gical correlations may be necessary, we see
that there is great risk of including among them some which are not.

With regard to the great mass of the correlations, however, including
all the _indirect_ ones, Professor Huxley seems to us warranted in
denying that they are necessary; and we now propose to show deductively
the truth of his thesis. Let us begin with an analogy.

Whoever has been through an extensive iron-works, has seen a gigantic
pair of shears worked by machinery, and used for cutting in two, bars of
iron that are from time to time thrust between its blades. Supposing
these blades to be the only visible parts of the apparatus, anyone
observing their movements (or rather the movement of one, for the other
is commonly fixed), will see from the manner in which the angle
increases and decreases, and from the curve described by the moving
extremity, that there must be some centre of motion--either a pivot or
an external box equivalent to it. This may be regarded as a necessary
correlation. Moreover, he might infer that beyond the centre of motion
the moving blade was produced into a lever, to which the power was
applied; but as another arrangement is just possible, this could not be
called anything more than a highly probable correlation. If now he went
a step further, and asked how the reciprocal movement was given to the
lever, he would perhaps conclude that it was given by a crank. But if he
knew anything of mechanics, he would know that it might possibly be
given by an eccentric. Or again, he would know that the effect could be
achieved by a cam. That is to say, he would see that there was no
necessary correlation between the shears and the remoter parts of the
apparatus. Take another case. The plate of a printing-press is required
to move up and down to the extent of an inch or so; and it must exert
its greatest pressure when it reaches the extreme of its downward
movement. If now anyone will look over the stock of a printing-press
maker, he will see half a dozen different mechanical arrangements by
which these ends are achieved; and a machinist would tell him that as
many more might readily be invented. If, then, there is no necessary
correlation between the special parts of a machine, still less is there
between those of an organism.

From a converse point of view the same truth is manifest. Bearing in
mind the above analogy, it will be foreseen that an alteration in one
part of an organism will not