members to be formed. The logical form of the membership predicate in English
does not match the logical form of the membership relation. In the metaphysically
perfect language, these two would not come apart.
A Rylean view of the metaphysically perfect language is attractive. What makes a
“category mistake”such as“my table has three subsets”worse than other falsehoods
is that no paraphrase of such a sentence into the metaphysically perfectly language is
even in principle possible. (Sometimes what can be said should not be said.) Any
purported paraphrase of that sentence would require conjoining incompatible intra-
categorial expressions. Our language would be better off (metaphysically speaking)
were Ryle to be right about it.
Recall the general principle of recombination wefirst considered: any logically
consistent sentence in the metaphysically perfect language is possibly true. Assume
that“is a member of”is a perfectly natural expression, as are the proper names of
fundamental particles. It might initially seem then that this principle of recombin-
ation licenses the claim that“this electron is a member of this proton”is a possibly
true sentence. However, on the picture described moments ago, this sequence of
expressions is not even a sentence in the metaphysically perfect language, let alone an
atomic sentence. And so the powerful principle of recombination does not license the
claim that some electron is a member of some proton.^52
On a view like this, there is an intimate connection between logical possibility and
metaphysical possibility: all logically consistent sentences in the metaphysically
perfect language express metaphysically possible propositions. In short, in the meta-
physically perfect language, logical possibility suffices for metaphysical possibility.
However, one cannot straightforwardly claim that a proposition is metaphysically
possible if and only if it is expressible by a logically consistent sentence in the
metaphysically perfect language. The English sentence“No electron has a member”
expresses a metaphysically necessary, and hence metaphysically possible, truth. Yet
this truth isnotexpressible in the metaphysically perfect language. So we need further
principles to bridge logical possibility and metaphysical possibility if we hope to offer
a reductive analysis of metaphysical possibility in terms of logical possibility. My
hope is that pursuing such principles might be a fertile research project.
Note that for similar reasons we can’t say that anecessarycondition for being
metaphysically possible is having a logically consistent translation in the metaphys-
ically perfect language, since, as just noted,“no electron has a member”is necessarily
true (and hence possibly true) and yet lacks a translation.^53 In the remainder of this
section, I will focus on strategies for formulating an appropriate necessary condition
for metaphysical possibility. Finding such a condition would be a goodfirst step
(^52) Recall that we might wish to restrict this principle further, so that it generates onlyde dictomodal
truths. As suggested earlier, one plausible way of restricting it is to have it apply to sentences without
proper names or other rigid designators. 53
Such sentences lack what Sider (2011) calls a metaphysical semantics.