A 'gap' test of the sensitivity of explosives to shock is described. An RDX pellet generates a shock in a metal wafer (the 'gap') which is in contact with the sample. A sharp dent in a steel bar at the other end of the sample signifies detonation. Sensitivity is determined in terms of the minimum gap thickness that inhibits detonation. The sensitivity of a mixture of explosives has been found to be simply related to the sensitivities and amounts of its components. Ionization probes and photography have been used to record the course of detonation; a hypothesis explains the observed facts. There is for each charge a critical initial shock velocity below which transition to detonation will not occur. The entering shock wave initiates reactions which release energy. The energy available from these reactions will counteract losses and, if sufficient, reinforce the shock which will then accelerate smoothly into a detonation. The probability that any test will result in a detonation is thus seen to be controlled by the balance of energy lost and gained by the shock wave.