§ 1. Assuming the substantial correctness of much of the evidence for phantasms which have markedly coincided with an event at a distance, how can it be known that these coincidences are not due to chance alone? In examining this question, we must be careful to distinguish waking cases from dreams - in which latter class (as we have seen) the scope for chance-coincidences is indefinitely large .......... 372-374

§ 2. The answer to this question depends on two points - the frequency of phantasms which have markedly coincided with real events, and the frequency of phantasms which have not. If the latter class turned out to be extremely large - e.g., if we each of us once a week saw some friends' figure in a place which was really empty - it is certain that occasionally such a subjective delusion would fall on the day that the friend happened to die. The matter is one on which there have been many guesses, and many assertions, but hitherto no statistics 374-376

§ 3. To ascertain what proportion of the population have had experience of purely subjective hallucinations, a definite question must be asked of a group large and varied enough to serve as a fair sample of the whole. The difficulty of taking such a census has been much increased by a wide misunderstanding of its purpose ........... 376-377

§ 4. But answers have been received from a specimen group of 5705 persons; and there is every reason to suppose this number sufficient . . 377_379

§ 5. It may be objected that persons may have wrongly denied such experiences (1) through forgetfulness - but the experiences of real importance for the end in view are too striking to be readily forgotten; (2) by way of a joke or a hoax - but this would lead rather to false confessions than false denials; (3) in self-defence - but such error as may have been produced by this motive has probably been more than counter-balanced in other ways . , . 379-381

§ 6. As to visual hallucinations, representing a recognised face or form - in the last 12 years such an experience has, according to the census, befallen 1 adult in every 248; but it would have had to befall every adult once, and some adults twice, to justify the assumption that the cases recorded in the present work on first-hand testimony, of the coincidence of the experience in question with the death of the person represented, were due to chance. The odds against the accidental occurrence of the said coincidences are counted in trillions . 381-383

§ 7. The extreme closeness of some of the coincidences affords the basis for another form of estimate, which shows the improbability of their accidental occurrence to be almost immeasurably great . . . . . 384-386

And a number of further cases and further considerations remain, by which even this huge total of improbability would be again swelled. The conclusion, therefore, after all allowances, that at any rate a large number of the coincidences here adduced have had some other cause than chance seems irresistible . 386

§ 8. An argument of a quite different sort may be drawn from certain peculiarities which the group of coincidental hallucinations present, when compared, as a whole, with the general mass of transient hallucinations of the sane 386-389