Abstract
According to Kant, arithmetic judgements are not analytic since they are about our practice of operating with figures and things in a certain way. Hence the empiricist thesis that any meaningful assertion is either analytic or synthetic a posteriori seems to be refuted (§§ 1, 2). Using syntax and semantics of truth-conditional logic Frege nevertheless shows that arithmetic can be understood as a system of quasi-analytic sentences speaking about numbers as abstract entities (§§ 3, 4). Axiomatic set theory, however, conceals the connection between (internal) truth-functional arithmetic and our (external) practice of counting and computing (§§ 5-7). --- In spite of the insights truth-conditional semantics provides for a non-psychological understanding of mathematical thinking, it is neither a general theory of meaning and analyticity nor a foundation of a general sense-criterion (§§ 8, 9).