§ 13. But, on the whole, the danger that the closeness of the coincidence may be exaggerated depends rather on misstatement of the date of the percipient's than of the agent's share in the alleged occurrence. Clearly the fact that some one has died or has had a serious accident, or has been placed in circumstances of some unusual sort, is likely to be known to more persons, and to be more frequently recorded in some permanent form, than the fact that some one has had, or says he has had, an odd hallucination. And clearly also, if one of the points is fixed, and the other, by hasty assumption or defective memory, is moved up to it, the movable date is likely to be that of the event which has no ascertainable place in the world of objective fact. As a rule, it is at any rate possible at the time to obtain certainty as to the date of what has befallen the agent; and therefore if the percipient has been struck by his experience and retains evidence of its date, either in writing or in the memory of others to whom he mentioned it, he will very likely be prompted, when he hears of the other event, to assure himself as to what the degree of coincidence really was. But the converse case is very different.

If the percipient does not record his experience at the time of its occurrence, even a week's interval may destroy the possibility of making sure what its exact date was; and therefore, however certain the date of the other event may be, assurance as to the degree of coincidence will here be unattainable. It is often expressly recognised as such by the percipient himself; and then one can only regret that the importance of the class of facts - if facts indeed they are - has been so little realised that the simple measures which would have ensured accurate evidence have not been taken. But where the account given is one of accurate coincidence, we cannot be satisfied without good evidence that the point was critically examined into at the time. It may, of course, happen that the percipient has a clear recollection that the coincidence was adequately made out at the time, although he can produce no documentary evidence which would establish i t; and if others confirm his memory in this respect, that is so far satisfactory.

Such unwritten confirmation, however, will have little independent force, unless the person who gives it was made aware of the percipient's experience within a very short time of its occurrence.

But though the danger here must be explicitly recognised, it is important not to exaggerate its practical scope. The coincidence may have been reported as closer than it was; but it may still, in a majority of cases, be fairly concluded to have fallen within the 12 hours' limit. As a rule, the news of what has befallen the agent arrives soon enough for not more than a space of two days to intervene between the percipient's knowledge of this event and the time when, to make the coincidence complete, his own experience must have taken place. We are not, therefore, making a large demand on his memory; we are only requiring that he shall remember that an experience, which he represents as remarkable, befell him, or did not befall him, on the day before yesterday. No doubt, after a lapse of years, the evidential value of what a person reports ceases to have a close relation to the knowledge of the facts which it seems pretty certain he must have had at the time. But the demands made at the time on the intelligence either of the percipient, or of anyone else who had the opportunity of asking questions and forming conclusions, are so slight that we may fairly take contemporary written records of the matter, or even later verbal corroboration, as having a considerable claim to attention, even when the best evidence of all - evidence whose existence preceded the arrival of the news - is wanting.

And it is important to notice that, while we have had several coincidences reported to us as having been close to the hour, which turned out, on further inquiry or examination of documents, to have been only close to the day, we have had few cases where a similar correction has proved that the 12 hours' limit was really overpassed.1 A good many coincidences, no doubt, have been represented as extremely close, where no independent evidence on this point has been accessible, and closer inquiry has occasionally revealed that the assertion rested only on a guess. But wholly to neglect cases where the exactitude of the coincidence is not brought within the 12 hours' limit would clearly be unreasonable, provided that - on the evidence - it is not likely that this limit was much exceeded, and not certain that it was exceeded at all. Such cases must, of course, be excluded from any numerical estimate based on precise data; but they may fairly be allowed their own weight on the mind.