In the last chapter of Representation and Reality Hilary Putnam remarks that reason can transcend whatever bounds we set for it. That is: for any theory of what reason dictates in some given domain–arithmetic, meaning, though in general–reason is liable to dictate more than that theory generates. This could be read as just a picturesque way of putting a well-known formal result. But Putnam does not so mean it. (His immediate target here is the idea that representation–in particular, in attitudes towards the world–reduces to something else (specifically 'functional role').) I think the slogan is a useful way of getting at what has always been most central in Putnam's thought. The first step in getting at this is to try to demarcate some area in which there is room for reason to transcend any bounds a theory of meaning might set for it. The obvious place to look is in a theory of the extensions (to use a familiar term) fixed in words meaning what they do, or in concepts being the concepts they are. The relation between meaning, or concepts, and extensions is one between ways of representing things as being and historical episodes or their participants. When one thinks of things in this way, it is difficult to see how reason could fail to transcend any bounds we set (or how we could even set very much in the way of bounds). This is a good framework in which to place Putnam's discussions of (to put it tendentiously) the world-dependence of conceptual structure (or structures). Which is, in turn, a good way to see the right reaction to the problems which have emerged, since 1905, with traditional notions of analyticity, necessity and a priority.