Suppose, for example, that the ancestral species of a certain forest-butterfly habitually reposed on branches which hung near the ground and were covered with dry or rotten leaves; such a species would assume on its under surface a protective coloring which by its dark, brown, yellow, or red tints would tend toward similarity with such leaves. If, however, the descendants of this species should be subsequently compelled, no matter from what cause, to adopt the habit of resting on the green-leafed branches higher up, then from that period on the brown coloring would act less protectively than the shades verging towards green. And a process of selection will have set in which consisted first in giving preference only to such persons whose brown and yellow tints showed a tendency to green. Only on the assumption that such shades were possible by a displacement in the quantitative proportions of the different kinds of biophores composing the determinants of the scales affected, was a further development in the direction of green possible. Such being the case, however, that development _had to_ result; because fluctuations in the numerical proportions of the biophores are always taking place, and consequently the material for germinal selection is always at hand. At present it is impossible to determine exactly the magnitude of the initial stages of the deviations thus brought about and promoted by the sexual blending of characters; but it may perhaps be ascertained in the future, with exceptionally favorable material. Pending such special observations, however, it can only be said _a priori_ that slight changes in the composition of a determinant do not necessarily {50} condition similar slight deviations of the corresponding character,--in this case the color,--just as slight changes in the atomic composition of a molecule may result in bestowing upon the latter widely different properties. As soon, however, as the beginning has been made and a definite direction has been imparted to the variation, as the result of this or that primary variation's being preferred, the selective process must continue until the highest degree of faithfulness required by the species in the imitation of fresh leaves has been attained. That the foregoing process has actually taken place is evidenced not only by the presence of the beginnings of such transformations, as found for example in some greenish-tinted specimens of Kallima, but mainly by certain species of the South American genus Catonephele, all of which are forest-butterflies, and which, with many species having dark-brown under surfaces, present some also with bright green under surfaces--a green that is not like the fresh green of our beech and oak trees, but resembles the bright under surface of the cherry-laurel leaf, and is the color of the under surfaces of the thick, leathery leaves, colored dark-green above, borne by many trees in the tropics.