[Illustration: FIG. 52. The fertilization of the Yucca. _A_, ovipositor of the Yucca-moth. _op_, its sheath. _sp_, its apex. _op_^1, the protruded oviduct. _B_, two ovaries of the Yucca, showing the holes by which the young moths escape, and (_r_) a caterpillar in the interior. _C_, head of the female moth, with the sickle-shaped process (_si_) on the maxillary palps for sweeping off the pollen and rolling it into a ball. _mx_^1, the proboscis. _au_, eye. _p_^1 base of first leg. _D_, longitudinal section through an ovary of the Yucca, soon after the laying of two eggs (_ei_). _stk_, the canal made by the ovipositor.] Here, then, we find an adaptation of certain parts of the moth's body in relation to the fertilization of the flower, but in this case it is as much in the interest of the moth as of the plant. By carrying the pollen to the stigma the moths secure the development of the ovules, which serve their offspring as food, so that we have here to do with a peculiar form of care for offspring, which is not more remarkable than many other kinds of brood-care in insects, such as ants, bees, Sphex-wasps, ichneumon-flies, and gall-flies. It might be objected that this case of the Yucca is not so much one of effecting fertilization as of parasitism; but the eggs, which are laid in the seed-pods, are very few, and the caterpillars which emerge from them only devour a very small proportion of the seeds, of which there may be about 200 (Fig. 52, _B_). Thus the plants also derive an advantage from the moth's procedure, for quite enough seeds are left. The form and position of the stamens and of the stigma seem to be as exactly adapted to the visits of the moth as the moth is to the transference of the pollen, for the Yucca can only be fertilized by this one moth, and sets no seed if the moth be absent. For this reason the species of Yucca cultivated in Europe remain sterile. Thus the apparent contradiction is explained, and the facts everywhere support the hypothesis that the adaptations between flowers and insects depend upon processes of selection. This origin is incontrovertibly proved, it seems to me, in another way, namely, by the merely _relative_ perfection of the adaptations, or rather, by their relative imperfection. I have already pointed out that all adaptations which depend upon natural selection can only be relatively perfect, as follows from the nature of their efficient causes, for natural selection only operates as long as a further increase of the character concerned would be of advantage to the existence of the species. It cannot be operative beyond this point, because the existence of the species cannot be more perfectly secured in this direction, or, to speak more precisely, because further variations in the direction hitherto followed would no longer be improvements, even though they might appear so to us.