The prediction of prison riots

Zeeman et al. (1976) proposed a model of the escalation of disorder in a prison and the subsequent riot. The theoretical foundation of this model was an analogy with Lorentz's (1963) study of the behaviour of a cornered and threatened dog and was couched in terms of catastrophe theory (Isnard & Zeeman, 1975). The escalation to a riot in Gartree in 1972 illustrated the uses of their model. Since 1975 there have been two more riots, one in Hull in 1976 and one in Gartree in 1978. In the light of these riots Zeeman's model is reformulated using a utility or pay‐off approach. By adjusting his model in this way we found a more efficient method of monitoring disturbances. The changes in the monitored variables over the time preceding each of these three riots are discussed and conclusions drawn from the new model.