This section is from the book "Phantasms Of The Living", by Edmund Gurney, Frederic W. H. Myers, Frank Podmore. Also available from Amazon: Phantasms of the Living.
The estimate from the above data is as follows. The probability that a person, taken at random, will have a vivid dream of death in the course of 12 years is 1/26; the probability that any assigned member of the general population, and therefore that any particular dreamt-of person, will die within 12 hours of an assigned point of time is 22/1000 x 1/365; hence the probability that, in the course of 12 years, a vivid dream of death, dreamt by any previously specified individual, and the death of the person dreamt of will fall within 12 hours of one another is 1/26 x 22/1000 x1/365 = 1/431.363. (If x, the number of the dreamer's circle of friends, be taken into account, we then have, as the probability that any one of the dreamer's circle should die within the particular 24 hours defined by the dream, x X 1/26 x 22/1000 x 1/365;and as the probability that, if some one of the dreamer's circle dies within the particular 24 hours, it should be the particular one dreamt of, 1/x; whence the probability of this double event = 1/x X x X 1/26 x 22/1000 x 1/365, i.e the result is unaltered.) That is to say, each group of 431,363 persons in the population of the United Kingdom will on an average, if chance alone rules, produce one such coincidence in the given time.
Now let it be supposed for a moment that our appeal for evidence has effectively reached as large a section of the population as this: let it be supposed, that is to say, that the number of persons from whom we should, directly or indirectly, have received examples of such coincidences, if they had had them to communicate, amount to 431,363. In that case, then, the number of such coincidences which we ought, by the doctrine of chances, to have encountered is 1. The number which we have actually encountered, of vivid dreams of death, narrated to us at first-hand,1 dreamt since January 1st, 1874, by persons free from anxiety, and falling within 12 hours of the death of the person dreamt of2 - is 24: that is, a number 24 times larger than the doctrine of chances would have allowed us to expect.3 And this number is very much below what we are justified in assuming. For while my colleagues and I are probably disposed rather to overrate than to underrate the extent to which the world is acquainted with our proceedings, we cannot suppose that the actual number of persons from whose collective experiences our examples are drawn really approaches half a million, as above supposed.
Moreover, 7 of the 24 coincidences are represented as having been extremely close; in 3 other cases the interval was at any rate not more than 4 hours, and in another was from 3 to 6 hours; while in 9 more, where death and dream are merely stated to have fallen on the same night, the coincidence may have been exact, and is not likely to have been inexact to the extent of anything like 12 hours. Again, dreams are excluded where the actual fact of death was not in some way presented, even though the dying person was dreamt of in a remarkable way.4 But most markedly have I understated the case in this respect, that I have introduced nothing but the bare fact of death, and have neglected the points of detail which in some instances add indefinitely to the difficulty of regarding the coincidence as a chance.
The above is a tolerably clear computation; and its validity could only be rebutted in two ways - (1) by impugning, on evidential grounds, the cases of coincidence that are alleged; (2) by impugning the approximate accuracy of the initial datum - that within the last 12 years not more than 1 person in 26 has, without clear cause, had a markedly distressing and haunting dream of the death of an acquaintance.
1 In three cases the evidence, though not actually from the dreamer, is of the sort described on p. 119, as on a par with first-hand.
2 With 5 possible exceptions, where the 12 hours' limit may not have been nearly reached, but where it may have been exceeded. In 2 of these cases, the death took place in the afternoon and the dream followed the same night; in the 3 others, the death was either on the night of the dream or on a day contiguous to that night but its hour is not known. Two dreams are excluded which arc known to have followed the death by a very little more than 12 hours, the death not having been heard of in the interval.
3 Of the 24 included dreams, 10 were of the deaths of near relatives; and 4 or 5 concerned persons who were outside the circle of intimate acquaintance.
4 I would draw special attention to this point. For when we come to deal with the waking cases, where a phantasm of a person is seen or heard at the time of his death, they may seem to present a marked difference from the dreams that will be quoted as having coincided with death, - these last being distinctly dreams of death, whereas it is only in a minority of the waking cases that any idea of death was conveyed. The waking percipient may have surmised the death, from the fact that he had seen or heard the phantasm; but the phantasm itself, more often than not, was simply the natural-looking appearance or natural-sounding voice of the "agent." We must remember, then, that for aught we know, telepathically caused dreams of just this type may occur; but unless they present specially remarkable features (as ill case 23, p. 142 above) we should not cite them as having even a prima facie claim to be considered telepathic, just because of the immense scope for chance-coiucidence that dreams afford.
We demand more of a dream - that it shall suggest the right event, and not merely the right person - before we think it worth considering; and the dreams to be quoted correspond with the rarer typo of waking eases, where the phantasmal representation has in some way distinctly suggested death.
The evidential value of the alleged coincidences will be better estimated when we consider some of the actual specimens. But as regards the initial datum, on which the calculation depends, there are objections the force of which must be at once admitted. Dreams in general, it may be said, fade away from our memory because there is nothing to stamp them there; but if it happens that some real event recalls a recent dream, then, by the principle of association, this dream will obtain a more permanent lodgment in our mind. Now the death of relatives or friends is the sort of real event which it is practically certain that we shall hear of very soon after it occurs. A dream of such an occurrence is therefore practically certain to get stamped in the way described, if it has been at all synchronous with the fact. And it is thus allowable to suppose that a large number of such dreams may occur which lapse unnoted from the mind, but any one of which, had news of the real event been received immediately afterwards, would have been recalled and associated with it, and would have then added a case to the list of "remarkable coincidences".
This argument is to a considerable degree met by the terms of the question. What is asked is not merely whether people have had a dream of death; but whether they have had one which has haunted them for at least an hour after rising in the morning; and it will not be maintained that an experience of that sort is so likely to vanish utterly from the memory as an ordinary dream. But it might, no doubt, be rejoined that perhaps a good many of the coincident dreams were not marked at the time by any special vividness or impressiveness; and that the dreamer came to imagine this peculiarity of character in his dream, after it had come to assume importance in his eyes from the discovery that it had coincided with the reality. And I most fully recognise that when the argument begins to turn on such a point as the degree of vividness which a dreamer, looking back to a dream through the medium of subsequent impressions, can swear that it possessed, anything like positive proof becomes impossible. A dream so looked back to may get charged with an emotional character, just as we saw above that it might get rilled in with a* precision of detail, which it did not really possess. But I must here draw special attention to the safeguards already mentioned.
Our collection includes a remarkably high proportion of cases where the coincident dream was marked as exceptional in character - at the time, and before the real event was known - by being immediately narrated as such to someone else (who, if accessible, has of course been questioned as to his memory of the fact); or by being noted in writing; or by being in some way acted on. [Figures are here given which I omit, as I have omitted many dream cases. - Ed.] And I must further point out that, in order to explain away the result of the above computation in the way suggested, it would have to be assumed not only that a great many dreams of death pass unnoted and leave no impression, while still of such a nature that a vivid impression of them would revive if news corresponding to them were subsequently heard; it would have to be assumed that such an experience befalls very nearly every adult in the country at least once within the 12 years. For our conclusion was that coincident dreams of death in this country were 24 times as numerous as the law of chances - according to the datum which the census gave us - would allow.
Therefore to make the law of chances applicable as an explanation, we must multiply our initial datum by 24; that is, instead of assuming 1 person out of every 26 to have had the required dream, we must assume 24 out of every 26 - that is nearly every one - to have had it; nay, on the more probable estimate of our area of inquiry (p. 213), we must assume that on the average every one has had it as many as four times within the given period; though 96 per cent. of them forget all about it.l A good many of my readers, will, I think, altogether repudiate such a supposition in their own case. I believe, indeed, that a perfectly impartial census would have given a decidedly smaller proportion than 1 in 26. For it is practically impossible to carry out a census of the sort required without getting an unfair proportion of Yeses, Persons who have only No to say in answer to such a question as was propounded, are apt to think that there is no good in saying it; and if they receive a printed form, instead of writing their answer on it and returning it, they are apt to consign it to the waste-paper basket - probably often with a vague notion that what was wanted was a Yes, and that sensible people, who do not have exceptional experiences, ought not to encourage the superstitions of those who do.2
 
Continue to: