of the plantation, and call the remainder the costs of the raw
cotton. But this is really to reason in a circle. For in either case
the magnitude of the deduction depends on the marginal utility of the
cotton-seed. And the notion of the cost of anything becomes blurred
and blunted if we so use it that it must be deduced from the utility
of something else, which is not an agent in the production of the
thing in question.

This point is not merely an academic one. It means that we cannot
explain the _relative_ prices of cotton lint and cotton-seed in terms
of cost at all, whether marginal or otherwise. The influence of cost
will be confined to the _sum_ of the prices of the two things. Upon
this sum it will exert precisely the same influence as it exerts upon
price in general, by affecting the total quantities of the two things
that will be supplied. But upon the distribution of this sum between
lint and seed, cost will exert no influence whatever, because it
cannot affect the proportions in which they are supplied. It may
assist some readers if I state the matter in more concrete terms. Cost
of production will be one of the factors which will result in the
production of an annual cotton crop in the United States of, let us
say, 10 million tons of seed cotton. This crop will yield roughly
6-2/3 million tons of cotton-seed, and 3-1/3 million tons (or rather
more than 13 million bales) of lint. The combined price received by
the planter of (let us say) 14.4 cents for 1 pound of lint plus 2
pounds of seed should correspond roughly to the marginal joint costs
of production. But the factor of cost has no influence at all in
determining that this combined price is made up of a price of 12 cents
per pound for lint, and only 1.2 cents per pound (or $24 per ton) for
cotton-seed. To account for this we must rely entirely upon demand. We
can say, shortly, that the respective prices must be such as will