But men, we may assume, were not, at the assumed stage of thought, so besotted as not to take a great practical distinction between sleeping and waking experience on the whole.  As has been shown, the distinction is made by the lowest savages of our acquaintance.  One clear waking hallucination, on the other hand, of the presence of a person really absent, could not but tell more with the early philosopher than a score of dreams, for to be easily forgotten is of the essence of a dream.  Savages, indeed, oddly enough, have hit on our theory, ‘dreams go by contraries.’  Dr. Callaway illustrates this for the Zulus, and Mr. Scott for the Mang’anza.  Thus they do discriminate between sleeping and waking.  We must therefore examine waking hallucinations in the field of actual experience, and on such recent evidence as may be accessible.  If these hallucinations agree, in a certain ratio, beyond what fortuitous coincidence can explain, with real but unknown events, then such hallucinations would greatly strengthen, in the mind of an early thinker, the savage theory that a man at a distance may, voluntarily or involuntarily, project his spirit on a journey, and be seen where he is not present.

When Mr. Tylor wrote his book, the study of the occasional waking hallucinations of the sane and healthy was in its infancy.  Much, indeed, had been written about hallucinations, but these were mainly the chronic false perceptions of maniacs, of drunkards, and of persons in bad health such as Nicolai and Mrs. A. The hallucinations of persons of genius—­Jeanne d’Arc, Luther, Socrates, Pascal, were by some attributed to lunacy in these famous people.  Scarcely any writers before Mr. Galton had recognised the occurrence of hallucinations once in a life, perhaps, among healthy, sober, and mentally sound people.  If these were known to occur, they were dismissed as dreams of an unconscious sleep.  This is still practically the hypothesis of Dr. Parish, as we shall see later.  But in the last twenty years the infrequent hallucinations of the sane have been recognised by Mr. Galton, and discussed by Professor James, Mr. Gurney, Dr. Parish, and many other writers.

Two results have followed.  First, ‘ghosts’ are shown to be, when not illusions caused by mistaking one object for another, then hallucinations.  As these most frequently represent a living person who is not present, by parity of reason the appearance of a dead person is on the same level, is not a space-filling ‘ghost,’ but merely an hallucination.  Such an appearance can, prima facie, suggest no reasonable inference as to the continued existence of the dead.  On the other hand, the new studies have raised the perhaps insoluble question, ’Do not hallucinations of the sane, representing the living, coincide more frequently than mere luck can account for, with the death or other crisis of the person apparently seen?’ If this could be proved, then there would seem to be a causal nexus, a relation of cause