but on a _selection of relevant instances_, and never claim to be based
upon an _exhaustive_ knowledge of particulars. Hence _in form_ the most
satisfactory induction is always incomplete, and differs in no wise from
a bad one. 'All bodies fall to the ground' is an induction which has
worked. 'All swans are white' broke down when black swans were
discovered in Australia. The validity of an induction, then, is not a
question of form.

The necessity for such selection no intellectualist theory of Induction
has understood. All have aimed at exhaustiveness, and imagined that if
it could be attained, inductive reasoning would be rendered sound, and
not impossible. Their ideal 'cause' was the totality of reality,
identified with its 'effect,' in a meaningless tautology. Nothing but
voluntarism can enable logicians to see that our actual procedure in
knowing is the reverse of this, that causal explanation is the
_analysis_ of a continuum, and that 'phenomena,' 'events,' 'effects,'
and 'causes' are all creations of our selective attention; that in
selecting them we run a risk of analyzing falsely, and that if we do,
our 'inductions' will be worthless. But whether they are right or wrong,
valuable or not, real reasoning from 'facts' can never be a 'formally
valid' process.

We are thus brought to see the hollowness of the contention that 'Pure
Reason' can ignore its psychological context and dehumanize itself. A
thought, to be thought at all, must seem _worth_ thinking to someone, it
must convey the meaning he intends, it must be true in his eyes and
relevant to his purposes in the situation in which it arises--_i.e._, it
must have a motive, a value, a meaning, a purpose, a context, and be
selected from a greater whole for its relevance to these. None of these
features does intellectualist logic deign to recognize. For if truth is