That new characters actually predominate in the last stage of the ontogeny, may also be demonstrated from the markings of caterpillars. It is, of course, not hereby implied, that throughout the whole animal kingdom new characters can only appear in the last ontogenetic stage. Haeckel is quite correct in maintaining that the power of adaptation of an organism is not restricted to any particular period. Under certain circumstances transformations may occur at any period of development; and it is precisely insects undergoing metamorphosis that prove this point, since their larvæ differ so widely from their imagines that the earlier stages may be completely disguised. It is here only signified that, with respect to the development of caterpillars, new characters first appear in the adult. The complexity of the markings, which so frequently increases with the age of the caterpillar, can scarcely bear any other interpretation than that the new characters were always acquired in the last stage of the ontogeny. In certain cases we are able, although with some uncertainty, to catch Nature in the act of adding a new character. I am disposed to regard the blood-red or rust-red spots which occur in the last stage of the three species of _Smerinthus_ larvæ in the neighbourhood of the oblique stripes as a case in point. It has already been shown that these red spots must be regarded as the first rudiments of the linear coloured edges which reach complete development in the genus _Sphinx_. In some specimens of _Smerinthus Tiliæ_ the spots coalesce so as to form an irregular coloured edge to the oblique stripes. In _S. Populi_ they occur in many individuals, but remain always in the spot stage; whilst _S. Ocellatus_ is but seldom, and _S. Quercus_ appears never to be spotted. The spots both of _S. Tiliæ_ and _Populi_ certainly do not show themselves exclusively in the fifth (last) stage, but also in the fourth, and sometimes in _Populi_ even as early as the third stage, from which we might be disposed to conclude that the new character did not first appear in the last stage. But the majority of the spotted individuals first acquire their spots in the fifth stage, and only a minority in the fourth; so that their occasional earlier appearance must be ascribed to the backward transference of a character acquired in the fifth stage. Moreover, the fourth and fifth stages of the caterpillars are closely analogous both in size, mode of life, and marking, and are therefore analogous with reference to the environment, so that it is to be expected that new characters, when depending on adaptation, would be rapidly transferred from the fifth stage to the fourth.[128] We should thus have a case of the acceleration by natural selection, of processes determined by innate causes. Why changes should predominate in the last stage, is a question closely connected with that of the causes of larval markings in general, and may therefore be investigated later.