My second nuclear related question of the week! In reading "Making of the Atomic Bomb" by Richard Rhodes, I was struck by a curiosity. The ultimate impetus for the U.S. Manhattan Project was the transmittal of the MAUD Committee's findings, which concluded that the creation of a bomb was not only possible, but feasible, and feasible on a time-frame relevant to the war. MAUD Committee's findings were based on the original Frisch-Peierls Memorandum, which arrived at its conclusions because Peierls decided to calculate what would happen if he assumed a reaction mediated by fast-neutrons in pure U-235; he explicitly discounts the possibility of a slow-neutron reaction in U-235 OR a fast-neutron reaction in U-238/natural uranium.
Why didn't U.S. scientists, notably those who were most concerned about the use of fission for a bomb (e.g., Leo Szilard) emphasize the use of fast neutrons in pure U-235 as proof that a bomb was feasible?
In the book, the pre-MAUD transmittal U.S. work is almost entirely concerned with slow neutron reactions in moderated systems. Even Szilard, who was the most notable figure concerned with the use of fission as a bomb, tended to emphasize the use of slow neutrons in moderated systems?
Because the idea of separating large amounts of U-235 from U-238 seemed pretty infeasible. Remember that before the Frisch-Peierls calculation in 1940 on the critical mass of a pure-U-235 weapon, nobody had ever tried to do that calculation. And so they assumed that the critical mass would be pretty high, similar to the critical mass of an unenriched reactor of some size. But the real impact of F-P was that they found that 1) the reaction might be pretty impressive, and 2) the size of material needed to get the reaction was actually pretty small. F-P actually estimated that the "critical size" was "roughly a pound or so" — a really far cry from what people had expected, and much more achievable-sounding.
By the time of F-P, at least one method — thermal diffusion — had been suggested for enriching uranium that might be feasible, and more would eventually be developed.
So that's the impact of the F-P memo in a nutshell: 1) there are ways to possibly produce enriched uranium, 2) you don't need that much uranium for a bomb, and 3) the bomb would be devastatingly explosive. That's what it is offering up that hadn't been said before.
If you don't think that uranium can be easily enriched, or that you could ever produce that much material, then you are going to focus on situations that don't involve enriching it — like a moderated "reactor-bomb" idea.
Now, the amusing thing about Frisch-Peierls is that its estimates are in retrospect somewhat ridiculously optimistic. Thermal diffusion is pretty crap as a means of enrichment and would never have been adequate; the size of the plants necessary is actually gigantic and enriching uranium is much more prohibitively expensive in all resources than they ever dreamed. And you need a lot more material than they imagined: for U-235, the bare sphere critical mass is 52 kg / 115 lbs, and a working bomb might require more than that based on the design. The Little Boy bomb used 64 kg / 141 lbs.
So the thing that convinced the US to pursue nukes, interestingly enough, was erring on the side of thinking it was too easy. That was not only appealing because of the possibility of the US having nukes, but because the same assumption about it being too easy could be applied to the Germans as well. So it was doubly-encouraging to the Americans who could see the appeal of a bomb in the abstract but especially if they were racing the Germans for it! But it turns out the Germans had a somewhat more accurate understanding of the difficulty of such a program, and did not pursue one of their own.
Often in the history of science and technology it is very difficult to understand why something that later seems obvious didn't seem obvious earlier. In this case, it is because pretty much everyone assumed large-scale uranium enrichment was unlikely to be feasible on any kind of short timeframe.
Just to add more commentary (won't let me edit the original post), the critical question here is about how exactly the U.S. scientists who were most worried about the possibility of fission based bombs (eg, Szilard) thought the bomb was going to be made, if not through fast-fission of U-235. Did they think that it was feasible to do a fast-neutron/U-238 or a slow-neutron/U-235 bomb? If they DID think that the bomb could only be made with fast-neutrons and U-235 (as ended up being the case, and the R-P memorandum uses as a basis), then why was that not emphasized more by them in attempt to shape the early research efforts?