ype. In neither
workmanship nor acquisition can one fix upon routine as the "normal"
type, hoping to derive or to explain away the inevitable residue of
"outstanding cases." For as a matter of fact the outstanding cases prove
to be our only clue to a knowledge of how routine is made.[50]

The above formula will apply, with the appropriate changes of emphasis,
to buyers and sellers in an organized market, as well as to the parties
to a simple transaction of barter. Two main empirical characteristics of
the economic situation are suggested in putting the statement in just
these terms. In the first place, the primary problem in such a situation
is that of "exchange valuation," the fixation of a "subjective" (or
better, a "personal") price ratio between what the agent wishes to
acquire and whatever it is that he offers in exchange. The agent thus is
engaged in determining what shall be the relative importance for himself
of _two_ commodities or exchangeable goods. And in the second place
these goods get their values determined together and in relation to each
other, never singly and with a view to _subsequent_ comparison. These
values when they have been determined will be measured in terms of
marginal utility in accordance with familiar principles, but the
marginal utilities that are to express the attained and accepted ratio
at which exchange eventually takes place are not known quantities at all
in the inception of the process of comparison. If these dogmatic
statements seem to issue in hopeless paradox or worse, then let us not
fear to face the paradox and fix its lines with all possible
distinctness. Can a man decide to offer so much of one commodity for so
much of another unless he _first_ has settled what each is worth to him
in some intelligible terms or other? And is not this latter in point of
fact the real decision--at all events clearly more than half the battle?
Does not the exchange ratio to which one can agree "leap to the eyes,"
in fact, as soon as the absolute values in the case have been once
isolated and given numerical expression?

In a single word we here join issue. For the comparison in such a case
is _constructive comparison_, not a mechanical measuring of fixed
magnitudes, as the above objection tacitly assumes. And constructive
comparison is essentially a transitive or inductive operation whereby
the agent moves from one level to another, altering his standard of
living in some more or less import