As regards the accuracy of the records, though it has been possible to draw up a sort of table of degrees, such a table affords, of course, no final criterion. Each case must be judged on its merits by reference to a considerable number of points. It is essential for judgment that the narrative should be given in the percipient's own words, - not converted into second-hand evidence by being paraphrased. This principle has been followed throughout, as well as that of obtaining, whenever possible, corroborative evidence of all sorts, whether from private sources, public notices, or official records.

Further, a very large proportion of the narratives in Phantasms have stood the test of cross-examination of the witnesses in personal interviews. This part of the investigation was carried out by the authors of the book and their colleagues, especially Professor and Mrs. Sidgwick, and it greatly added to their own confidence that the testimony they published came from trustworthy and intelligent witnesses. The practice of making the personal acquaintance of the witnesses has been continued as far as possible with all the best evidence received since by the Society for Psychical Research.

Another argument for the general trustworthiness of the evidence is the fact that amid all their differences, the cases present one general characteristic - an unusual affection of one person, having no apparent relation to anything outside him except the unusual condition, otherwise unknown to him, of another person. This characteristic gives them the appearance of a true natural group, and involves the hypothesis that the facts, if truly stated, are probably due to a single cause. It involves further, a very strong argument that the facts are truly stated; since it is extremely unlikely - if all the accounts are erroneous - that the various supposed errors of inference, lapses of memory, and exaggerations and perversions of narration should issue in a consistent body of evidence, presenting one well-defined type of phenomenon, free in every case from excrescences or inconsistent features, and explicable and completely explicable by one equally well-defined hypothesis. What is the likelihood that a number of narratives, if we assume them to have diverged in various ways from the actual facts, should thus converge to a single result ? We find that all of them stop short at or within a given line - the line being the exact one up to which a particular explanation, not of the witnesses but of ours, can be extended, and beyond which it could not be extended.

Tempting marvels lie further on - marvels which in the popular view are quite as likely to be true as the facts actually reported, and which the general traditions of the subject would connect with those facts. But our reporters one and all eschew them. To take, for instance, he group of cases which the reader will probably find to be the most interesting, as it is also the largest, in our collection - apparitions at the time of death. Why should not such apparitions hold prolonged converse with the waking friend ? Why should they not produce physical effects - shed tears on the pillow and make it wet, open the door and leave it open, or leave some tangible token of their presence? It is surely noteworthy that we have not had to reject, on grounds like these, a single narrative which on other grounds would have been admitted.

1 Except in the Supplement, which includes a good many second-hand accounts, as well as first-hand accounts where the evidence, from lack of corroboration or other causes, falls short of the standard attained in the body of the work. The principle in selecting cases for the Supplement was to take only those which - supposing telepathy to be established as a fact in Nature - would reasonably be regarded as examples of it.

On the other hand, these details are found abundantly in second-hand cases, and they are precisely of the sort which the telepathic hypothesis could by no possibility be made to cover. The existence of such features in second-hand narratives shows how wide is the possible range of incidents in stories where ordinary human faculties are alleged to have been transcended. Of this wide field, the hypothesis of the action of mind on mind covers only a single well-defined portion. We can hardly suppose, then, that if error were widely at work in the case of our first-hand evidence, its results would always fall inside and not outside this very limited area - should all, that is, conform to the purely telepathic type.

Meanwhile, it is not, of course, claimed that the evidence is such as must convince every candid inquirer, and after setting forth the standard desired, and discussing the force of that actually attained, we may pass on to a consideration of some of the principal criticisms that have been directed against the latter.