§ 9. Cases of this type are of course, as a class, less satisfactory. It is here that some of the recognised tendencies to error - the impulse to make vague things definite, and the impulse to make a group of facts compact and harmonious - may find their opportunity. The error will, of course, not arise without a certain foundation in fact: the news that a friend has died is not in itself calculated to create a wholly fictitious idea that one has had an unusual experience shortly before the news arrived. But an experience which has been somewhat out of the common may look quite different when recalled in the light of the subsequent knowledge. It may not only gain in significance; its very content may alter. A person perhaps heard his name called when no one was near, and, not being subject to hallucinations of hearing, he was momentarily struck by the fact, but dismissed it from his mind. A day or two afterwards he hears of a friend's death. It then occurs to him that the events may have been connected. He endeavours to recall the sound that he heard, and seems to hear in it the tones of the familiar voice.

Gradually the connection that he has at first only dimly surmised, becomes a certainty for him; and in describing the occurrence, without any idea of deceiving, he will mention his friend's voice as though he had actually recognised it at the time. In the same way something dimly seen in an imperfect light may take for subsequent memory the aspect of a recognisable form; or a momentary hallucination of touch may recur to the mind as a clasp of farewell.

Now such possibilities cannot be too steadily kept in view, during the process of collecting and sifting evidence. At the same time, the interrogation of witnesses, and the comparison of earlier and later accounts, have not revealed any definite instance of this sort of inaccuracy. Now the number of alleged telepathic cases which we have examined (a number of which the [700] narratives given in this book form less than a third) seems sufficiently large for the various types of error that really exist to have come to light; and, as a matter of fact, certain types have come to light, and have helped us to a view of what may be called the laws of error in such matters. If, then, a particular form of inaccuracy is conspicuous by its absence from our considerable list of proved inaccuracies, it may be concluded, we think, not to have been widely operative. It would be a different matter if the cases of the lower evidential class stood alone - if we were unable to present any cases where the percipient's identification of his impression with the particular personality of the agent had been established beyond dispute.

But in face of the large number of those stronger instances, it would be unwarrantably violent to suppose that in all, or nearly all, the other cases where the percipient declares that the identification was clear and unmistakable, he is giving fictitious shape and colour to a purely undistinctive experience.

But there is yet another reason for allowing this inferior evidence to stand for what it is worth. For even if we make very large allowance for inaccuracy, and suppose that in a certain number of these cases the visible or audible phantasm, afterwards described as recognised, was really unrecognised at the moment, the evidence for a telepathic production of it does not thereby vanish. If, indeed, a witness's mental or moral status were such that he might be supposed capable of giving retrospective and objective distinctness to what was an utterly indefinite impression, with no external or sensory character at all, his testimony would, of course, be valueless; simply because we could not assure ourselves that he had not had experiences of that sort daily, so that the coincidence with the real event would lose all significance. But in the case of a witness of fair intelligence, the point remains that the presence of a human being was suggested to his senses in a manner which was in his experience markedly unusual or unique, at the time that a human being at a distance with whom he was more or less closely connected, was in a markedly unusual or unique condition. By itself such evidence might fairly, perhaps, be regarded as too uncertain to support any hypothesis.

But if a ease for telepathy can be founded on the stronger cases, where the immediate reference of the impression to the agent is as much established as the fact of the impression itself, then we have no right to lay down as an immutable law of telepathic experience that such a reference is indispensable. Recognition is beyond doubt the best of tests; and in a vast majority of our cases we have the percipient's testimony, and in a very large number corroborative testimony as well, to the fact of recognition. But distinctness and unusualness in the experience are also evidential points. We have, indeed, a whole class of cases where the percipient has expressly stated that a phantasm which coincided with the supposed agent's death was unrecognised, and where, therefore, the distinctness and unusualness of the impression were the only grounds for paying any attention to the coincidence. Such eases may be far from proving telepathy; yet if telepathy be a vera causa, it would be unscientific to leave them out of account.

§ 10. So much for the evidence of the state of the agent, and of the experience of the percipient, regarded simply as events, of which we want to know (1) to what extent we can rely on the description that we receive of them; (2) to what extent the presumption of a telepathic connection between them is affected by the sort of inaccuracies that may be revealed or surmised. The sketch that has been given is, of course, a mere outline. It must wait for further amplifications of detail till we come to examine the evidence itself. Meanwhile it may serve to prepare the reader's mind, and to indicate what special points to be on the look-out for. But of those four essential items of a case, as to which the opportunities and the effect of misrepresentation were to be specially considered, two still remain, namely, the precise times of the two items already discussed - of the agent's and the percipient's respective shares in the incident. It is clearly essential to a telepathic case that these times should approximately coincide; and error in the assertion of this coincidence is a possibility requiring fully as much attention as error in the description of the two events.