Wallace, on the other hand, refuses to do this, and says that "the best answer to the difficulty" of supposing natural selection to have been the only cause of co-adaptation may be "found in the fact that the very thing said to be impossible by variation and natural selection, has been again and again affected by variation and artificial selection[41]." This analogy (which Darwin had already and very properly adduced with regard to the _blending of adaptations_) he enforces by special illustrations; but he does not appear to perceive that it misses the whole and only point of the "difficulty" against which it is brought. For the case which his analogy sustains is not that which Darwin, Spencer, Broca and others, mean by _co-adaptation_: it is the case of a blending of _adaptations_. It is not the case where adaptation is _first initiated in spite of intercrossing_, by a fortuitous concurrence of variations each in itself being without adaptive value: it is the case where adaptation is _afterwards increased by means of intercrossing_, through the blending of variations each of which has always been in itself of adaptive value.
From this I hope it will be apparent that the only way in which the "difficulty" from co-adaptation can be logically met by the ultra-Darwinian school, is by denying that the phenomenon of co-adaptation (as distinguished from the blending of adaptations) is ever to be really met with in organic nature. It may be argued that in all cases where co-adaptation _appears_ to occur, closer examination will show that the facts are really due to a blending of adaptations. The characters A + B + C + D, which are now found united in the same organism, and, as thus united, all conspiring to a common end, may originally have been distributed among different organisms, where they _severally_ subserved some other ends--or possibly the same end, though in a less efficient manner. Obviously, however, in this case their subsequent combination in the same organism would not be an instance of co-adaptation, but merely of an advantageous blending together of already existing adaptations. This argument, or rejoinder, has in point of fact been adopted by Professor Meldola, he believes that all cases of seeming co-adaptation are thus due to a mere blending of adaptations[42]. Of course, if this position can be maintained, the whole difficulty from co-adaptation would lapse. But even then it would lapse on the ground of _fact_. It would not have been overturned, or in any way affected, by Wallace's _argument_ from artificial selection. For, in that event, no such argument would be required, and, if adduced, would be irrelevant, since no one has ever alleged that there is any difficulty in understanding the mere confluence of adaptations by free-intercrossing of the best adapted.