A well-known paradox of strict verificationism is this one. Suppose we distinguish between evidence-statements and statements for the truth or falsity of which evidence statements are support, and suppose we could not come to know the non-evidential statements except by knowing the truth of the evidential ones. We must say: what we know is after all some set of evidential statements, and what we mean when we assert the non-evidential statement is after all a set of evidential statements. But the point of calling a set of evidential statements evidential is simply to draw the distinction between evidence and what it is evidence for. What was the initial contrast therefore, and how did we come to think of it as a contrast? If what I mean by S is what would normally be considered evidence for the truth of S, then it cannot be evidence for the truth of S.