making liberal allowance for this unconscious exaggeration, and for another disturbing cause—the possible influence of selection on the results,[1]—the probable number of death coincidences is reduced to 30.
We have then these 30 coincidences with death in 1300 apparitions. But the death rate for the last completed decade (1881–1890) of the period under review was 19.15—i. e., the probability that any person taken at random would die within any given 24 hours was 19.15 in 365,000 = about 1 in 19,000. If there is no causal connection between the hallucination and the death, we should find but I coincidence in 19,000—we actually find 1 in 43.
We may dismiss, then, the suggestion of explanation by chance coincidence. But it need hardly be said that we are not, therefore, entitled to claim that we have found an irrefragable proof of telepathy. The coincidences, it is true, did not occur by chance, if the facts have been correctly reported. But on the one hand, the frequency of non-coin-
- ↑ The non-coincidental hallucinations, which are much less interesting, would probably not be known beforehand to the collector: and even if they were, the collector would not be likely to go out of his way to collect such an account. Further, apparitions at the time of death are naturally more talked about, the collectors would probably know of some such amongst their acquaintance, and unless, in recording the answers, they systematically canvassed the whole of the neighbourhood accessible to them, it is almost certain that they would yield to the temptation to "bag" a death coincidence. even though it did not, properly speaking. come within their ground. See Proceedings, S. P. R., vol. x.. pp. 210 and 243.
of age are excluded. there are 26 years included in the remoter period. If 11 experiences occur in to years. we should look for 29 at most in the remaining 26 years—we find 51!
