During this crucial period of uncertainty, I visited him in Princeton. Fermi happened to be there, too, on a brief visit, and we discussed the prospects all afternoon, during dinner in Johnny's house, and all evening. The next day we talked with Oppenheimer. He knew about the results obtained by Everett and me. He seemed rather glad to learn of the difficulties, whereas von Neumann was still searching for ways to rescue the whole thing. Johnny outlined some hydrodynamical calculations. Fermi concurred. They came to estimate a certain velocity of expansion which seemed to me much too slow. With the experience of all the work I had done in the past months, I noticed that they had erred in assuming the density of liquid deuterium to be 1, whereas it is only a small fraction of 1. This error of per unit mass instead of per unit volume made the velocity appear indeed smaller and Johnny realized it and exclaimed: "Oh my! It is indeed much faster than a train." Oppenheimer winked at me. He liked having the difficulties confirmed and enjoyed catching von Neumann and Fermi in a small and trivial arithmetic error.
My calculations with Everett concerned the first phase of the explosion, the problem of the initial ignition. An important part of the story has been overlooked in the official accounts and concerns some quite fundamental work that Fermi and I did following the first calculation of the progress of the reaction, its propagation and explosion. In numerous joint discussions we outlined the possibilities of propagation, assuming that some way or other (perhaps by the expenditure of large amounts of tritium) the initial ignition could be achieved. There again we had to use guesses in place of the enormously difficult detailed calculations that would have required computers even faster than those in existence. We did this again in time-step stages with intuitive estimates and marvelous simplifications introduced by Fermi.
The numerical work was done on desk computers with the assistance of a number of programmers from the laboratory's computing group, managed by a good-humored New Yorker, Max Goldstein. Much to Max's annoyance, Fermi wanted to encourage the girls to use slide rules; the machine precision was not really warranted because of our simplifications. But Max insisted on the usual routines with desk calculators. Reading from slide rules and using logarithms as Fermi did was much less accurate, but with his marvelous sense he had the ability to judge the right amount of' accuracy which would be meaningful. The girls, who were merely making calculations without knowing the physics or general mathematics behind them, could not do that, of course, so in a way Max was quite right in insisting on the standard routines.
I particularly remember one of the programmers who was really beautiful and well endowed. She would come to my office with the results of the daily computation. Large sheets of paper were filled with numbers. She would unfold them in front of her low-cut Spanish blouse and ask, "How do they look?" and I would exclaim, "They look marvelous!" to the entertainment of Fermi and others in the office at the time.
A joint report was written by Fermi and myself. Enrico exercised very great caution in its conclusions. In fact, one conclusion stating the unpromising nature of the reaction as it was planned contained this sentence: "If the cross sections for the nuclear reactions could somehow be two or three times larger than what was measured and assumed, the reaction could behave more successfully."
I believe this work with Fermi to have been even more important than the calculations made with Everett. It turned out to be basic to the technology of thermonuclear explosions. Fermi was satisfied with both its execution and with the fact that it put a limit to the size of such explosions. As he said: "One cannot make trees grow skyward indefinitely."
In the meantime Teller continued to be very active both politically and organizationally at the moment when things looked at their worst for his original wartime "super" design, even with the modifications and improvements he and his collaborators had outlined in the intervening period.
Perhaps the change came with a proposal I contributed. I thought of a way to modify the whole approach by injecting a repetition of certain arrangements. Unfortunately, the idea or set of ideas involved is still classified and cannot be described here.
Psychologically it was perhaps precipitated by a memorandum from Darol Froman, an associate director of the laboratory, who asked various people what should be done with the whole "super" program. While expressing doubts about the validity of Teller's insistence on his own particular scheme, I wrote to Froman that one should continue at all costs the theoretical work, that a way had to be found to extract great amounts of energy from thermonuclear reactions.
Shortly after responding I thought of an iterative scheme. After I put my thoughts in order and made a semi-concrete sketch, I went to Carson Mark to discuss it. Mark, who was by then head of the theoretical division, was already in charge of the very extensive theoretical work supporting Teller's and Wheeler's special groups. The same afternoon I went to see Norris Bradbury and mentioned this scheme. He quickly grasped its possibilities and at once showed great interest in pursuing it. The next morning, I spoke to Teller. I don't think he had any real animosity toward me for the negative results of the work with Everett so damaging to his plans, but our relationship seemed definitely strained. At once Edward took up my suggestions, hesitantly at first, but enthusiastically after a few hours. He had seen not only the novel elements, but had found a parallel version, an alternative to what I had said, perhaps more convenient and generalized. From then on pessimism gave way to hope. In the following days I saw Edward several times. We discussed the problem for about half an hour each time. I wrote a first sketch of the proposal. Teller made some changes and additions, and we wrote a joint report quickly. It contained the first engineering sketches of the new possibilities of' starting thermonuclear explosions. We wrote about two parallel schemes based on these principles. The report became the fundamental basis for the design of the first successful thermonuclear reactions and the test in the Pacific called "Mike." A flurry of activity ensued. Teller lost no time in presenting these ideas, perhaps with most of the emphasis on the second half of our paper, at a General Advisory Committee meeting in Princeton which was to become quite famous because it marked the turning point in the development of the H-bomb. A more detailed follow-up report was written by Teller and de Hoffmann. New physicists were brought to Los Alamos, and work toward experimental verification started in earnest.
John Wheeler came to New Mexico to help Teller. He brought with him several of his brightest students. Among them was Ken Ford, with whom I was to do unrelated work later on. There was John Toll, now President of the State University of New York at Stony Brook, then a promising young physicist; Marshall Rosenbluth, who had been in Los Alamos during the war as a soldier with the SED; Ted Taylor, who contributed so many new ideas to atomic fission bombs; a brilliant mathematical physicist, Conrad Longmire; and other talented young people. Intense, fast work went on, and the plans for "Mike" were ready just a few months after our fateful conversation.
During their Los Alamos year, the Wheelers lived in a house next to ours and we saw them frequently. Wheeler was a very interesting type of physicist. To my mind, he had all the right desires for novelty in theoretical ideas without sticking too rigidly to preconceived notions and existing schemes. Sometimes he thought of outlandish-sounding schemes in physics or in cosmology, so much so that some of his ideas would strike me as lacking in an element of common sense or connection with possible experiments. Or perhaps they were, on the contrary, as Pauli once said about some ideas proposed by Heisenberg, "not crazy enough." Wheeler's great merit is his work on general relativity, pursued to extreme situations like black holes and beyond; he also has great didactical talents. Of his students, the best by far, I think, is Feynman. Long before, they wrote a very nice joint paper on a generalization of Mach's principle.