Wittgenstein is accused by Dummett of radical conventionalism, the view that the necessity of any statement is a matter of express linguistic convention, i.e., a decision. This conventionalism is alleged to follow, in Wittgenstein's middle period, from his ‘concept modification thesis’, that a proof significantly changes the sense of the proposition it aims to prove. I argue for the assimilation of this thesis to Wittgenstein's ‘no-conjecture thesis’ concerning mathematical statements. Both flow from a strong verificationist view of mathematics held by Wittgenstein in his middle period, and this also explains his views on the law of excluded middle and consistency. Strong verificationism is central to making sense of Wittgenstein's middle-period philosophy of mathematics.