Wolfgang Stegm?ller is one philosopher who has consistently chal? lenged the orthodox view that the reduction of one theory to another must embody the derivation of laws, and, where appropriate, the definability or translatability of primitive terms, as well as the idea that no member of a pair of incommensurable theories may be reduced to the other. As I understand him, Stegm?ller is not merely attempting to modify our customary usage of terms like 'reduction' and incom? mensurability'; though this would be a consequence of his proposal. His chief aim, I take it, is to argue that the correct response to the incommensurability thesis is not to denounce it as false or incoherent, but to provide a rationalist account of scientific progress even in the face of revolutionary changes of theory, that is even in the presence of potential or actual incommensurability. In the process, Stegm?ller hopes to defend Kuhn against the charges of subjectivity and irra? tionality. The task is a formidable one, and Stegm?ller approaches it by urging nothing less than a complete revision of our concepts of 'theory', 'paradigm', 'reduction', and so forth; the idea being that, by switching to a new metascientific paradigm, we can make sense of the claim that, whilst a scientific theory may be incommensurable with its predecessor, still the latter may be reduced to the former so that progress can be seen to have been achieved. The new metascientific framework that Steg? m?ller propounds is of course the structuralist view of theories. Some time ago I carried out a fairly extensive study of the struc? turalist concept of reduction,1 and I later applied some results of that study to shed doubts on the adequacy of Stegm?ller's defence of Kuhn,2 at least those aspects of it that hinge on reduction. What I claimed, very roughly, was that if a theory T" is reducible to a theory T according to structuralist criteria, then there will be a translation of a certain sort which correlates every sentence in the language of V with some sentence in the language of T. But if the laws of V can be translated into the framework of T, then, or so I argued, the two theories can only, contrary to Stegm?ller's intentions, be commensurable. This is the gist of my claim; but the argument rests on several technical assumptions