It must be remembered, in the first place, that all these veridical or coincidental cases stand out together as a single group from a background of hallucinations which involve no coincidence, which have no claim to veridicality. If purely subjective hallucinations of the senses affected insane or disordered brains alone,--as was pretty generally the assumption, even in scientific circles, when our inquiry began,--our task would have been much easier than it is. But while there can be no question as to the sound and healthy condition of the great majority of our percipients, Edmund Gurney's "Census of Hallucinations" of 1884, confirmed and extended by the wider inquiry of 1889-1892, showed a frequency, previously unsuspected, of scattered hallucinations among sane and healthy persons, the experience being often unique in a lifetime, and in no apparent connection with any other circumstance whatever.[109] Since casual hallucinations of the sane, then, are thus _frequent_, we can hardly venture to assume that they are all _veridical_. And the existence of all these perhaps merely subjective hallucinations greatly complicates our investigation of veridical hallucinations. It prevents the mere existence of the hallucinations, however strangely interposed in ordinary life, from having any evidential value, and throws us upon evidence afforded by external coincidence;--on the mere fact, to put such a coincidence in its simplest form, that I see a phantom of my friend Smith at the moment when Smith is unexpectedly dying at a distance. A coincidence of this general type, if it occurs, need not be difficult to substantiate, and we have in fact substantiated it with more or less completeness in several hundred cases. The _primâ facie_ conclusion will obviously be that there is a causal connection between the death and the apparition. To overcome this presumption it would be necessary either to impugn the accuracy of the informant's testimony, or to show that chance alone might have brought about the observed coincidences. On both of these questions there have been full and repeated discussions elsewhere. I need not re-argue them at length here, but will refer the reader to the "Report on the Census of Hallucinations," _Proceedings_ S.P.R., vol. x., where every source of error as yet discovered has been pretty fully considered. To that volume also I must refer him for a thorough discussion of the arguments for and against chance-coincidence. The conclusion to which the Committee unanimously came is expressed in the closing words: "Between deaths and apparitions of the dying person a connection exists which is not due to chance alone." We have a right, I think, to say that only by another census of hallucinations, equally careful, more extensive, and yielding absolutely different results, could this conclusion be overthrown.