MM Bennett: "Existential" but Analytic Statements

I would like to post now about (ii). Although Karen Bennett suggests (p. 2) that she would argue against semanticism in general, in effect she seems to argue just against one possible way of implementing one possible semanticist position with respect to one particular debate—in particular against the analyticity-involving way of implementing Hirsch 2005’s semanticist position with respect to the composition debate. Even when so restricted, I have some doubts about her argument.

In essence, her claim is that what she calls ‘linking principles’ of the sort of

(*) if there are simples arranged tablewise in R, then there is a table in R that is numerically distinct from the simples arranged tablewise.

cannot be analytic. She offers the following reason for this claim:

Saying that (*) is analytic … amounts to saying that we can define things into existence. But surely an analytic claim cannot be existence entailing in this way; surely the existence of a new object cannot follow by meaning alone. Who knew ontological arguments were so easy? (p. 19)

I find this reasoning puzzling. The relevant sentences involve existence claims, but as consequents in conditionals. And we seem to be familiar enough with existential statements of that form being, sometimes, analytic. The following seems to me to have quite a good claim to be one such:

Whenever something is a proper part of another, there is something that is numerically distinct from them which is part of the latter but not of the former.

(I am having some discussion with Ross Cameron at bleb on this, as it seems to me to be relevant also against his argument that principles of composition need not be necessary.)