Let us now press the actual evidential question somewhat closer. Let us consider, for it is by no means evident at first sight, what conditions a visual or auditory phantasm is bound to fulfil before it can be regarded as indicating _primâ facie_ the influence of a discarnate mind. The discussion may be best introduced by quoting the words in which Edmund Gurney opened it in 1888.[132] The main evidential lines as there laid down retain their validity, although the years which have since passed have greatly augmented the testimony, and in so doing have illustrated yet other tests of true post-mortem communication,--to which we shall presently come. "It is evident that in alleged cases of apparitions of the dead, the point which we have held to distinguish certain apparitions of _living_ persons from purely subjective hallucinations is necessarily lacking. That point is _coincidence_ between the apparition and some critical or exceptional condition of the person who seems to appear; but with regard to the dead, we have no independent knowledge of their condition, and therefore never have the opportunity of observing any such coincidences. "There remain three, and I think only three, conditions which might establish a presumption that an apparition or other immediate manifestation of a dead person is something more than a mere subjective hallucination of the percipient's senses. Either (1) more persons than one might be independently affected by the phenomenon; or (2) the phantasm might convey information, afterwards discovered to be true, of something which the percipient had never known; or (3) the appearance might be that of a person whom the percipient himself had never seen, and of whose aspect he was ignorant, and yet his description of it might be sufficiently definite for identification. But though one or more of these conditions would have to be fully satisfied before we could be convinced that any particular apparition of the dead had some cause external to the percipient's own mind, there is one more general characteristic of the class which is sufficiently suggestive of such a cause to be worth considering. I mean the disproportionate number of cases which occur _shortly after_ the death of the person represented. Such a time-relation, if frequently enough encountered, might enable us to argue for the objective origin of the phenomenon in a manner analogous to that which leads us to conclude that many phantasms of the living have an objective (a telepathic) origin. For, according to the doctrines of probabilities, a hallucination representing a known person would not _by chance_ present a definite time-relation to a special cognate event--viz., the death of that person--in more than a certain percentage of the whole number of similar hallucinations