§ 1. When we pass to spontaneous exhibitions of telepathy, the nature of the evidence changes; for the events are described by persons who played their part in them unawares, without any idea that they were matter for scientific observation. The method of inquiry will now have to be the historical method, and will involve difficult questions as to the judgment of human testimony, and a complex estimate of probabilities ....... 95-96

§ 2. The most general objection to evidence for phenomena transcending the recognised scope of science is that, in a thickly populated world where mal-observation and exaggeration are easy and common, there is (within certain limits) no marvel for which evidence of a sort may not be obtained. This objection is often enforced by reference to the superstition of witchcraft, which in quite modern times was supported by a large array of contemporary evidence ........... 96

But when this instance is carefully examined, we find (1) that the direct testimony came exclusively from the uneducated class; and (2) that, owing to the ignorance which, in the witch-epoch, was universal as to the psychology of various abnormal and morbid states, the hypothesis of unconscious self-deception on the part of the witnesses was never allowed for . . . . 96-97

Our present knowledge of hypnotism, hysteria, and hystero-epilepsy enables us to account for many of the phenomena attributed to demonic possession, as neither fact nor fraud, but as bona fide hallucinations . . . 97

While for the more bizarre and incredible marvels there is absolutely no direct, first-hand, independent testimony ..... 98-99

The better-attested cases are just those which, if genuine, might be explained as telepathic; but the evidence for them is not strong enough to support any definite conclusion ......... 99-100

§ 3. The evidence for telepathy in the present work presents a complete contrast to that which has supported the belief in magical occurrences. It comes for the most part from educated persons, who were not predisposed to admit the reality of the phenomena; while the phenomena themselves are not strongly associated with any prevalent beliefs or habits of thought, differing in this respect, e.g., from alleged apparitions of the dead. Still we must not, on such grounds as these, assume that the evidence is trustworthy . . 101-102

§ 4. The errors which may affect it are of various sorts. Error of observation may result in a mistake of identity. Thus a stranger in the street may be mistaken for a friend, who turns out to have died at that time, and whose phantasm is therefore asserted to have appeared. But it is only to a very small minority of the cases which follow that such a hypothesis could possibly be applied 102-103

Error of inference is not a prominent danger; as what concerns the telepathic evidence is simply what the percipient seemed to himself to see or hear, not what he inferred therefrom ....... 103-104

§ 5. Of more importance are errors of narration, due to the tendency to make an account edifying, or graphic, or startling. In first-hand testimony this tendency may be to some extent counterbalanced by the desire to be believed; which has less influence in cases where the narrator is not personally responsible, as, e.g., in the spurious and sensational anecdotes of anonymous newspaper paragraphs, or of dinner-table gossip ...... 104-106

§ 6. Errors of memory are more insidious. If the witness regards the facts in a particular speculative or emotional light, facts will be apt, in memory, to accommodate themselves to this view, and details will get introduced or dropped out in such a manner as to aid the harmonious effect. Even apart from any special bias, the mere effort to make definite what has become dim may fill in the picture with wrong detail; or the tendency to lighten the burden of retention may invest the whole occurrence with a spurious trenchancy and simplicity of form ............ 106-108

§ 7. We have to consider how these various sources of error may affect the evidence for a case of spontaneous telepathy. Such a case presents a coincidence of a particular kind, with four main points to look to: - (1) A particular state of the agent, e.g., the crisis of death; (2) a particular experience of the percipient, e.g., the impression of seeing the agent before him in visible form; (3) the date of (1); (4) the date of (2) . . . . . . . . . 108

§ 8. The risk of mistake as to the state of the agent is seldom appreciable: his death, for instance, if that is what has befallen him, can usually be proved beyond dispute .......... 108-109

For the experience of the percipient, on the other hand, we have generally nothing but his own word to depend on. But for what is required, his word is often sufficient. For the evidential point is simply his statement that he has had an impression or sensation of a peculiar kind, which, if he had it, he knew that he had; and this point is quite independent of his interpretation of his experience, which may easily be erroneous, e.g., if he attributes objective reality to what was really a hallucination . . . . • . 109

The risk of misrepresentation is smallest if his description of his experience, or a distinct course of action due to his experience, has preceded his knowledge of what has happened to the agent ...... 109-110

§ 9. Where his description of his experience dates from a time subsequent to his knowledge of what has happened to the agent, there is a possibility that this knowledge may have made the experience seem more striking and distinctive than it really was. Still, we have not detected definite instances of this sort of inaccuracy. Nor would the fact (often expressly stated by the witness) that the experience did not at the time of its occurrence suggest the agent, by any means destroy - though it would of course weaken - the presumption that it was telepathic . ........ . 111-112