van Fraassen’s overall view of science is known as constructive empiricism, whose cental tenet is that theories are formulated in order to be empirically adequate with respect to observational data or phenomena. The instance view of explanation presupposes that there are laws of nature - otherwise there would be no explanations. van Fraassen agrees with the Bayesian in that the starting point of analysis is an initial state of opinion which comprises probability assignments to a range of initial opinions. In any case a good empiricist like van Fraassen will demand something more solid that the leap of faith which it seems is needed to appreciate universals. The most pressing difficulty is one that van Fraassen mentions in passing, which is the problem of distinguishing regularities that really are laws from those that are accidents. The role of theoretical terms is to organise the embedding’ of empirical substructure in full models of the theory.