510 



SCIENCE. 



Vol. IV., No. 96. 



he means to include in this heterogeneous group 

 visual hallucinations of waking persons, which we 

 regard as by far the most important phenomena from 

 an evidential point of view. If any one, in his 

 waking moments, experiences apparitions of human 

 forms as often as once a week, which is the degree 

 of frequency that Professor Newcomb's calculation 

 assumes,' it is obvious that the approximate coinci- 

 dence of one of these apparitions with the death of 

 the corresponding human being will be an insignifi- 

 cant accident. But we have not ourselves met with 

 any specimen of this class. We have collected more 

 than a hundred first-hand cases of apparitions closely 

 coinciding with the time of death of the person seen ; 

 and it is only in a small minority of such cases that 

 our informants, according to their own account, have 

 had any other hallucination than the apparition 'in 

 question. 



The following sketch may serve to show the lines 

 on which our own reasoning in the matter will pro- 

 ceed. We are making a census, which, so far, shows 

 that in this country the proportion of sane persons, 

 in good health and awake, who within the last ten 

 years have had a visual hallucination representing 

 some living person known to them, is about one in 

 three hundred. Now, let us make a supposition far 

 below the actual mark, and confine the number of 

 the acquaintances of each of these hallucinated per- 

 sons to five. Let us further suppose that one of these 

 five persons does actually die in the course of the ten 

 years. This seems fair, on the whole ; for, though in 

 some cases more than one may die within that time, 

 in others none may die. According to this estimate, 

 then, the chance that the death will take place within 

 twelve hours of the apparition will be one in 365 X 2 

 X 10 X 5; that is, one in 36,500: in other words, only 

 one out of every 36,500 of the hallucinated persons 

 will, in the course of ten years, hit off the coinci- 

 dence by chance. But since the hallucinated persons 

 are only a three-hundredth of the whole population, 

 this means that the proportion of the whole popu- 

 lation who will by chance have an apparition of a 

 person known to them within twelve hours of that 

 person's death is only one in 10,950,000. Now, we 

 ourselves have a large collection of such recent cases, 

 resting on good first-hand testimony; but let us put 

 the number far below the mark, and say thirty cases. 

 If, then, these thirty coincidences are to be fairly at- 

 tributed to chance, the population of the country 

 will have to be 328,500,000. But we cannot suppose 

 that our appeal for evidence has reached the whole 

 population; and we shall be making a sober estimate, 

 if we reckon that within the given time ten times as 

 many cases must have occurred as those we happen 

 to have encountered. This brings the necessary 

 population up to 3,285,000,000; and the number will 

 be further immensely increased if we take count of 

 the fact that many of the coincidences are extremely 

 close, that the times of the two events fall not only 

 within twelve hours, but within one. Thus the 

 theory that chance would account for the cases could 

 only be justified if the population of the country were 

 several hundred times what it actually is. The re- 

 ductio ad absurdum seems tolerably complete. 



The case of dreams is of course very different. 

 We are most of us constantly dreaming. A very large 

 number of 'odd coincidences' between dreams and 

 external events is certain to occur by mere chance, 

 and the cases are rare where the correspondence is of 

 a kind which strongly suggests telepathic influences. 

 Here, therefore, Professor Newcomb's estimate is far 

 more applicable ; and we have always felt that dreams, 

 by themselves, could not be expected to afford conclu- 



sive proof of telepathy. This, however, does not seem 

 a sufficient reason for ignoring them ; since, if the fact 

 of telepathic communication be otherwise established, 

 they may throw light which we could ill afford to 

 neglect, on the nature of the mental and cerebral 

 processes involved. 



As regards 'haunted houses,' we readily admit, and 

 have expressly pointed out, the far greater uncertainty 

 of the evidence as compared with the best telepathic 

 cases. But even here we differ from Professor New- 

 comb in seeing a distinction between the experiences 

 which we deem of some prima facie importance, and 

 the experience which he supposes when a person, 

 lying awake an hour after midnight, hears some 

 sound the cause of which is beyond his power to 

 guess. Sounds are the very weakest sort of evidence. 

 What strength the prima facie case has, depends, not 

 on things heard, but on things seen ; and seen, not by 

 one person only, but by several independently and at 

 different times, and, as the seers affirm, without any 

 knowledge, on their part, that the house was supposed 

 to be ' haunted.' 



Professor Newcomb's concluding remarks, dealing 

 with the experimental side of telepathy, deserve care- 

 ful attention. But his objections here rest entirely 

 on the hypothesis of visual and auditory indications 

 consciously or unconsciously given by the ' agent ' 

 to the ' percipient; ' and though it is difficult, I know, 

 to convince persons who have not been present that 

 sufficient precautions have been taken to eliminate 

 this source of error, it must surely be admitted that 

 such precautions are possible. As regards sight, no 

 one will deny the possibility; and, as regards hear- 

 ing, we think, that, if a careful watch is kept, the 

 means of communication resolve themselves into 

 slight variations of breathing. Such variations were 

 never detected in our experiments, and in any case 

 could hardly be supposed capable of rapidly convey- 

 ing to the percipient's mind the form of an irregular 

 diagram; and the difficulty would be increased in 

 cases where the signs would have had to be uncon- 

 scious, as in many of our experiments where we were 

 able not only to vary the ' agent,' but to act our- 

 selves as 'agents.' As for ' indications whether the 

 subject is going right or wrong,' they must, of course, 

 be prevented by taking care that the 'agent' shall 

 not watch what the 'percipient' is doing. Most of 

 the spurious ' thought-reading ' of the ' willing-game ' 

 would be prevented, if the 'w 7 iller,' instead of the 

 ' willed,' were effectively blindfolded. 



But we find ourselves once more wholly in sympa- 

 thy with Professor Newcomb, when he insists that 

 the experiments must be repeated again and again, 

 under the strictest conditions, before we can reason- 

 ably expect thought-transference to be accepted as 

 an established scientific fact. So far from resenting 

 the demand for more evidence, we are ourselves un- 

 ceasingly reiterating it. The responsibility for such 

 novel observations cannot be too widely spread, and 

 glad indeed shall we be to shift some of it to Ameri- 

 can shoulders. Edmund Gtjrney, 



Hon. sec. of the Society for psychical research, 



14 Dean's yard, Westminster, S.W., 

 Nov. 4. 



Mr. Gurney's letter suggests many interesting 

 reflections on the probabilities involved in questions 

 of telepathic phenomena, and I hope for an early 

 opportunity to engage in a further discussion of the 

 subject in the columns of Science. This will natu- 

 rally involve the consideration of the points raised in 

 his letter. Meanwhile there are two numerical data: 



