In the first place, then, when I talk of "a reason," I mean _only_ a good reason and _not_ a bad one. A bad reason is, no doubt, a reason, in one sense of the word; but I mean to use the word "reason" exclusively in the sense in which it is equivalent to "good reason." But what, then, is meant by a good reason for a belief? I think I can express sufficiently accurately what I mean by it in this connection, as follows:--A good reason for a belief is a proposition which is true, and which would not be true unless the belief were also true. We should, I think, commonly say that when a man knows such a proposition, he has a good reason for his belief; and, when he knows no such proposition, we should say that he has no reason for it. When he knows such a proposition, we should say he knows something which is a reason for thinking his belief to be true--something from which it _could_ be validly inferred. And if, in answer to the question "How do you know so and so?" he were to state such a proposition, we should, I think, feel that he had answered the question which we meant to ask. Suppose, for instance, in answer to the question "How do you know that?" he were to say "I saw it in the _Times."_ Then, if we believed that he had seen it in the _Times_, and also believed that it would not have been in the _Times_, unless it had been true, we should admit that he had answered our question. We should no longer doubt that he did _know_ what he asserted, we should no longer doubt that his belief was true. But if, on the other hand, we believed that he had not seen it in the _Times_--if, for instance, we had reason to believe that what he saw was not the statement which he made, but some other statement which he mistook for it; or if we believed that the kind of statement in question was one with regard to which there was no presumption that, being in the _Times_, it would be true: in _either_ of these cases we should, I think, feel that he had _not_ answered our question. We should still doubt whether what he had said was true. We should still doubt whether he _knew_ what he asserted; and since a man cannot tell you how he _knows_ a thing unless he does know that thing, we should think that, though he might have told us truly how he _came to believe it,_ he had certainly not told us how he _knew_ it. But though we should thus hold that he had _not_ told us _how he knew_ what he had asserted, and that he had given us no reason for believing it to be true; we must yet admit that he had given us a reason in a sense--a _bad_ reason, a reason which was no reason because it had no tendency to show that what he believed was true; and we might also be perfectly convinced that he had given us _the reason_ why he believed it--the proposition by believing which he was induced also to believe his original assertion.