But apart from any bias of an emotional or speculative sort, we must certainly admit a general tendency in the human mind to make any picture of facts definite. To many people vagueness of emotion or of speculation is a delight; but no one enjoys vagueness of memory. In thinking of an event which was in any way shadowy or uncertain, there is always a certain irksomeness in realising clearly how little clear it was. The same applies, of course, to events at which we look back through any considerable interval of time. The very effort to recall them implies an effort to represent them to the mind as precisely and completely as possible, and it is often not observed that the precision thus attained is not that of reality.

Lastly, there is a general tendency to lighten the burden of memory by simplifying its contents - by bringing any group of connected events into as round and portable a form as possible. This may, of course, only result in the loss of excrescences and subordinate features, while the essential incident is left intact. But we shall find instances further on-where simplification really alters the character of the evidence. Details may not simply drop out; they may undergo a change, and group themselves conveniently round some central idea. It might reasonably be expected, and we ourselves certainly began by expecting, that error from this source would always tell in the direction of actual distortion and exaggeration; if the aspect of the case was to some extent striking and insignificant to begin with, it would seem likely that this aspect of it should become more pronounced as it assumed a more isolated place in the mind, and lost its connection with the normal stream of experience in the course of which it appeared. As a matter of fact, however, this is by no means always what happens. For instance, we have met with several cases of the following sort.

An impression of a remarkable kind, and which, if telepathy exists, may fairly be regarded as telepathic, has been produced on a percipient while in a state which he recognised at the time as one of complete wakefulness, and which was practically proved to be so by the fact that he did not wake from it - that it formed a connected part of his waking life. But in the natural gravitation towards easy accounts of things, he gradually gets to look back on this experience as a dream; that is, he allows the verdict of subsequent memory to supplant the verdict of immediate consciousness. We must not then say in our haste, all men - or all memories - are exaggerators. Even where evidence has been modified in passing through several mouths, a comparison between later and earlier versions of the same occurrence has sometimes shown that its more striking and significant characteristics have lost rather than gained by the transmission. But this is no doubt the exception.

1To take a single instance - a lady sends us an unsensational narrative of the ordinary type, as to how one day in 1882, when just about to sit down to the piano, she saw close to her the figure of an old school-friend, who, as it turned out, died on that day at a distance. "I am confident," she says, " of having seen the vision, though my common sense makes mo wish to put it down to imagination. I never saw any vision of any kind before or since." "But we are withheld from quoting the account in a form which could have any evidential value, by her feeling that such publication would be wrong.

§ 7. Such, in brief outline, are the principal sources of error which may in a general way be supposed to affect the sort of evidence with which we are concerned; and our next step must be to fix with precision what the actual opportunities for perversion are. The evidence for telepathy has a certain type and structure of its own, and we must realise what this is, in order to know where to look for the weak points. What, then, are the essential elements of a typical telepathic phenomenon? They consist in two events or two states, of a more or less remarkable kind, and connected, as a rule, by certain common characteristics; and of a certain time-relation between the two. For example, if a flawless case is to be presented, it would be of the following type and composition: It would comprise (1) indisputable evidence that A (whom we call the agent) has had an unusual experience - say, has died; (2) indisputable evidence that B (whom we call the percipient) has had an unusual experience which includes a certain impression of A - say, has, while wide awake, had a vision of A in the room; (3) indisputable evidence that the two events coincided in time - which, of course, implies that their respective dates can be accurately fixed.

When I call such evidence as this flawless, I do not, of course, mean that it is conclusive: the fact that the two events occurred, and the fact that they occurred simultaneously, might be placed beyond dispute, and the coincidence might, for all that, be due not to telepathy, but to chance alone. But though no single case can prove telepathy, no case where the above conditions are not to some extent realised can even help to prove it. Briefly, then, if the account of some alleged instance of telepathy is evidentially faulty, there must be misrepresentation as to one or more of the following items: (1) the state of the agent; (2) the experience of the percipient; (3) the time of (1); (4) the time of (2).

Now the evidence where the chances of misrepresentation have primarily to be considered is clearly that of the percipient. It is the percipient's mention of his own experience which makes, so to speak, the groundwork of the case; unless the percipient gives his own account of this experience, the case is in no sense a first-hand one; whereas if such an account is given we should consider the evidence first-hand, even though the account of the agent's state is not obtained from himself. Of course when the agent is in a position to give an account, it is important that his evidence should be procured; but this is impossible in the numerous cases where his share in the matter consists simply in dying. In these cases, then, we are dependent on others for evidence as to the agent's side of the occurrence; and primarily often on the percipient, who is our first and indispensable witness for the whole matter. This being premised, we shall have no difficulty in discovering where the risks of misrepresentation really lie.