a manner that the motion of the common ancestral form must occur as the resultant. To such demonstration there is however as yet a long step. Haeckel himself moreover points out that his hypothesis is by no means a “mechanical theory of heredity,” but only an introduction to this theory, which he hopes “will be capable of being elevated to the rank of a genetic molecular theory” (_loc. cit._ p. 17). But although we must also confess with the critic of the “Philosophy of the Unconscious,” that “the facts of heredity have hitherto defied every scientific explanation,”[299] this furnishes us with no excuse for flying to a metaphysical explanation, “which is here certainly least able to satisfy the inability to understand the connection arising from natural laws.” It is not to be wondered at that Von Hartmann, on the ground of the “Unconscious” on which he takes his stand, speaks of the law of correlation as an unconscious acknowledgment of a “non-mechanical universal principle on the side of Darwinism.” By “correlation” he understands something quite different to the idea which we attach to this expression. He supposes that “Darwinism sees itself compelled to acknowledge through empirical facts the uniform correlation of characters pertaining to the specific type; but it thereby contradicts its mechanical principles of explanation, all of which amount to the same thing as conceiving the type as a mosaic, chequered, superficial, and accidental aggregate of characters, which have been singly acquired, contemporaneously or successively, by selection or habit.” I do not believe, however, that any such conception has ever been admitted either by Darwin or any one else. The admission that not all, but only every deep-seated _physiological_ detailed modification, is or may be bound up with a system of correlated changes, indeed implies that we on our side also acknowledge an internal harmony of parts--an equilibrium, as I have above expressed it. But does this include the admission of a teleological principle, or exclude a mechanical explanation? Do we thereby acknowledge a “specific type” in the sense of an inseparably connected complex of characters, none of which can be taken away without all the others becoming modified? Does such a view agree generally with the empirical facts? Neither of these views appears to me to represent the case.