There are two common ways of elucidating the notion of a natural kind, one referring to micro-structure, the other referring to inductive projectibility. I argue that these are not just two ways of saying what natural kinds are, but that they yield two different conceptions. I survey some arguments against the micro-structuralist conception. Then I discuss the relation between the projectibility conception and essentialism. I suggest that a reasonable variant of the projectibility conception allows us to be realists about kinds without embracing essentialism.