remain obscure, whether, with Mill, we make the intermediaries into 'possible sensations,' or, with Kant, hold the substructure of the order to be the work of an impersonal understanding. But suppose that my conscious perception has an entirely practical destination, that it simply indicates, in the aggregate of things, that which interests my possible action upon them: I can then understand that all the rest escapes me, and that, nevertheless, all the rest is of the same nature as what I perceive. My consciousness of matter is then no longer either subjective, as it is for English idealism, or relative, as it is for the Kantian idealism. It is not subjective, for it is in things rather than in me. It is not relative, because the relation between the 'phenomenon' and the 'thing' is not that of appearance to reality, but merely that of the part to the whole.
Here we seem to return to realism. But realism, unless corrected on an essential point, is asThe mistake is to set up homogeneous space as a real or even ideal medium prior to extension. inacceptable as idealism, and for the same reason. Idealism, we said, cannot pass from the order manifested in perception to the order which is successful in science, that is to say to reality. Inversely, realism fails to draw from reality the immediate consciousness which we have of it. Taking the point of view of ordinary realism, we have, on the one hand, a composite matter made up of more or less independent parts, diffused through-
