The error which lies at the opposite pole from Dr. Hibbert’s mistake in not collecting instances, is the error of collecting only affirmative instances.  We hear constantly about ’hallucinations of sight, sound, or touch, which suggest the presence of an absent person, and which occur simultaneously with some exceptional crisis in that person’s life, or, most frequently of all, with his death’. {192} Now Mr. Gurney himself was much too fair a reasoner to avoid the collection of instantiae contradictoraes, examples in which the hallucination occurs, but does not coincide with any crisis whatever in the life of the absent person who seems to be present.  Of these cases, Dr. Hibbert could find only one on record, in the Mercure Gallant, January, 1690.  The writer tells us how he dreamed that a dead relation of his came to his bedside, and announced that he must die that day.  Unlike Miss Lee, he went on living.  Yet the dream impressed him so much that he noted it down in writing as soon as he awoke.  Dr. Johnson also mentions an instantia contradictoria.  A friend of Boswell’s, near Kilmarnock, heard his brother’s voice call him by name:  now his brother was dead, or dying, in America.  Johnson capped this by his tale of having, when at Oxford, heard his name pronounced by his mother.  She was then at Lichfield, but nothing ensued.  In Dr. Hibbert’s opinion, this proves that coincidences, when they do occur, are purely matters of chance. {193a} There are many hallucinations, a death may correspond with one of them, that case is noted, the others are forgotten.  Yet the coincidences are so many, or so striking, that when a Maori woman has a hallucination representing her absent husband, she may marry without giving him recognised ground for resentment, if he happens to be alive.  This curious fact proves that the coincidence between death and hallucinatory presence has been marked enough to suggest a belief which can modify savage jealousy. {193b}

By comparing coincidental with non-coincidental hallucinations known to him, Mr. Gurney is said to have decided that the chances against a death coinciding with a hallucination, were forty to one,—­long odds. {194a} But it is clear that only a very large collection of facts would give us any materials for a decision.  Suppose that some 20,000 people answer such questions as:—­

1.  Have you ever had any hallucination?

2.  Was there any coincidence between the hallucination and facts at the time unknown to you?

The majority of sane people will be able to answer the first question in the negative.

Of those who answer both questions in the affirmative, several things are to be said.  First, we must allow for jokes, then for illusions of memory.  Corroborative contemporary evidence must be produced.  Again, of the 20,000, many are likely to be selected instances.  The inquirer is tempted to go to a person who, as he or she already knows, has a story to tell.  Again, the inquirers are likely to be persons who take an interest in the subject on the affirmative side, and their acquaintances may have been partly chosen because they were of the same intellectual complexion. {194b}