The horizontal control rods and vertical safety rods were amply able to shut down the pile, as long as it was possible to insert them into their thimbles in the pile. However, this might not be possible in the event of a major disruption to the pile, such as an earthquake, enemy bombing, or internal explosion. If the pile shifted several inches, the rod thimbles might be shattered and the rods would then be useless.
Initally, a third pile-control mechanism consisted of a group of five 105-gallon tanks sitting on top of the pile. Each was filled with a 1.0 to 1.5 percent boron solution, kept under 75 psi of air pressure. Six pipes ran from each tank to six of the VSR thimble openings. In the event of a safety circuit trigger or a manual command from the control room, a fast-acting Belfield valve on each tank would open, and the boron solution would run into all the VSR thimbles, whether the rods were inserted or withdrawn from the pile.
This liquid poison, although never needed for an emergency, posed a potential problem several years later when the VSR thimbles began to leak helium from the pile. If the solution had been dumped into a leaking thimble and spread to the graphite in the interior of the pile, the boron could have severely degraded the pile’s ability to maintain a chain reaction.
Additionally, due to the higher power levels achieved in B Reactor throughout the late 1940s and mid-1950s , Hanford operators expressed concern that, at the higher operating temperatures, the boron liquid would "flash" to steam at initial contact with the hot aluminum thimbles that lined the VSRs. If this happened the solution could boil and there might not be enough liquid left to shut down the pile. Furthermore, the vapor formed from the boron solution might rupture the thimbles, thus wetting the graphite. This risk was considered so severe that operators did not dare to test the third safety system at all after the summer of 1950.
As a result, the liquid boron was replaced with 29 "ball hoppers" (one at the top of each VSR channel) that contained 3/8-inch to 7/16-inch nickel-plated carbon steel balls. These balls, which also acted to shut down the pile through neutron absorption, could funnel down into the VSR channels in the event of an emergency or a test. The balls could then be removed by a vacuum system. In January 1952, B Reactor became the first to be fitted with the new "Ball-3X" system.