if the germ-plasm contains determinants at all by whose fluctuations in a plus or minus direction the appropriate variation is attainable.”[69] But why _must_ they? There is no ‘must’ about Chance, unless one extends its operations to infinity. Why is it so certain that the inequalities of nutriment, on which hereditary variability is supposed to depend, must necessarily run the gamut of all possible variations? There is no ‘must’ in this theory, except that it is the last ditch of the “mechanical conception of the economy of life.” It ‘must’ be true—or that conception must quit the field. Were evolution to depend on the occurrence, by pure chance, of a few appropriate variations among a vast multitude of indifferent or disadvantageous ones, is it conceivable that we should find in nature anything like the infinite wealth of closely and beautifully adapted structure which is actually present? In particular, how are we to account for the cases in which a number of parts are so modified as to work together in harmonious co-adaptation? Each of these parts, according to Weismann, originates quite independently of the others. Take the case of the Indian leaf-butterfly already referred to.[70] The first beginnings of the midrib on Weismann’s theory had nothing to do with the rest of the rib, nor had any of the veinings with this, or with one another; and the contour of the leaf, sending out a little projection like a stalk _exactly where the midrib starts_, originated quite independently of that marking, and equally so of the leaf it mimics! To explain co-adaptations like this on Weismann’s theory is really much the same as to suppose that a picture could be painted by simply plastering the scrapings of a palette on a canvas, if only one continued the process long enough. And the marvel in question, the co-adaptation of various parts, has not been attained once or twice but, to a greater or less degree, in _every_ organism possessing any structural complexity.