It may therefore be argued, adversely, that in the selected coincidental
hallucinations, the persons seen were in the class most usually beheld in
non-coincidental and, probably, purely subjective hallucinations
representing real persons; also, that knowledge of their illness, even
when no anxiety existed, kept them in some cases before the mind; also,
that several cases are foreign, and that 'most foreigners are fools.' On
the other hand, affection, familiarity, and knowledge of illness had _not_
produced hallucinations even in the case of these percipients, till
within the twelve hours (often much less) of the event of death.

It would have been desirable, of course, to publish all the
_non_-coincidental cases, and show how far, in these not _veridical_
cases, the recognised phantasms were those of kindred, dear friends, known
to be ill, and subjects of anxiety[17].

The Census, in fact, does contain a chapter on 'Mental and Nervous
Conditions in connection with Hallucinations,' such as anxiety, grief,
and overwork. Do these produce, or probably produce, many empty
hallucinations _not_ coincident with death or any great crisis? If they
do, then all cases in which a coincidental hallucination occurred
to a person in anxiety, or overstrained, will seem to be, probably,
fortuitous coincidences like the others. All percipients, of all sorts
of hallucinations, hits or misses, were asked if they were in grief or
anxiety. Now, out of 1,622 cases of hallucination of all known kinds
(coincidental or not), mental strain was reported in 220 instances; of
which 131 were cases of grief about known deaths or anxiety. These mental
conditions, therefore, occur only in twelve per cent. of the instances. On
the whole, it does not seem fair to argue that anxiety produces so much
hallucination that it will account by itself for those which we have