[83] See, for example, the case quoted in Chapter X., No. 63.
[84] Phantasms of the Living, vol. i. pp. 164-166.
[85] Proceedings American S.P.R., pp. 350, 351.
[86] Two other examples are referred to in Phantasms, vol. i. p. 189, but in neither case is the evidence obtainable at first-hand.
[87] Except, of course, in cases of rudimentary hallucinations, such as after-images and bright spots in the eyes and singing in the ears, which are caused by the physical condition of the external organ.
[88] See case No. 51, later; and compare Mr. Galton's observations in his lecture at the Royal Institution on "The Just Perceptible Difference" (reported in the Times, January 30th, 1893). Mr. Galton found that the ideal auditory impressions called up by reading the printed substance of a lecture enabled him to hear the lecturer's voice at a greater distance than when he had not the printed text before him; the ideal appears to have supplemented the real impression, as, in the case given in the text, the real reinforced the ideal.
[89] Phantasms, vol. i. pp. 303-310. The statement in the text must not be regarded as having more weight than its author himself would have assigned to it. Mr. Gurney certainly regarded his estimate as little more than a guess—a guess indeed made by one who had carefully studied and weighed the facts, so far as they could be known, but because of our inevitable ignorance a guess still, rather than an estimate on the approximate accuracy of which it would be safe to rely. The calculation depends on several assumptions, one or two of which, at least, are highly controvertible; for instance, the accuracy of the 5187 persons who asserted that they had not within a given period of twelve years had an exceptionally vivid and distressing dream relating to the death of a friend; and the accuracy of the twenty-four persons who described themselves as having had within the same period a similar dream actually occurring within twelve hours of the death of the person represented. Probably the estimate given requires modification by large allowances being made in both directions for defects of memory. But even when thus discounted the coincidences will, it is thought, by any one who carefully studies the subject be found to be more numerous than can plausibly be attributed to chance.
[90] A man writing on Wednesday would almost certainly say "last night" if he meant to indicate the preceding night, whereas, having just before written of "Friday last," it was natural to describe the Tuesday in the previous week as simply "Tuesday."
[91] These letters are omitted for want of space. They are given in full in the Journal of the S.P.R. for April 1888, pp. 255, 256.
[92] Proc. S.P.R., vol. v. pp. 453-455.