INTRODUCTION

At the 1983 conference of the Auseralasian Association of Philosophy, there were two papers about the project of using abstract entities as ersatz possible worlds. One was mine, which distinguished three versions of ersatzism and raised different objections against different ones. The other was Peter Forrest's, which proposed that structural universals should serve as ersatz worlds; the actualised one is instantiated by the concrete world, the rest are uninstantiated (or instantiated only by proper parts of the concrete world – I omit this complication henceforth).

Forrest and I both wondered where his proposal would fall in my classification, and which of my objections I might raise against it. I found it unexpectedly difficult to give a straight answer. I ended up posing trilemmas, and needing to know more about the doctrine of structural universals on which Forrest's proposal was to be based.

I concluded that, after all, I had little objection to Forrest's use of structural universals, for the most part uninstantiated, as abstract ersatz worlds. Instead, I objected to the structural universals themselves. But I needed to distinguish different versions of the doctrine of structural universals and raise different objections against different ones. And I found that, for the most part, what I had to say would parallel what I had to say against different versions of ersatzism.

Not long before, in ‘New Work for a Theory of Universals’, I had taken a favourable but noncommital view of D. M. Armstrong's theory of universals - a theory which accepts structural universals, though not uninstantiated ones. I said that it gave us one tenable way to draw an indispensable distinction between natural and unnatural classes.