§ 2. But though dreams thus present a logical point of departure, they also form in many ways the most assailable part of our case. They are placed almost in a separate category by their intimate connection with the lowest physical, as well as the highest psychical, operations. The grotesque medley which constantly throng through the gate of ivory thrust into discredit our rarer visitants through the gate of horn. And before proceeding further, it will be well to examine with some care the general evidential value of dreams, in relation to a theory of transferred impressions. The field may seem a fair one enough, as long as we keep to general expectations; if telepathy is a reality, here is a probable scene of telepathic events. But what opportunities does it afford for confirming these expectations by accurate and convincing evidence? This is a question which may rapidly convert our hopes into doubts.

The first objection to dreams, as evidence for transferred impressions of distant conditions or events, is this - that dreams being often somewhat dim and shapeless things, subsequent knowledge of the conditions or events may easily have the effect of giving body and definiteness to the recollection of a dream. When the actual facts are learnt, a faint amount of resemblance may often suggest a past dream, and set the mind on the track of trying accurately to recall it. This very act involves a search for details, for something tangible and distinct; and the real features and definite incidents which are now present in the mind, in close association with some general scene or fact which actually figured in the dream, will be apt to be unconsciously read back into the dream. They make part of the original, of which the mind conceives the dream to have been a picture; and the picture, when evoked in memory, will only too probably include details drawn from the original. After we have once realised the matter in its full distinctness, it becomes almost impossible to recall with due indistinctness the distant and shadowy suggestion of it.

Dreams in this way resemble objects seen in the dusk; which begin by puzzling the eye, but which, when once we know or think we know what they are, seem quite unmistakable and even full of familiar detail. For our purposes, therefore, it is of prime importance that the dream shall be told in detail to some one on whose memory we can rely; or, better still, written down, or in some way acted on, at the time, and before the confirmation arrives. Nearly all the evidence to be brought forward has, at any rate, this mark of accuracy.

1 This term implies that sensation, from a physical point of view, is inverted; that the initial stimulation takes place in the higher tracts of the brain, and that the stimulation of the special sensory apparatus is produced by a downward centrifugal current. The point will bo further noticed in connection with waking hallucinations (see pp. 311-12).

But there is a more general and sweeping objection. Millions of people are dreaming every night; and in dreams, if anywhere, the range of possibilities seems infinite; can any positive conclusion be drawn from such a chaos of meaningless and fragmentary impressions? Must not we admit the force of the obvious a priori argument, that among the countless multitude of dreams, one here and there is likely to correspond in time with an actual occurrence resembling the one dreamed of; and that when a dream thus "comes true," unscientific minds are sure to note and store up the fact as something extraordinary, without taking the trouble to reflect whether such incidents occur oftener than pure chance would allow? Can the chances be at all estimated? Are any valid means at hand for distinguishing between a transferred impression and a lucky coincidence? What degree of exactitude of date and circumstance must be reached, before we consider even a striking correspondence as worth attention? And what proportion of striking correspondences are we to demand, before we consider that the hypothesis of chance is strained in accounting for them?

In the first place, it is to be noted that there has, so far, been a complete lack of the statistics which alone could form the basis for an answer to these questions. It has never been known with any certainty what proportion of people habitually dream, what proportion of dreams are remembered at all, in what proportion of these remembered dreams the memory is evanescent, and in what proportion it is profound and durable. This latter point may be specially hard to establish satisfactorily in a particular case, as it is affected by the question whether a person's attention is habitually directed to his dreams, and also by the question whether he has happened to recount a particular dream to others, and so to stamp it on his own memory. By making inquiries on a large scale, however, a considerable approximation to certainty may be attained on these and various other points of importance. A good deal has been done in this way during the last three years; and though I cannot say that the results are such as would allow us to base a theory of telepathy upon the facts of dreams alone, I think that they do much to diminish the a priori plausibility of the theory of chance, as a sufficient explanation of all cases of marked correspondence between a dream and an external event: