't have the radiochemistry yet.”

But Sakharov himself said, “No, we did not get any help from the Mike shot debris.”

Metropolis: Yes.

Rhodes: They argue strongly that it was an independent discovery on their part to come up with radiation compression.

Metropolis: But when Sakharov was in this country, and he visited I think only once or perhaps at most twice. But he wanted to meet Edward.

Rhodes: I spent some time with Yelena Bonner [wife of Andrei Sakharov] when I was in Moscow last year. She’s a real champion of Teller’s. She feels that Teller received treatment here something like what her husband received there.

Metropolis: Yes, yes. There is the many-facetted story of the Hungarians and the fact that [Leo] Szilard essentially monopolized the conversation as an object of conversation at last night’s talk.

Rhodes: Oh really?

Metropolis: Yeah, at last night’s dinner.

Rhodes: Why so? Why did he monopolize? Was he the subject of conversation?

Metropolis:  Well, I guess––I don't know how these conversations go. But they somehow find a way of–– Szilard was, I guess, known to everyone there, especially Edward and Mici. So it had a way of sustaining itself.

Rhodes:  I’ll tell you a story that I heard from the Russians that I haven’t told around because I want to save it for my book and not let the other historians know, but I think you'd find it interesting.

When they were locked away at Arzamas working on their weapons, it was even worse than Los Alamos during the war. They couldn’t even leave for vacations. They were just there. They were very much cut off from the world. One of the ways that they kept in touch with what was going on over here, they had a subscription to the Bulletin of the Atomic Science. And one day, one of them told me, Andrei Sakharov, young Andrei Sakharov, walked in with a copy of The Voice of the Dolphins, Szilard’s––

Metropolis: Old book.

Rhodes: Yes. Szilard’s story about being tried as a war criminal as a result of having lost the war really hit them very hard. They hadn’t thought about these moral issues very much.

Metropolis: I see.

Rhodes: And it was, this physicist thought, the beginning perhaps even of Sakharov’s consideration of all these issues. So, Szilard once again––

Metropolis:  Oh yes.

Rhodes:  Like a note in a bottle, drifted over into that world and brought them some of his thought. Isn’t that an extraordinary story?

Metropolis:  That’s right. I went twice to Chicago, once in ’46 and then another time in late ’57. I arrived in Chicago when Sputnik was––October 4, 1957 is the date of Sputnik. I arrived in Chicago on that date and I went there and stayed there for about eight years. But what I wanted to say, that when I arrived there––do you know Studs Turkel, by any chance?

Rhodes:  Yes.

Metropolis:  He is a good friend of my sister’s, my older sister. And, I happen to know him reasonably well. So, the first thing he asked me was what about, “Do I know about The Voice of the Dolphin?”

I said, “Yes, I have a copy of it with an inscription on it.”

So he says, “Well, may I borrow it?” He said, “I will keep it only about a month until I read it and then I’ll return it.”

And so a year passed. I wanted it back because of the inscription and because I think that––no, I think Szilard was still alive. I still liked to have it. So I asked him, what about this book that I had loaned you a year ago, more than a year ago? And he returned it.

Rhodes: One last question that comes to mind: what was the response here when the Soviets tested their first bomb, Joe-1, in ’49?

Metropolis: Yes. That is a secret of the fact that Stan was giving some talks at Harvard, I think. The group met one evening and said, “Well what do we do, because Stan is coming back tomorrow?” So we thought of a story. This was a poker club and we all played poker. So we sat around, and it was a very dull game. You know how these games fluctuate.

But I returned home. This was in ’49. So, I was a bachelor then. One of the players had a stomach cramp and was taken to the hospital after he had left and arrived at home. His good wife took him over to the hospital. So we thought of telling Stan that we had a big poker game and there were fluctuations, very serious fluctuations. We normally would break up at 12 o’clock. We’d have our last round at 12 o’clock, at midnight. But then we decided that there was always these fluctuations, that the game went on and on and on. Bradbury even interceded, the director had interceded to tell us that we could not play any poker. So we had all this drummed up.

So, when Stan came back––and we lived about, in those days––you know, this apartment is where the old lab used to be?

Rhodes: Oh really?

Metropolis: Yes. That’s Los Alamos Canyon out there. So, the inn is where the lab used to be for the theoretical building.

Stan lived nearby and he came down and he said, “What’s new?”

So I said, “Well, you know, the Russians have fired the bomb.”

And we knew it ahead of the public because of Truman, he passed the word on to us.

So, he says, “I don't believe it.” He said, “Anything else happen?” He said, “For instance, was there a poker game?”

So, we had this cock and bull story ready to tell, and everybody had agreed to it. In fact, the one serious mistake we had made was to tell Francoise about it.

And then Stan asked about the poker game. I told him, “Look, we’re not supposed to talk about this thing.” And Bradbury––there were lots of serious fluctuations and the game went on and on and on and on past midnight and past the one usual round of talks. So he went back to the house. He left. And then Francoise saw him fretting, and said it was a cock and bull story because she was just taking pity on this man.

Rhodes: He was missing the idea of playing poker.

Metropolis: Yes, that’s right. He was always a great poker player, a great poker player. There are lots of stories, lots of stories. Wait a minute. When I told him about the Russians firing the first atomic bomb, he said, “I don't believe it.” But then he believed the whole of the story of the poker game.

Rhodes:  [Laughing]

Metropolis: He believed the whole story of the poker game. And he believed all of that. The only thing that was true was that the fact that the Russians had fired the bomb in ’49.

Rhodes: That’s funny. It depends on your expectations.

Metropolis: That’s right. And he loved to play poker.

You know, you would think that Johnny von Neumann would love to play poker, but he didn't. Johnny was very bored, very bored. The height of my ambition was to win $20 in a poker game, and then buy the book that he and Morgenstern wrote on the theory of games.

Rhodes: With the winnings.

Metropolis: With the winnings, that’s right. And paste that into the front page.

Rhodes: He did play poker though, did he not?

Metropolis: He did, yeah.

Rhodes: Without any interest.

Metropolis: Not with any interest. That’s exactly right.

Rhodes: Did he, as I’ve heard, have an amazing collection of dirty limericks that he could recite on queue?

Metropolis: But that was true of—Edward, you know, liked limericks.

Rhodes: Ah, yes, right.

Metropolis:  Edward liked limericks. I think that Johnny liked a lot of scatological stories.