That no member of a natural kind can switch kinds is a consequence of David Wiggins’ view that the identity conditions for such things are given by the natural kind itself. If dog is a natural kind, then dogs must be dogs and one dog cannot ‘turn into’ something else, say, by gradually ‘becoming’ a mass of tissue (as Marjorie Price had held). Were such a transition to involve the persistence of the same thing, then the thing in question would (...) comply with conditions of identity not restricted to dogs. To provide for such transitions James Baillie suggests alternative conditions of identity, based on Andrew Brennan’s notion of ‘survival’ (itself borrowed from Parfit). Survival requires that the earlier thing be ‘structurally similar to’, and ‘play a significant and direct role in the production of’ the later thing (sufficiency is gained if, in addition, they are composed of the same kind of matter). Unlike identity, however, survival is what Wiggins calls a ‘branching’ relation, capable of linking a single thing with two or more equal ‘competitors’ (as apparently happens in cases of brain fission). To provide conditions of identity per se, Baillie follows Parfit, who claimed that our identity is preserved by psychological continuity when there is no branching. Members of natural kinds, such as dogs, are now said to be reidentifiable over time by the relation of non-branching survival. I argue against this on the ground that we can know when this relation obtains only when we are already able to individuate things by a different sort of relation, such as those associated with natural kinds. (shrink)