There is another crucial point in pragmatism which the defenders of
the system are apt to pass over lightly, but which Mr. Russell regards
(justly, I think) as of decisive importance. Is, namely, the pragmatic
account of truth intended to cover all knowledge, or one kind of
knowledge only? Apparently the most authoritative pragmatists admit
that it covers one kind only; for there are two sorts of self-evidence
in which, they say, it is not concerned: first, the dialectical
relation between essences; and second, the known occurrence or
experience of facts. There are obvious reasons why these two kinds of
cognitions, so interesting to Mr. Russell, are not felt by pragmatists
to constitute exceptions worth considering. Dialectical relations,
they will say, are verbal only; that is, they define ideal objects,
and certainty in these cases does not coerce existence, or touch
contingent fact at all. On the other hand, such apprehension as seizes
on some matter of fact, as, for instance, "I feel pain," or "I
expected to feel this pain, and it is now verifying my expectation,"
though often true propositions, are not _theoretical_ truths; they are
not, it is supposed, questionable beliefs but rather immediate
observations. Yet many of these apprehensions of fact (or all,
perhaps, if we examine them scrupulously) involve the veracity of
memory, surely a highly questionable sort of truth; and, moreover,
verification, the pragmatic test of truth, would be obviously
impossible to apply, if the prophecy supposed to be verified were not
assumed to be truly remembered. How shall we know that our expectation
is fulfilled, if we do not know directly that we had such an
expectation? But if we know our past experience directly--not merely
knew it when present, but know now what it was, and how it has led
down to the present--this amounts to enough knowledge to make up a
tolerable system of the universe, without invoking pragmatic
verification or "truth" at all. I have never been able to discover