In his influential paper "Mathematical Truth'¹ Paul Benacerraf states two requirements for any account of mathematical truth to be worth considering, namely: (i) that the semantic treatment of mathematical statements does not differ essentially from the semantic treatment of non—mathematical statements, and (ii) that the account of mathematical truth harmonize with what he calls a reasonable epistemology. According to him, combinatorial accounts of mathematical truth, which tend to identify mathematical truth with derivability in a formal system, violate the first requirement, whereas, platonist philosophies of mathematics (like Gödel's) violate the second requirement. Such a violation of the second requirement, however, depends on Benacerraf s understanding of "reasonable epistemology'. It should be clear that if one identifies "reasonable epistemology' with empiricist theory of knowledge (causal or not), platonist philosophies of mathematics are not easy to reconcile with reasonable epistemologies. But such an identification need not be taken for granted.