There is a well-known tension within Sellars' scheme arising from commitments to both an anti-foundationalist epistemology and a Peircean scientific realism. This tension surfaces conspicuously in his treatment of ontological category theory. On the one hand, Sellars applies and extends Carnap's metalinguistic deflation of ontology. On the other hand, however, Sellars is not prepared to 'go conventionalist' but upholds the possibility of a "positive ontology" (Rosenberg). I offer a new reading of Sellars’ Carus Lectures in which I combine two projects. (...) First, I argue that Sellars provides us here with the sketch of a method of ‘category projection’ which can be used, within the setting of Sellars' scheme, to 'transcend from within' the limitations of category theories developed in non-Peircean conceptual structures and to enable us non-Peirceans to make any justifiable descriptive claims about the structure of reality. In the course of doing so I also offer a new reading of Sellars' Carus Lectures, highlighting the systematic advantages that motivated Sellars to view a process ontological interpretation of sensation. (shrink)