§I. Anti-realism of the sort which Michael Dummett has expounded takes issue with the traditional idea that an understanding of any statement (here, declarative sentence) is philosophically correctly analysed as involving grasp of conditions necessary and sufficient for its truth. Many kinds of statement to which, as we ordinarily think, we attach a clear sense would have to be represented, according to this tradition, as possessing verification-transcendent truth-conditions; if true that is to say, they would be so in virtue of circumstances of a type transcending our range of possible awareness. Exactly where to draw the boundaries of our possible awareness might be controversial; but there is clearly no being aware, in the relevant sense, of the kind of state of affairs which would make true a generalization of theoretical physics, an assertion about James II weight on his twenty-eighth birthday, a claim about what would have happened if Edward Heath had not sought a fresh mandate during the miners' strike, or—from your point of view—the statement that my left ear aches. In each of these kinds of case the traditional view, while granting that we (or you) cannot experience the truth-conferring states of affairs as such, would nevertheless credit us with a clear conception of the type of thing they would be. To be sure, there is then no possibility of a straightforward construal of this conception as a recognitional capacity. But the traditional view tends to conceal from itself the problematic status which the alleged grasp of truth-conditions therefore assumes by working with the picture that the ‘conception’ is indirectly recognitional, that it issues in a cluster of unproblematic recognitional capacities; in particular, the ability to recognize what is or is not good evidence for the relevant statement and the ability to recognize its logical relations to other statements.