The results of the CPC’s calculations could be punched onto new cards or printed onto fifteen-inch-wide continuous-feed form paper. The printer, already mechanically sophisticated because it was part of the previously developed accounting machine, could provide a row of up to 132 digits in a single stroke. Between the input and the output came the arithmetic. That was controlled by plug boards a little less than one square foot in size that could be individually wired to meet the needs of a particular kind of calculation. For instance, suitable wiring could allow a single command on a punched card to order a square root, leading to the kind of hiccup mentioned above.
The CPC was slow, ponderous… and reliable. The ENIAC, a nightmare to program and already obsolete, had served its evolutionary purpose. Princeton’s MANIAC, revving at the starting gate and ready to launch a revolution, was to the CPC what a prototype Aston Martin is to a Model T. It was fast, sleek… and unreliable. The new breed of stored-program machines, offspring of the MANIAC, would soon take over, but for a year or two, the CPC did its job, churning out the calculations that kept the nuclear weapons programs on track. (And not just the weapons programs. Of the more than 500 CPCs delivered by IBM to government and industrial concerns, many, like antiquated Model T’s, kept on serving the needs of their owners for many years.{8})
There were, if I recall correctly, only three or four CPCs in Los Alamos during my year there. By the spring of 1951 this was not enough to meet demand, even with late-night use of them.
So, shortly before my return to Princeton, I was dispatched to Sandia Labs in Albuquerque to carry out calculations there, where CPC usage was not yet saturated. At least not saturated around the clock—time in the middle of the night was made available to me. For a place to sleep during the day I was assigned a room in the BOQ (bachelor officer quarters), which gave me a certain wry pleasure since as an enlisted man in the Navy half a dozen years earlier I had perceived officers as light-years distant.
These Sandia calculations, like the ones just before at Los Alamos, were on the propagation of a thermonuclear flame in a cylinder of deuterium that had undergone radiation implosion (using a lot of hypothetical numbers, since the degree of compression and even the dimensions of the cylinder were unknown). The results began to look encouraging.
No sooner was I installed in Princeton than the Los Alamos authorities arranged for me to continue these calculations on a CPC located in an IBM building on the East Side of Manhattan (not the principal IBM building in midtown). I was—of course—assigned an extended graveyard shift, 8:00 p.m. to 8:00 a.m. if I remember correctly. Typically, I took the train from Princeton Junction to New York in the evening and returned by train in the morning, checking in with my colleagues before seeking some rest. To this day I have fond memories of the Hamburger Heaven across the street from the IBM building. It stayed open all night. And good memories of John Sheldon, the aspiring executive who ran this small branch of IBM. He constantly fussed because IBM culture dictated that he wear a hat on his way to and from work. On his staff was a blind young man who expertly wired plug boards.
Chapter 14
The Garwin Design
An important meeting of an advisory panel convened by the Atomic Energy Commission was scheduled for Saturday and Sunday, June 16 and 17, 1951, and was to be held at the Institute for Advanced Study in Princeton, fortuitously soon after Matterhorn got going.{1} It was a truly “blue-ribbon” panel, attended by all five Atomic Energy Commissioners;[74] Robert Oppenheimer and four other members of the General Advisory Committee; a Los Alamos contingent that included Norris Bradbury, Carson Mark, Lothar Nordheim, and Edward Teller; John Wheeler from Matterhorn; and what I call the “big-three” consultants: Hans Bethe, Enrico Fermi, and John von Neumann—in total about twenty people.{2} Progress toward a thermonuclear weapon was to be a principal agenda item.
I spent that Saturday night in New York nursing a CPC, as I had spent every night that week. I tweaked the input parameters and assumptions this way and that way and could not escape the conclusion that the compressed deuterium would probably burn. On Sunday morning I took an earlier-than-usual train to Princeton Junction and carried the computer output with me to the Matterhorn building. After the guard let me in I translated the numerical results I was carrying into a couple of graphs inked with marker pens on oversized paper, perhaps as large as two feet by three feet. By now it was after 9:00 a.m. I rolled up the graphs, got back in my car, and drove to the Institute for Advanced Study, where the advisory panelmeeting was taking place in a first-floor conference room. I walked across the grass to the low window of the meeting room and saw John Wheeler standing, as if about to deliver some remarks. Either by gesture or by tapping on the window, I caught his attention. He walked over to the window and opened it. I handed him the graphs, saying “Looks good.” He went back to the front of the room and taped the graphs to a chalkboard. I went in search of breakfast and some rest.
My graphs reinforced whatever message Wheeler and Teller were delivering to the assembled group. This was, apparently, the meeting at which the GAC—and, in particular, its chairman, Robert Oppenheimer—switched from doubt and questioning to a conviction that the “new” Super would very likely work. Oppenheimer famously called the Teller-Ulam approach to a thermonuclear weapon “technically sweet,” a perspective that may date from this 1951 meeting. And, who knows, one that my graphs may have helped to solidify.
(I should explain the context of Oppenheimer’s “technically sweet” comment. At the 1954 hearing that resulted in the loss of his security clearance, he was being questioned about his 1949 opposition to going forward with a program to develop an H bomb. To his interrogator he said, in effect, that he felt free to oppose it on moral grounds at that time because, on technical grounds, it seemed doubtful that it could be done. It was his judgment, he said, that “when you see something that is technically sweet, you go ahead and do it and you argue about what to do about it only after you have had your technical success. That is the way it was with the atomic bomb.”{3} By inference, he was saying, had we felt confident in 1949 that the H bomb could be built, we would have been less likely to oppose it, and instead would have waited until its completion before deciding what to do about it. His actual words were “I cannot very well imagine if we had known in late 1949 what we got to know by early 1951 that the tone of our report would have been the same.” Later in his testimony, and still in response to challenges about his 1949 opposition, he said, “The program we had in 1949 was a tortured thing that you could well argue did not make a great deal of technical sense. It was therefore possible to argue also that you did not want it even if you could have it. The program in 1951 was technically so sweet that you could not argue about that.”{4})
74
Sumner Pike, an original Commissioner appointed in 1946; Gordon Dean, Chair, and Harry Smyth, both appointed in 1949; and T. Keith Glennan and Thomas Murray, both appointed in 1950.