Nor would it be safe to assume that the tendencies which have demonstrably operated to falsify the more remote records have been altogether inactive during the last ten years. The causes which tend to sophisticate narratives of this kind, as already shown, are many and difficult to detect; the kind of evidence required to place the alleged death-coincidence beyond reasonable doubt has in some cases never existed; in others, through the destruction of documents, the death of friends, or the mere lapse of time, it is now unattainable. Of the 65 reported coincidences perhaps not more than one-fifth reach the evidential standard of the cases included in this volume. And whilst there is a strong presumption that some proportion of those, which from one or other of the causes suggested inevitably fall below the standard, yet represent facts with substantial accuracy, we have no test which will enable us to determine with precision what narratives and to what extent are worthy of credence. Many of the best-attested cases are printed in full in the Report already referred to, and any reader who is interested in the matter will be able to form an estimate for himself. Meanwhile, an attempt has been made, by means of a careful examination of each narrative in detail, to estimate its evidential value. In the result it would appear that about 44 narratives rest on evidence that may be regarded as fairly good. Of these 44 cases, however, 12 must be struck out, 3 as having been imported into the census,[108] and 9 because a certain amount of anxiety may be presumed to have existed, and may be supposed--though the evidence for such action is very slight--to have caused the hallucination. We thus have 32 cases remaining, in which we have evidence of the occurrence of a hallucination, without apparent cause, within twelve hours of the death of the person seen.