that "Epistemology clears the way for metaphysical construction or hypothesis."[1] But he treats epistemology as if it were a science clearly separable from metaphysics, so much so that he thinks it possible for us to be 'realists' in our epistemology, while we are 'idealists' in our metaphysics.[2] There is an intelligible sense in which it can be said that mere subjective idealism—the assertion that we never can know anything beyond the 'states of consciousness' which are the hypostatised abstractions with which the psychologist may profess to work—is inconsistent with idealism in the sense in which that means a belief in the ultimate rationality of the universe. But Mr. Seth sets up 'reals' in epistemology—the supposed absolutely existing 'things' of ordinary picture-thinking—in order to knock them down in metaphysics, by regarding them only as "moments in the being of an intelligently directed Life." It would seem easier, at least, and more obviously logical, to base such a metaphysical theory on an epistemology which denied the possibility of knowing anything that existed independently of all thought, and to base a denial of such a metaphysical theory on an epistemology which made the fact of knowledge require the existence of a plurality of absolutely existing 'reals.'
If metaphysics be strictly limited to speculative metaphysics, the attempt to frame an all-embracing hypothesis about the ultimate nature of the universe as a whole, we can, of course, distinguish that part of philosophy (whether possible or not) from an inquiry into the conditions of knowledge; but we cannot safely separate such speculations from the preliminary inquiry. If our epistemology gives us no ground for any belief in any unity of the cosmos whatever or in any rationality in the process of it, the attempt to explain it as a whole is condemned at the outset. The attempt to construct a speculative metaphysics, however tentative and hypothetical, is only defensible if we feel some justification for believing that there is a cosmos to be explained, and that it must be to some extent intelligible by us. That is to say, in our epistemology, we are already, if we are
