We see, then, as we go up the taxonomic, and undoubtedly also the phylogenic series, that there is a cross-breeding of more and more diverse individuals, a funding of more and more divergent characteristics in a common offspring. Why is this? I answer, for the sake of _better results in the offspring_. This is abundantly shown by direct experiment. In hermaphroditic plants in which there may be either self-fertilization or else cross-fertilization with other individuals of the same species, the latter produces better results in number and vigor of offspring. But there are other advantages, more difficult to prove but none the less certain, and of the greatest importance in evolution: First, as already stated, complexity of inheritance, like complexity of composition in a chemical substance, gives instability to the embryo, and thus liability to variation in the offspring; and this in its turn furnishes the material for selection of the fittest. Again, it seems to me that there is a direct tendency to improve the offspring by a sort of struggle in the embryo among the various qualities inherited from both sides, and a survival of the best and strongest--a sort of pre-potency of strong qualities. Can divergence of uniting individuals and the funding of diverse characteristics go any further? It may. The differences of the uniting individual may be still further increased, and the resulting offspring still further improved by the cross-breeding of different varieties of the same species, for we thus add varietal differences to sexual differences in the uniting individuals. It is well known that too close breeding, or consanguineous breeding, or breeding in and in, as it is variously called, if continued long, has a bad effect on the offspring, weakening the stock, while judicious crossing of varieties within certain limits of difference has a good effect, strengthening the stock and increasing its fertility. It probably does so in two ways: one direct, by funding many diverse qualities from both sides, and the survival in the offspring of the strongest and best; the other indirect, by giving _plasticity_, instability to the embryo, and variability to the offspring, and therefore abundant material for the operation of selection, either by man or by Nature. We said, “within certain limits of difference.” If the difference is extreme, as in extreme varieties and races, then the effect becomes again bad, and more and more so as the limit of specific difference is approached; at which limit at last Nature shuts down and forbids the bans. Thus, then, there is in cross-breeding a regular law of effect, increasing to a maximum and again decreasing, which may be graphically represented by a curve (Fig. 69). In this figure the horizontal line represents the ordinary level of the type; distances on this line represent differences, individual, varietal, or specific; ordinates above or below represent the effect, good or bad, of crossing. Thus s s′