On learning of the nuclear fission discovery, Fermi, working with Szilard, immediately added a new hypothesis to his list. His tentative explanation for the high energy generated by the uranium atom on splitting assumed that the original single atom causing the split produced two neutrons. It would be possible that these two would then collide with other uranium atoms and produce four neutrons, and so the chain of reactions would grow as would the number of neutrons. The process would continue until all uranium atoms were used up. Each nuclear collision releases huge amounts of energy.
Fermi's hypothesis described an ideal condition. It had to be acknowledged that in real situations, energy production and reaction rates would be reduced by missed collisions due to high nuclear speed and slowed interaction from intra-nuclear absorption, as Fermi had already discovered in his paraffin wax experiments.
Experimental confirmation required a reliable neutron source, a suitable vessel to house the chain reaction, and a method of controlling, or moderating, the reaction rate. The neutron source was created by bombarding beryllium with highly accelerated deuterons (nuclei of deuterium atoms) created in a cyclotron. The vessel was a room-size pile of ultra-pure graphite seeded with a lattice of uranium and the moderators were removable graphite rods inserted at intervals in the pile.
The chain reaction occurs at critical mass, that point at which there are just sufficient neutrons to sustain the reaction after accounting for neutron loss from the pile due to escape and absorption. The pile that provided the first chain reaction was a spherical pile almost 26 ft. in diameter.