A more important source of error has next to be considered, namely, for-getfulness; either a temporary and superficial forgetfulness, which may lead people to answer "No" when further care and thought would have produced the answer "Yes": or the real oblivion which in course of time may banish many experiences beyond the power of recall. In order to judge of the effect of this, the number of visual hallucinations reported were tabulated according to the time that had elapsed in each case between their occurrence and the date when the percipient answered the Census question. It was found that the number was comparatively large for the most recent year, and decreased rapidly as the years became more remote - at eight, nine, and ten years ago becoming less than half what it was at first (Report, table iv., pp. 43 and 66). The conclusion (based on this table and on other considerations given fully on pp. 63-65 of the Report) was arrived at that, roughly speaking, the number of visual hallucinations reported must be multiplied by four to give the true number experienced over the age of ten.

At the same time, the numbers reported are valuable as a record of the number of hallucinations remembered, and afford means of comparing the number and impressiveness of different kinds of hallucinations.

Chapters IV. to XI. of the Report contain discussions of -

IV. The phenomena of hallucinations in general, different degrees of externalisation of hallucinations, and their relation to sensory impressions which are not fully externalised, and to the power of visualisation.

V. The relation of illusions to hallucinations, and the part played by points de repère in the latter. VI. The form and development of hallucinations. VII. Their physiology; the physiological action involved in most hallucinations is probably exclusively cerebral. VIII. Age, sex, heredity, nationality, and health considered in relation to hallucinations. Number of persons who have had more than one hallucination, and cases in which the same form of hallucination is repeated several times. IX. Mental and nervous conditions; effects of emotion and anxiety. X. Expectancy and suggestion; the operation of suggestion in the production and development of hallucinations. XI. Organic effects sometimes accompanying hallucinations, and the operation of suggestion in producing these.

These chapters are profusely illustrated with examples of all kinds.

For our present purpose the main question relates to the reported proportion of veridical hallucinations, and especially of "death-coincidences," that is to say, cases in which a recognised apparition occurred within twelve hours of the death of the person represented by it, the death being unknown to the percipient at the time {Report, Chapters XII. and XIII.).

Excluding cases occurring to percipients who have had other hallucinations, the exact number of which was not reported, and experiences of children under ten, we find 65 death-coincidences (Report, p.209) reported at first-hand out of the total of 350 recognised apparitions of living persons {ibid., p. 246). Since death-coincidences are better remembered and more spoken of than the generality of hallucinations, a disproportionate number may be expected to have been known by the collectors, and, in fact, 25 per cent, of death-coincidences, as against 8 per cent, of all other recognised apparitions of living persons, are stated to have been so known. Some allowance must therefore be made in this class of cases for the possibility of error arising from selection.

Of the 65 cases, 3 were known to have been selected, and therefore left out of account in the calculation. Of the remaining 62, 16 were known beforehand to the collector; 26 were not known; and for the remaining 20 there was no evidence either way. On the supposition that the influence of previous knowledge and selection was the same proportionately in those cases as in the cases where its operation was ascertainable (v. Report, p. 243, footnote 3 for detailed calculation), a deduction of 8 was made to cover the possible disproportion due to selection. A further special allowance was made for the possibility of exaggeration, since in comparing death-coincidences with hallucinations in general, it was found that the remoter cases of death-coincidences were unduly numerous in proportion to the recent ones, which suggested that some of the remote cases reported were not really coincidental. On this ground, 22 cases were deducted, leaving 32 death-coincidences {Report, pp. 242-43).