But when we consider the aggregation of these variations in a process of evolution, his theory seems unsatisfactory. We have already seen that what we commonly call a variation involves not one change, but a series of changes, each term of which is necessary. Muscle, nerve, and ganglion must all vary simultaneously and correspondingly. Correlation and combination are just as essential as variation. And evolution often demands the disappearance of less fit structures just as much as the advance of the fittest. Says Osborne, "It is misleading to base our theory of evolution and heredity solely upon entire organs; in the hand and foot we have numerous cases of muscles in close contiguity, one steadily developing, the other degenerating." Weismann offers the explanation that "if the average amount of food which an animal can assimilate every day remains constant for a considerable time, it follows that a strong influx toward one organ must be accompanied by a drain upon others, and this tendency will increase, from generation to generation, in proportion to the development of the growing organ, which is favored by natural selection in its increased blood-supply, etc.; while the operation of natural selection has also determined the organ which can bear a corresponding loss without detriment to the organism as a whole."[A] [Footnote A: Weismann, Essays, p. 88.] Here again natural selection of individuals, not the diminished supply of nutriment, has to determine which of many muscles shall be poorly fed and which favored. But natural selection can favor special organs only indirectly through the individuals which possess such organs. Variation is fortuitous, and there is nothing, except natural selection, to combine or direct them. And, I think, we have already seen that any theory which neglects or excludes such directing and combining agencies must be unsatisfactory and inadequate. Weismann has promised us an explanation of correlation of variation in accordance with his theory; and if such an explanation can be made, it would remove one of the strongest objections. But for the present the objection has very great weight. Furthermore, as Osborne has insisted, linear variations, or variations proceeding along certain single and well-marked lines, would seem inexplicable by, if not fatal to, Weismann's theory. And yet Osborne, Cope, and others have shown that the teeth of mammals have developed steadily along well-marked lines. They have apparently not resulted at all by selection from a host of fortuitous variations.