Scientific realists believe that an ontology adequate for science must include theoretical entities of various kinds, and that it is reasonable to accept such an ontology as the foundation for a general theory of what there is. J. J. C. Smart elaborated this doctrine in his Philosophy and Scientific Realism (1963). The theory he proposed was mainly about what really exists, that is, it had an ontological orientation. Theories with similar orientations have been defended recently by Nancy Cartwright (1983) and Michael Devitt (1984). The theory I called “scientific entity realism” was also a theory of this kind. These theories may be contrasted with those semantically oriented versions of scientific realism that often seem to be more concerned with the theories of truth and reference than with what there is.

The ontology of scientific realism is supported by an argument from the best explanation: if the world behaves as if entities of the kinds postulated by science exist, then the best explanation of this fact is that they really do exist. Properly understood and used, this is a good and powerful argument. But there has been little attempt made to find out what its scope and limitations are; and scientific realists have generally presented it crudely as an argument for the existence of “things like atoms and electrons” – as if this were all that needed to be said. However, the argument is not a good argument for the existence of some kinds of theoretical entities (such as space-time points), and it is a good argument for the existence of certain kinds of properties and relationships that many scientific realists do not believe in.