ractions which are
not the elements out of which objects of sense are constructed. They are
abstractions made from those objects whose ground for isolation is found
in the peculiar problems of experimental psychology, such as those of
color or tone perception. It would be impossible to make anything in
terms of Berkeleyan sense-data and of symbolic logic out of any
scientific discovery. Research defines its problem by isolating certain
facts which appear for the time being not as the sense-data of a
solipsistic mind, but as experiences of an individual in a highly
organized society, facts which, because they are in conflict with
accepted doctrines, must be described so that they can be experienced by
others under like conditions. The ground for the analysis which leads to
such facts is found in the conflict between the accepted theory and the
experience of the individual scientist. The analysis is strictly _ad
hoc_. As far as possible the exception is stated in terms of accepted
meanings. Only where the meaning is in contradiction with the experience
does the fact appear as the happening to an individual and become a
paragraph out of his biography. But as such an event, whose existence
for science depends upon the acceptance of the description of him to
whom it has happened, it must have all the setting of circumstantial
evidence. Part of this circumstantial evidence is found in so-called
scientific control, that is, the evidence that conditions were such that
similar experiences could happen to others and could be described as
they are described in the account given. Other parts of this evidence
which we call corroborative are found in the statements of others which
bear out details of this peculiar event, though it is important to note
that these details have to be wrenched from their settings to give this
corroborative value. To be most conclusive they must have no intentional
connection with the experience of the scientist. In other words, those
individuals who corroborate the facts are made, in spite of themselves,
experiencers of the same facts. The perfection of this evidence is
attained when the fact can happen to others and the observer simply
details the conditions under which he made the observation, which can
be then so perfectly reproduced that others may repeat the exceptional
experience.

This process is not an analysis of a known world into ultimate elements
and their relations. Such an analysis never is