do
it by bidding against each other, in case any of them are making
additions to their mills or shops, and also by bidding against
any new employers who may appear. Perfect competition requires
that this bidding for labor and capital shall continue up to the
profit-annihilating point. Here, as elsewhere in the purely static
part of the discussion, we have to make assumptions that are
rigorously theoretical and put out of view in a remorseless way
disturbing elements which appear in real life. The static state
requires that all _entrepreneurs_ who survive the sharp tests of
competition should have equally productive establishments, which means
that they should all be able to get the same amount of product from a
given amount of labor and capital. The actual fact is that differences
of productive power still survive. There are some small establishments
which, within the little spheres in which they act, are as productive
as large ones; but there are also some which are struggling hopelessly
against large rivals in the general market and are destined erelong to
give up the contest. In other words, the centralizing and leveling
effects of competition are approximated but never completely realized
in actual life.

A fact that it is well to note is that the test of final productivity
is inaccurately made when unduly large amounts of labor and capital
are made the basis of the measurement. Take away, for instance, a
quarter of the working force, estimate the reduction of the product
which this withdrawal occasions, and attribute this loss entirely to
the labor which has been taken away, and you estimate it too highly.
With so large a section of the labor withdrawn the capital would work
at a disadvantage, and a part of the reduction of the product would be
due to this fact. If we should take away all the labor, the capital
would be completely paralyzed, and the product would become _nil_. It
would obviously be inaccurate to say that the whole product is
attributable to the labor, on the ground that withdrawing the labor
annihilates it all. With any large part of the labor treated as a
single unit, the loss of product occasioned by a withdrawal of such a
unit is more than can be accurately imputed to it as its specific
product. The smaller the increments or units are made, the less
important is this element of inaccuracy, and it becomes a wholly
negligible quantity when they become very small. A study of the forms
of the pro