hazards. It would be almost impossible to demonstrate empirically that altering remuneration systems in this way will reduce the number of major accidents. But the recommendation does not depend on such evidence. It is based on a theory of human motivation, viz, that people are likely behave in ways that please their boss, especially when their material interests are dependent on pleasing the boss. The strength of the recommendation depends on the strength of this theory of human motivation. For many people the theory will be regarded as self-evident, without need for further evidence. The logic here is that the particular accident analysis has shown how a certain causal factor can work and the theory of human motivation has provided the basis for generalisation and for the belief that if this recommendation were implemented the risk of major accident would be reduced.