uncommon. But the general result of the combined details was one which
could not possibly be known to A. before his death; nor to B. could it be
known at all. Yet B.'s dream represented this general result with perfect
accuracy, which cannot be accounted for by the revival of subconscious
memory in sleep. Neither asleep nor awake can a man remember what it is
impossible for him to have known. The dream contained no _prediction_ for
the results were now fixed; but (granting the good faith of the narrator)
the dream did contain information not normally accessible.

However, by way of psychological explanation of the dream, my friend cited
Coleridge's legend, as to the German girl and her unconscious knowledge of
certain learned languages. 'And what is the evidence for the truth of
Coleridge's legend?' Of course, there is none, or none known to all the
psychologists who quote it from Coleridge. Neither, if true, was the
legend to the point. However, psychology will accept such unauthenticated
narratives, and yet will scoff at first baud, duly corroborated testimony
from living and honourable people, about recent events.

Only a great force of prejudice can explain this acceptance, by
psychologists, of one kind of marvellous tale on no evidence, and this
rejection of another class of marvellous tale, when supported by first
hand, signed and corroborated evidence, of living witnesses. I see only
one escape for psychologists from this dilemma. Their marvellous tales are
_possible_, though unvouched for, because they have always heard them and
repeated them in lectures, and read and repeated them in books. _Our_
marvellous tales are impossible, because the psychologists know that they
are impossible, which means that they have not been familiar with them,
from youth upwards, in lectures and manuals. But man has no right to have
'clear ideas of the possible and impossible,' like Faraday, _a priori_,
except in the exact sciences. There are other instances of weak evidence