The paper shows – contra what has been argued by Trenton Merricks – that counterpart theory, when conjoined with composition as identity, does not entail mereological essentialism. What Merrick’s argument overlooks is that contingent identity is but one of the effects of grounding identity across possible worlds on similarity.

In this paper, I consider the validity and proper formulation of the only-x-and-y principle, which states, roughly, that whether a later individual, y, is numerically identical to an earlier individual, x, can depend only on facts about x and y and the relationships between them. In the course of my investigation, I distinguish between two classes of physical entities - those that exist in a 'real' sense, and those that exist in a mere Cambridge sense. This distinction is grounded in (...) Peter Geach's distinction between 'real' and mere Cambridge change. I argue in favor of a modified version of the only-x-and-y principle - the qualified only-x-and-y principle - which applies to entities that exist in a 'real' sense, but not to mere Cambridge entities. It is also argued that the plausibility of the qualified only-x-and-y principle has more to do with facts about the nature of causality than with intuitions we have about existence or numerical identity. I finish by considering some traditional objections to the only-x-and-y principle, and conclude that they do not succeed in refuting the qualified only-x-and-y principle. (shrink)