Abstract

In a naïve realist approach to reading an ontology off the models of a physical theory, the invariance of a given theory under permutations of its property-bearing objects entails the existence of distinct possible worlds from amongst which the theory cannot choose. A brand of Ontic Structural Realism attempts to avoid this consequence by denying that objects possess primitive identity, and thus worlds with property values permuted amongst those objects are really one and the same world. Assuming that any successful ontology of objects is able to describe a universe containing a determinate number of them, I argue that no version of OSR which both retains objects and understands ‘structure’ in terms of relations can be successful. This follows from the fact that no set of relational facts is sufficient to fix the cardinality of the collection of objects implied by those facts. More broadly, I offer reasons to believe that no version of OSR is compatible with the existence of objects, no matter how ontologically derivative they are taken to be. Any such account would have to attribute a definite cardinality to a collection of objects while denying that those objects are possessed of a primitive identity. With no compelling reason to abandon the classical conception of cardinality, such an attribution is incoherent