In the first place, what do we mean by the initial credibility of an event? There are very few cases where this notion can be precisely defined. The simple instances dealt with in the elements of mathematical probability do, it is true, permit of precise definition. The chance that a white ball will be drawn from an urn containing five black balls and one white ball can be exactly estimated, for we are in possession of all the very simple relevant factors. But the probability that Romulus founded Rome obviously belongs to a very different category. And what is the initial credibility of a miracle? Hume, as is well known, thought that the _a priori_ incredibility of a miracle was so great that its occurrence could not be established by human testimony. He is here trying to establish a ratio between the initial credibility of a class of events and the initial credibility of human testimony to such events. He is taking some kind of average in both cases, but it is difficult to see how such an average can be arrived at. Vague considerations of this kind are of no value in forming conclusions on matters of real interest to us, although they may be sufficient to warrant a lazy scepticism regarding what William James calls “dead hypotheses,” or may form the basis for amusing and ingenious mathematical exercises. But we have no notion of an average initial credibility which is of any use in practice; each case must be judged on its own merits. And if, to take the second point, we reached some average for the value of human testimony in general, we should never, in practice, apply it. The utmost we can hope to do is to establish a more or less constant relation between the testimony of classes of witnesses and classes of events. We have to divide witnesses into types, and for each given type estimate the value of its testimony to different classes of events. We must investigate the difference it makes when the witness is taken as isolated and when he is taken as a member of a group of witnesses. In this way we may hope to reach results which are of value in judicial procedure, in the study of history, and in various particular investigations, including those modern substitutes for miracles, the phenomena of spiritualism. We are, in fact, to investigate man in his capacity as a truth-recording instrument.