The Making of the Atomic Bomb

church, but something that might have been compared in size with a church was discovered in Schermerhorn [Hall].

There, Fermi goes on, they began to build “this structure that at that time looked again in order of magnitude larger than anything that we had seen before… It was a structure of graphite bricks and spread through these graphite bricks in some sort of pattern were big cans, cubic cans, containing uranium oxide.” The cans, 8 by 8 by 8 inches, 288 of them in all, were made of tinned iron sheet; each could hold about 60 pounds of uranium oxide. Each cubic “cell” of the uranium-graphite lattice — a can and its surrounding graphite — was 16 inches on a side. Spheres of uranium in an arrangement of spherical cells would have been more efficient. In these beginning experiments, with materials of doubtful purity, Fermi was pursuing order-of-magnitude estimates, a first rough mapping of new territory. “This structure was chosen because of its constructional simplicity,” the experimenters wrote afterward, “since it could be assembled without cutting our graphite bricks of 4' by 4' by 12'. Although we did not expect that the structure would approach too closely the optimum proportions, we thought it desirable to obtain some preliminary information as soon as possible.” Promising results might also win further NDRC support.

“We were faced with a lot of hard and dirty work,” Herbert Anderson recalls. “The black uranium oxide powder had to be… heated to drive off undesired moisture and then packed hot in the containers and soldered shut. To get the required density, the filling was done on a shaking table. Our little group, which by that time included Bernard Feld, George Weil, and Walter Zinn, looked at the heavy task before us with little enthusiasm. It would be exhausting work.” Then Pegram to the rescue in Fermi's telling:

We were reasonably strong, but I mean we were, after all, thinkers. So Dean Pegram again looked around and said that seems to be a job a little bit beyond your feeble strength, but there is a football squad at Columbia that contains a dozen or so of very husky boys who take jobs by the hour just to carry them through college. Why don't you hire them?

And it was a marvelous idea; it was really a pleasure for once to direct the work of these husky boys, canning uranium — just shoving it in — handling packs of 50 or 100 pounds with the same ease as another person would have handled three or four pounds.

“Fermi tried to do his share of the work,” Anderson adds; “he donned a lab coat and pitched in to do his stint with the football men, but it was clear that he was out of his class. The rest of us found a lot to keep us busy with measurements and calibrations that suddenly seemed to require exceptional care and precision.”

For this first exponential experiment and the many similar experiments to come, Fermi defined a single fundamental magnitude for assessing the chain reaction, “the reproduction factor k.” k was the average number of secondary neutrons produced by one original neutron in a lattice of infinite size — in other words, if the original neutron had all the room in the world in which to drift on its way to encountering a uranium nucleus. One neutron in the zero generation would produce k neutrons in the first generation, k² neutrons in the second generation, k³ in the third generation and so on. If k was greater than 1.0, the series would diverge, the chain reaction would go, “in which case the production of neutrons is infinite.” If A: was less than 1.0, the series would eventually converge to zero: the chain reaction would die out. k would depend on the quantity and quality of materials used in the pile and the efficiency of their arrangement.

The cubical lattice that the Columbia football squad stacked in Scher-merhorn Hall in September 1941 extrapolated to a disappointing first k of 0.87. “Now that is by 0.13 less than one,” Fermi comments — 13 percent less than the minimum necessary to make a chain reaction go — “and it was bad. However, at the moment we had a firm point to start from, and we had essentially to see whether we could squeeze the extra 0.13 or preferably a little bit more.” The cans were made of iron, and iron absorbs neutrons. “So, out go the cans.” Cubes of uranium were less efficient than spheres; next time the Columbia group would press the oxide into small rounded lumps. The materials were impure. “So, now, what do these impurities do? — clearly they can do only harm. Maybe they make harm to the tune of 13 percent.” Szilard would continue his quest for materials of higher purity. “There was some considerable gain to be made… there.”

“Well,” concludes Fermi, “this brings us to Pearl Harbor.”

Arthur Compton had less than two weeks to throw together a program between his discussion with Vannevar Bush and James Bryant Conant at the Cosmos Club luncheon on December 6 and the first meeting on December 18 of the new leaders of what was now to be called the S-l program. (S-l for Section One of the Office of Scientific Research and Development: Conant would administer S-l, but the National Defense Research Committee was no longer directly involved; the bomb program had advanced from research into development.) On December 18, Conant notes in the secret history of the project he wrote in 1943, “the atmosphere was charged with excitement — the country had been at war nine days, an expansion of the S-l program was now an accomplished matter. Enthusiasm and optimism reigned.” Compton offered his program to Bush, Conant and Briggs the next day and followed up on December 20 with a memorandum. The projects that had come under his authority were scattered across the country at Columbia, Princeton, Chicago and Berkeley. For the time being he proposed leaving them there.

With the arrival of war, not to breathe a word of the mysteries they were exploring, the project leaders had adopted an informal code: pluto-nium was “copper,” U235 “magnesium,” uranium generically in the nonsensical British coinage “tube alloy.” “On the basis of the present data,” Compton wrote, optimism reigning, “it appears that explosive units of copper need be only half the size of those using magnesium, and that premature explosions can be ruled out.” Because of the difficulty of engineering a remotely controlled chemical plant to extract plutonium, however, he thought that “the production of useful quantities of copper will take longer than the production of magnesium.” For a timetable he offered:

Knowledge of conditions for chain reaction by June 1, 1942.

Production of chain reaction by October 1, 1942.

Pilot plant for using reaction for copper production, October 1, 1943.

Copper in usable quantities by December 31, 1944.

His schedule was designed to show that plutonium might be produced in time to influence the outcome of the war, the standard which Conant was insisting upon after Pearl Harbor even more vehemently than before. But the uranium-graphite work had not yet won even Compton's full confidence. If graphite proved impractical and “copper production” had to wait for heavy water (of which Harold Urey was urging the extraction at an existing plant in Canada), Compton's schedule would slip by “from 6 months to 18 months.” And that might be too late to make a difference.

For the next six months, Compton estimated, the pile studies at Columbia, Princeton and Chicago would cost $590,000 for materials and $618,000 for salaries and support. “This figure seemed big to me,” he remembers modestly, “accustomed as I was to work on research that needed not more than a few thousand dollars per year.”

He had met with Pegram and Fermi to prepare this part of his proposal and concluded that when metallic uranium became available the project should be concentrated at Columbia. Over Christmas and through the first weeks of January it fell to Herbert Anderson, the native son, to find a building in the New York City area large enough to house a full-scale chain-reacting pile. Not to be outdone in the matter of informal codes, the Columbia team had named that culmination “the egg-boiling experiment.” Anderson stumped the wintry boroughs and turned up seven likely locations for boiling uranium eggs. He proposed them to Szilard on January 21; they included a Polo Grounds structure, an aircraft hangar on Long Island that belonged to Curtiss-Wright and the hangar Goodyear used to house its blimps.

But as Compton reviewed the work of the several groups that had come under his authority, bringing their leaders together in Chicago three times during January, their disagreements and duplications made it obvious that all the developmental work on the chain reaction and on pluto-nium chemistry should be combined at one location. Pegram offered Columbia. They considered Princeton and Berkeley and industrial laboratories in Cleveland and Pittsburgh. Compton offered Chicago. No one wanted to move.