118 EDMUND GURNET. be taken necessarily to imply memory ; but experiences are recorded in which the scene or incident, though not one associated with the waking life, is distinctly recognised as familiar when it recurs, which of course does imply a sort of memory. Natural somnambulism may seem to present more distinct resemblances ; and certainly, if one judged from expressions used by the som- nambulist, particular ideas which have made no part of the waking life, are apt to recrudesce in the sleep- waking state. As a rule, however, it would be hard to represent this phenomenon as involving more than a mere reawakening into activity of certain nervous tracts, which naturally manifest the fact of their activity by the same external results as on previous occasions. Even in the rarer cases, which more strongly suggest the recurrence of the past events as such, the presence of true psychic memory is more doubtful than in hypnotism, or at any rate is harder to substantiate ; for the very tests which might substantiate it naturally tend to wake the somnambulist, and so to put an end to the condition. Perhaps, therefore, the clearest interest of the hypnotic alternations of memory is rather as illustrating the spontaneous alternations in cases of ' double consciousness,' where a single individual lives in turn two (or more) separate existences. There, as here, the transition may be almost instan- taneous; and there, as here, while the memory of the normal state is continuous (its events being remembered even in the abnormal condition, just as we have seen that the events of ordinary life are remembered in either of the hypnotic states), the memory of the past events of any abnormal state lapses and recurs with the disappearance and reappearance of that state. (3) If the phenomena mentioned under the last head are some- what uncertain, it is otherwise when the condition intervening between two hypnotic states of the sanir kind is not normal wake- fulness but an hypnotic state of the other kind i.e., when a deep state intervenes between two alert, or an alert between two deep states. I have then found that (with certain well-marked ex- ceptions to be mentioned hereafter) the ideas impressed in the one sort of state are invariably forgotten in the other, and are as invariably again remembered when the former state recurs. Thus the ' subject,' when in the alert state, is told something some anecdote or piece of ordinary information which we will call A. He is then thrown, or allowed to fall, into the deep state with closed eyes, and is asked ' What were you told just now ? ' He is quite unaware what is meant, nor will the broadest hints recall the missing idea. He is now told something else, which we will call B; and is then re-awakened into the alert state. Being asked the same question as before, he at once repeats not B but A, and it is impossible to evoke in him any memory of B. Thrown again into the deep state, he in a similar way recalls B, and A has once more vanished. Finally he is completely awakened, informed that two things have been told him within the last five
