_Journal_ of the Society for Psychical Research in this country.[71] He will then perhaps be prepared to endorse the verdict of a shrewd and genial critic on the evidence presented in _Phantasms of the Living_, viz., that it "can only be rejected as a whole by one who is prepared to repeat at his leisure what David is reported to have said in his haste."[72] It is of course not possible with our present knowledge to estimate with any precision the probabilities for the coincidence by chance of such a vision as that recorded by Dr. Dupré (No. 47), or such a dream as Mr. Hamilton's (No. 58), with the event represented. Neither the nature of the percipient's impression in these and similar cases, nor the event to which the impression corresponds, are sufficiently well defined to admit of any numerical argument being based upon them. We can only recognise that whilst dreams and mind's-eye pictures are not very uncommon experiences, dreams and visions which faithfully reflect external events of an unlikely kind occur, if rarely, with sufficient frequency to give us pause. The common sense which in such cases leads us to infer a connection between the event and the corresponding mental experience is our only guide. But one large class of our spontaneous evidences is susceptible of more exact treatment. Sensory hallucinations are affections at once well marked and unusual. If we can ascertain their relative frequency it is possible to calculate with more or less exactness the probabilities of the coincidence by chance with some definite event. Such a calculation has been attempted in Chapter IX. with regard to hallucinations of a certain well-defined type coinciding with the death of the person represented. The conclusion there reached is that such coincidences are far too numerous to be ascribed to chance. This part of the evidence cannot therefore be summarily dismissed, as suggested by more than one recent critic, on the plea that hallucinations which coincide with a death may be set off against hallucinations which occur without any coincidence, and both alike be regarded as purely subjective and without significance. Our own estimate of the probabilities is, of course, provisional, and may ultimately prove to be wide of the mark. But, meanwhile, it is at least proof against assault by conjectural statistics or the _obiter dicta_ of amateur psychologists.