The changing levels of activities in a physical science profession are modelled as a network of relations between different career stages. This cybernetic theory predicts that the dominance of elites undergoes 300 year cycles of sharp alternations whereas the challenge of embryonic elites fluctuates in 100 year cycles. These results seem confirmed by a survey of chemical histories: the birthrate of outstanding chemists oscillates in 300 year cycles and of lesser chemists in 100 year cycles, both with the waveprofile specified by the model. These fluctuations seem to correspond toKuhn's periods of revolutionary and normal science.