I may take this opportunity of remarking that naturalists have observed that in most of the great classes a series exists from very complicated to very simple beings; thus in Fish, what a range there is between the sand-eel and shark,--in the Articulata, between the common crab and the Daphnia{479},--between the Aphis and butterfly, and between a mite and a spider{480}. Now the observation just made, namely, that selection might tend to simplify, as well as to complicate, explains this; for we can see that during the endless geologico-geographical changes, and consequent isolation of species, a station occupied in other districts by less complicated animals might be left unfilled, and be occupied by a degraded form of a higher or more complicated class; and it would by no means follow that, when the two regions became united, the degraded organism would give way to the aboriginally lower organism. According to our theory, there is obviously no power tending constantly to exalt species, except the mutual struggle between the different individuals and classes; but from the strong and general hereditary tendency we might expect to find some tendency to progressive complication in the successive production of new organic forms. {479} Compare _Origin_, Ed. i. p. 419, vi. p. 575. {480} <Note in original.> Scarcely possible to distinguish between non-development and retrograde development. _Modification by selection of the forms of immature animals._ I have above remarked that the feline{481} form is quite of secondary importance to the embryo and to the kitten. Of course, during any great and prolonged change of structure in the mature animal, it might, and often would be, indispensable that the form of the embryo should be changed; and this could be effected, owing to the hereditary tendency at corresponding ages, by selection, equally well as in mature age: thus if the embryo tended to become, or to remain, either over its whole body or in certain parts, too bulky, the female parent would die or suffer more during parturition; and as in the case of the calves with large hinder quarters{482}, the peculiarity must be either eliminated or the species become extinct. Where an embryonic form has to seek its own food, its structure and adaptation is just as important to the species as that of the full-grown animal; and as we have seen that a peculiarity appearing in a caterpillar (or in a child, as shown by the hereditariness of peculiarities in the milk-teeth) reappears in its offspring, so we can at once see that our common principle of the selection of slight accidental variations would modify and adapt a caterpillar to a new or changing condition, precisely as in the full-grown butterfly. Hence probably it is that caterpillars of different species of the Lepidoptera differ more than those embryos, at a corresponding early period of life, do which remain inactive in the womb of their parents. The parent during