In this thesis I investigate Rudolf Carnap's philosophy of mathematics. Most philosophers assume that the nature of mathematics raises deep philosophical questions, which call for theories about how we manage to know about and interact with abstract objects. Carnap's position, in contrast, is deflationary: he aims to show that we can take mathematics at face value without having to answer questions about the metaphysical status of mathematical objects. If Carnap is right, there is thus no need for a philosophy of mathematics as it is usually understood at all. The main argument of my thesis is that Carnap's position is unstable, since his own commitments force him to make at least some ontological assumptions about syntax, i.e. entities such as letters, strings, and proofs.

My claim that Carnap needs to accept some ontological questions as being in good shape goes against the received view in the secondary literature. Since the late 1980s interest in Carnap's philosophy of mathematics has been growing, mostly in the wake of important papers by Michael Friedman, Warren Goldfarb, and Thomas Ricketts. These and other scholars have forcefully defended Carnap against objections by, among others, Kurt Gödel, W. V. Quine, and Hilary Putnam. The most powerful challenge to Carnap's view, however, can actually be found in a less well-known paper by the logician E. W. Beth. The core of my thesis is thus a new interpretation of what I call Beth's argument from non-standard models, which relies on Gödel's incompleteness theorems and targets Carnap's claim that mathematics is analytic. I show that my reconstruction of Beth's argument is more charitable to the text than competing interpretations in the secondary literature, and argue that it is also more powerful since extant defences of Carnap cannot be applied.