Full Idea
I don't think all substitution-instances of a valid schema are 'true'; some are clearly meaningless, such as 'If all boojums are snarks and all snarks are egglehumphs, then all boojums are egglehumphs'.
Gist of Idea
Having a valid form doesn't ensure truth, as it may be meaningless
Source
Hilary Putnam (Philosophy of Logic [1971], Ch.3)
Book Reference
Putnam,Hilary: 'Philosophy of Logic' [Routledge 1972], p.28
A Reaction
This seems like a very good challenge to Quine's claim that it is only form which produces a logical truth. Keep deductive and semantic consequence separate, with two different types of 'logical truth'.