<p>This paper examines philosophical and empirical discussions of the beauty of mathematical proofs. Drawing on Wittgenstein’s later philosophy of mathematics and aesthetics, it argues that a recurrent philosophical temptation is to treat beauty-talk as if it must identify something: a property of mathematical reality, a structural profile of proofs, or a psychological/social state of the appraiser. This is not intended as a criticism of empirical studies, which operationalise and test hypotheses about mathematical appraisal. The target is instead a philosophical inference sometimes drawn from beauty-talk, or from data about beauty-judgements, to the conclusion that beauty must have a single underlying nature. Against that temptation, the paper proposes a Wittgensteinian alternative. On the view defended here, to call a proof beautiful is not primarily to attribute an intrinsic aesthetic property to it, but to make a practice-internal normative appraisal. Such judgements can endorse mathematical routes, calibrate attention, transmit standards, and help form mathematical taste, while remaining answerable to reasons without presupposing a single essence of beauty. The beauty of proofs is thus grammatical rather than ontological.</p>