After these apologies for the limitations of this essay, we may survey the backward track. We began by showing that savages may stumble, and have stumbled, on theories not inconsistent with science, but not till recently discovered by science. The electric origin of the Aurora Borealis (whether absolutely certain or not) was an example; another was the efficacy of 'suggestion,' especially for curative purposes. It was, therefore, hinted that, if savages blundered (if you please) into a belief in God and the Soul, however obscurely envisaged, these beliefs were not therefore necessarily and essentially false. We then stated our purpose of examining the alleged supernormal phenomena, savage or civilised, which, on Mr. Tylor's hypothesis, help to originate the conception of 'spirits.' We defended the nature of our evidence, as before anthropologists, by showing that, for the savage belief in the supernormal phenomena, we have exactly the kind of evidence on which all anthropological science reposes. The relative weakness of that evidence, our need of more and better evidence, we would be the very last to deny, indeed it is part of our case. Our existing evidence will hardly support any theory of religion. Anyone who is in doubt on that head has only to read M. Réville's 'Les Religions des Peuples Non-Civilisés,' under the heads 'Mélanésiens,' 'Mincopies,' 'Les Australiens' (ii. 116-143), when he will observe that this eminent French authority is ignorant of the facts about these races here produced. In 1883 they had not come within his ken. Such minute and careful inquiries by men closely intimate with the peoples concerned, as Dr. Codrington's, Mr. Hewitt's, Mr. Man's, and the authorities compiled by Mr. Brough Smyth, were unfamiliar to M. Réville, Thus, in turn, new facts, or facts unknown to us, may upset my theory. This peril is of the essence of scientific theorising on the history of religion. Having thus justified our evidence for the savage _belief_ in supernormal phenomena, as before anthropologists, we turned to a court of psychologists in defence of our evidence for the _fact_ of exactly the same supernormal phenomena in civilised experience. We pointed out that for subjective psychological experiences, say of telepathy, we had precisely the same evidence as all non-experimental psychology must and does rest upon. Nay, we have even experimental evidence, in experiments in thought-transference. We have chiefly, however, statements of subjective experience. For the coincidence of such experience with unknown events we have such evidence as, in practical life, is admitted by courts of law.