But however carefully dreams are noted and described, the objection still holds good that with impressions of such frequent occurrence chance alone will account for a considerable number of coincidences. It is easy, however, on a superficial view to exaggerate the probabilities of chance coincidence. The great majority of dreams, vague at the time and fugitive in the retrospect, are like footsteps in the sand. Yet as, here and there, one set of footprints out of the millions impressed upon the shore of a long-forgotten sea has been preserved for us in sand now turned to stone, so now and then one dream stands out from all the rest, and leaves on the memory an imprint which the daily reflux of the tide of consciousness cannot efface. If we strike out of the account all the dreams which are too vague to leave any permanent impress on waking, all those which are purely inconsequent and fantastic, and all which can be readily traced to some physical cause, we shall find that the number which we have to deal with,--the number, that is, of vivid and passably realistic dreams,--though no doubt large, is perhaps not beyond the range of definite calculation. It could not, for instance, be plausibly contended that the correspondence of a dream such as that of Captain Campbell's, recorded below, with the death of the person portrayed, is on the same level as the prophetic vision of the City clerk, who, dreaming every other night of the success of some horse which he has backed, happens on some one occasion to dream of the future winner. It will be observed that of the nine dreams which are given in full in this chapter, no less than four are concerned with death. Of the much larger number--149--of coincident dreams published in _Phantasms of the Living_, no less than 79 relate to a death. Now, as dreams of death or suggesting death do not form a large proportion of dreams in general, their startling preponderance amongst coincident dreams constitutes in itself an argument for ascribing such dreams to telepathy; for if any power exists whereby one mind can affect another, it would appear _à priori_ probable that such a power would be exercised most frequently and effectually at times of exceptional crisis. As has been pointed out by Mr. Gurney (_Phantasms of the Living_, vol. i. p. 303), the preponderance amongst "true" dreams of dreams relating to death may indeed be explained on the assumption that such dreams are more frequently remembered than other "true" dreams. This assumption is no doubt in a measure justified, but the consequences of admitting its truth must not be overlooked; for it of course follows that a large number of coincident dreams are forgotten, _i.e._, that the grounds furnished by dreams for believing in telepathy are much stronger than would at first sight appear.