The total number of recognised apparitions of living persons recorded at first-hand as occurring in the circle of 17,000 persons from which these death-coincidences were drawn, is 322.[109] But if, in order to allow for forgetfulness, as already indicated (p. 221), we multiply the number recorded by 4, we shall arrive at a total of 1288, as representing the probable number actually experienced by our informants since the age of ten. We have, therefore, 32 cases of hallucinations coinciding with the death of the person seen, in an estimated total of nearly 1300 recognised apparitions of living persons--or about 1 in 40. But the death-rate for England and Wales in the last completed decade being 19.15 per 1000 per annum, the average probability that any particular person will die on any particular day is 1000 × 365 / 19.15 = about 1 in 19,000. That is, there is one chance in 19,000 that a man will die on the day on which his apparition is seen and recognised, supposing there to be no causal connection between the two events. Or in other words, for every hallucination which coincides with the death of the person seen, we should have to find about 18,999 similar hallucinations (_i.e._, recognised apparitions of living persons) which do not so coincide. But after making due allowance for forgetfulness on the one hand, and for the creative activity of the imagination on the other, we find the actual proportion to be 1 to 40. In the face of these figures it would be preposterous to ascribe the reported cases of hallucinations at the time of death to chance. And the argument for some causal connection between hallucinations and external events is of course considerably strengthened if, in addition to (_a_) the coincidences of visual hallucinations with death, we take account of (_b_) the coincidences of auditory hallucinations with death, and (_c_) the coincidences of both visual and auditory hallucinations with other events than death, and (_d_) the cases in which the coincidence of the apparition with the death is nearer than twelve hours, the limit assumed in the above calculations. It may not be superfluous to repeat (see _ante_, p. 27, footnote) that the calculation above given does not purport to establish thought-transference as the cause of these coincidences. The cause may be a greater prevalence of exaggeration and memory-illusion than the Committee have allowed for. What the calculation does is to bring us face to face with the problem: Here are certain phenomena, demonstrably not due to chance: do they reveal a new mode of communication between human minds, or merely a new source of fallacy in human testimony? It will hardly be disputed that, in either event, to find an answer to the question will justify much labour spent upon the search. FOOTNOTES: [Footnote 93: See _Animal Magnetism_, by Binet and Féré, in the International Science Series, and the references there given.]