Herr Parish next censures the probable selection of good cases by collectors, on which the editors of the Census have already made observations, as they have also made large allowances for this cause of error. He then offers the astonishing statement that, 'in the view of the English authors, a view which is, of course, assumed in all calculations of the kind, an hallucination persists equally long in the memory and is equally readily recalled in reply to a question, whether the experience made but a slight impression on the percipient, or affected him deeply, as would be the case, for instance, if the hallucination had been found to coincide with the death of a near relative or friend.'[5] This assertion of Herr Parish's is so erroneous that the Report expressly says 'as years recede into the distance,' the proportion of the hallucinations that are remembered in them to those which are forgotten, or at least ignored, 'is very large.' Again, 'Hallucinations of the most impressive class will not only be better remembered than others, but will, we may reasonably suppose, be more often mentioned by the percipients to their friends.'[6] Yet Herr Parish avers that, in all calculations, it is assumed that hallucinations are equally readily recalled whether impressive or not! Once more, the Report says (p. 246), '_It is not the case_' that coincidental (and impressive) hallucinations are as easily subject to oblivion as non-coincidental, and non-impressive ones. The editors therefore multiply the non-coincidental cases by four, arguing that no coincidental cases (hits) are forgotten, while three out of four non-coincidentals (misses) are forgotten, or may be supposed likely to be forgotten. Immediately after declaring that the English authors suppose all hallucinations to be equally well remembered (which is the precise reverse of what they do say), Herr Parish admits that the authors multiply the misses by four, 'influenced by other considerations' (p. 289). By what other considerations? They give their reason (that very reason which they decline to entertain, says Herr Parish), namely, that misses are four times as likely to be forgotten as hits. 'To go into the reason for adopting this plan would lead us too far,' he writes. Why, it is the very reason which, he says, does _not_ find favour with the English authors! How curiously remote from being 'coincidental' with plain facts, or 'veridical' at all, is this scientific criticism! Herr Parish says that a 'view' (which does not exist) is 'of course assumed in all calculations;' and, on the very same page, he says that it is _not_ assumed! 'The witnesses of the report--influenced, it is true, by other considerations' (which is not the case), 'have sought to turn the point of this objection by multiplying the whole number of (non-coincidental) cases by four.' Then the 'view' is _not_ 'assumed in all calculations,' as Herr Parish has just asserted.