in some sense a good reason for thinking them true. Accordingly I am going to assume that many propositions of the following kind are true. Propositions, namely, which assert that one man would probably not have certain perceptions which he does have, unless some other man had certain particular perceptions. That some of you, for instance, would probably not be having precisely the perceptions which you are having, unless I were having the perception of more or less black marks on a more or less white ground. And, in this sense, I say, we certainly have reasons for supposing that other people have perceptions similar, in some respects, to those which we sometimes have. But when I said I was going to ask the question: What reason have we for supposing that other people exist? you will certainly not have thought that I merely meant to ask the question which I have just answered. My words will have suggested to you something much more important than merely this. When, for instance, I said that to the question "How do you know that?" the answer "I saw it in the _Times"_ would be a satisfactory answer, you may have felt, as I felt, that it would not in all circumstances be regarded as such. The person who asked the question might, in some cases, fairly reply: "That is no answer: how do you know that, because you saw a thing in the _Times_, it is therefore true?" In other words he might ask fora _reason_ for supposing that the occurrence of a particular statement in the _Times_ was a reason for supposing that statement true. And this is a question to which we all believe that there may be an answer. We believe that, with some kinds of statements which the _Times_ makes--some kinds of statements with regard to Fiscal Policy for example--the fact that the _Times_ makes them is no reason for supposing them to be true: whereas with regard to other kinds of statements, which it makes, such a statement, for instance, as that the King was dead, the fact that it makes them _is_ a reason for supposing them true. We believe that there are some kinds of statements, which it is very unlikely the _Times_ would make, unless they were true; and others which it is not at all unlikely that the _Times_ might make, although they were not true. And we believe that a reason might be given for distinguishing, in this way, between the two different kinds of statement: for thinking that, in some cases (on points, for instance, which, as we should say, are not simple questions of fact) the _Times_ is fallible, whereas in other cases, it is, though not absolutely infallible, very unlikely to state what is not true.