My own ignorance, at any rate, I recognise to be such that my notions of the probable or improbable in the Universe are not of weight enough to lead me to set aside any facts which seem to me well attested, and which are not shown by experts actually to conflict with any better-established facts or generalisations. Wide though the range of established science may be, it represents, as its most far-sighted prophets are the first to admit, a narrow glance only into the unknown and infinite realm of law. The evidence, then, leading me thus unresisting along, has led me to this main difference from our early treatment of veridical phantasms. Instead of starting from a root-conception of a telepathic impulse merely passing from mind to mind, I now start from a root-conception of the dissociability of the self, of the possibility that different fractions of the personality can act so far independently of each other that the one is not conscious of the other's action. Naturally the two conceptions coincide over much of the ground. Where experimental thought-transference is concerned--even where the commoner types of coincidental phantasms are concerned--the second formula seems a needless and unprovable variation on the first. But as soon as we get among the difficult types--reciprocal cases, clairvoyant cases, collective cases, above all, manifestations of the dead--we find that the conception of a telepathic impulse as a message despatched and then left alone, as it were, to effect its purpose needs more and more of straining, of manipulation, to fit it to the evidence. On the other hand, it is just in those difficult regions that the analogies of other splits of personality recur, and that phantasmal or automatic behaviour recalls to us the behaviour of segments of personality detached from primary personality, but operating through the organism which is common to both. The innovation which we are here called upon to make is to suppose that segments of the personality can operate in apparent separation from the organism. Such a supposition, of course, could not have been started without proof of telepathy, and could with difficulty be sustained without proof of survival of death. But, given telepathy, we have _some_ psychical agency connected with man operating apart from his organism. Given survival, we have an element of his personality--to say the least of it--operating when his organism is destroyed. There is therefore no very great additional burden in supposing that an element of his personality may operate apart from his organism, while that organism still exists.