This article critically examines SaulKripke's (1972) argument for the separability ofnecessary truths from truths known a priori,focusing on his criticism of the standard meter casepresented by Wittgenstein (1968). It attempts to showthat Kripke's argument is unworkable on any of severalreadings. Wittgenstein's own broadly conventionalistaccount of necessary truth is then considered in thelight of the standard meter example.