as regards both qualities; and though there is reason, as we shall presently see, to conclude that on the whole the offspring are usually inferior to the parents, nevertheless, since they are generally numerous, and the process of selection almost always severe, the survivors in each generation, as a rule, surpass their parents in the sum of their qualities; and as a result, there is usually evolution as regards the whole race even when the essential qualities are many. Moreover, we have assumed for the sake of simplicity that each ancestor was pre-eminent in one quality only, but of course, as constantly happens in nature, a single individual may approach pre-eminence in several or many qualities, or conversely, may fall below the specific mean in several or many qualities: in the latter case he tends of course to be eliminated, and to leave no offspring, and therefore to have no influence on the future of the race; but in the former he tends, even more than if he were eminent in one quality only, to survive and to leave offspring who, inheriting his characteristics, tend as a result to survive in greater numbers than the offspring of the less happily endowed, and therefore to render concurrent evolution in several directions more rapid.
Lastly, the statement that artificial selection differs from natural selection in that but a few qualities are developed by it, is true in a very limited sense only; for instance, by causing the evolution of the greyhound man has necessarily caused the evolution of a great number of subsidiary qualities, each subserved by a number, in some cases by an immense number, of finely co-ordinated structures. Therefore though speed alone has been aimed at, yet to obtain it evolution in many directions has been necessary (e.g. in "wind," "limb," circulatory and nervous systems, &c.). On this score then there is no essential difference between artificial
