The movement required to make the raps may have become semi-automatic from long habit, but can hardly have been unconscious. I may add that, out of a good many words and sentences which were spelt out in the same way at several different sittings, the case recorded was (with a single doubtful exception) the only- one that contained the slightest indication of any abnormal faculty.

1 For a curious case of the automatic production of anagrams, see Proceedings of the S.P.R., vol. iii., pp. 226-31.

To return to M. Richet's experiments - a result of a different kind was the following, which is especially noteworthy as due to the agency of an idea that was itself on the verge of the unconscious. M. Richet chose a quotation at random from Littre's dictionary, and asked for the name of the author, which was Legouv6. The letters produced were JOSEPHCHD, which looked like a complete failure. But the quotation in the dictionary was adjacent to another from the works of Joseph Cheerier; and M. Richet's eye, in running over the page, had certainly encountered the latter name, which had probably retained a certain low place in his consciousness. Another very interesting case of a result unintended by the agent, though probably due to something in his mind, was this. The name thought of was Victor; the letters produced on three trials were - DALEN DAMES DANDS - seemingly complete failures. But it appeared that while the agent had been concentrating his thoughts on "Victor," the name of a friend, Danet, had spontaneously recurred to his memory.

We should, of course, be greatly extending the chances of accidental success, if we reckoned collocations of letters as successful on the ground of their resemblance to any one of the names or words which may have momentarily found their way into the agent's mind while the experiment was in progress. Here, however, the name seems to have suggested itself with considerable persistence, and the resemblance is very close. And if the result may fairly be attributed to "mental suggestion," then, of the two names which had a certain lodgment in the agent's mind, the one intended to be effective was ineffective, and vice-versa.

It is a remarkable fact that in the few hitherto recorded cases of experimental telepathy, where words have been indicated by writing or by other movements on the percipient's part, the idea or word transferred seems as often as not to have been one which was not at the moment occupying the agent's consciousness; that is to say, the influence has proceeded from some part of the agent's mind which is below the threshold of conscious attention. (See p. 67-8 below.) This conception of unconscious agency - of an "unconscious intelligence " in the agent as well as in the percipient - will present itself again very prominently when we come to consider the cases of spontaneous telepathy. But the experimental instances have a theoretic importance of another sort. They seem to exhibit telepathic production of movements by what is at most an idea, and not a volition, on the agent's part. This, indeed, is a hypothesis which seems justified even by M. Richet's less exceptional results. For we must remember that in a sense A is throughout more immediately the agent than F; it is what A's mind contributes, not what F's mind contributes, that produces the tilts at the right moments.1 But this is of course through no will of A's; he is ignorant of the required word, and has absolutely no opportunity of bringing his volition into play.

His "agency" is of a wholly passive sort; and his mind, as it follows the course of his pen, is a mere conduit-pipe, whereby knowledge of a certain kind obtains access to the "unconscious intelligence" which evokes the tilts. If, then, the knowledge manifests itself as impulse, can we avoid the conclusion that in this particular mode of access - in "mental suggestion" or telepathy as such - a certain impulsive quality is involved. We shall encounter further signs of such an impulsive quality among the spontaneous cases.2 (See pp. 204, 350-1).

But of course the relation between F and the "medium" plays also a necessary part in the result; the impulse to tilt when a particular letter is reached only takes effect when it falls (so to speak) on ground prepared by "mental suggestion" from F - on a mind in which the word imagined by him has obtained an unconscious lodgment. The unconscious part of the percipient's mind would thus be the scene of confluence of two separate telepathic streams, which proceed to combine there in an intelligent way - one proceeding from F's mind, which produces unconscious knowledge of the word, and the other proceeding from A's mind, which produces an unconscious image of the successive letters.l Another possible supposition would be that F's thought affects, not the ' medium," but A; or conversely, that A's thought affects not the " medium," but F; - that A obtains unconscious knowledge of the word, or that F obtains unconscious knowledge of the letter, and so is enabled to communicate an impulse to the "medium" at the right moment.

And we should then have to suppose a secret understanding between two parts of A's or F's mind, the part which takes account of the letters of the alphabet, and the part which takes account of the letters of the word - the former being conscious and the latter unconscious, or vice versa, according as A or F is the party affected.

1 When A, in pointing, began at the beginning of the alphabet, the sense of time might conceivably have led to an unconscious judgment as to the point arrived at. This idea had occurred to M. Richet. It seems, however, an unnecessary multiplication of hypothesis; for we learn from him that in some trials A began at uncertain places, and that under these conditions coherent words were obtained. The fact that so often the approximate letter was given, instead of the exact one, might seem at first sight to favour the hypothesis of unconscious reckoning; but it will be observed that exactly the same approximations took place in our own experiment (p. 62), where the alphabet was in the "medium's" sight.

2The impulse might no doubt be otherwise accounted for if we supposed that a close connection was established in F's mind between the idea of the object - i.e., the successive letters - and the idea of the movement, and that this complex idea was what was transferred and what ultimately took effect. But it is hard to apply this hypothesis to cases where a word is produced which, though latent in F's mind, has no resemblance to the word whose production he is willing. The transference of the idea of the latent word, even to the exclusion of the right word, can be quite conceived; but can we suppose that, subconsciously or unconsciously, an idea of movement was combined with the idea of its letters in the agent's mind, at the very moment when that on which his attention was fixed, and with which ex hypothesi the conscious idea of movement was connected, was a quite different set of letters? Can we suppose that the idea of movement overflowed into the unconscious region of his mind, and there on its own account formed an alliance with alien elements, the effect of which on the percipient would prevent the effect intended? It must be remembered that where a word which is not the one intended gets transferred from F to the "medium," there is no knowledge, conscious or unconscious, on F's part, as to what that word will be.