colour-splashes of an 'impressionist' landscape the different splashes must be exactly where they are, otherwise on stepping back from the picture one would see, not a Haarlem hyacinth-field, or an avenue of poplars with their golden autumn leaves, but a mere unintelligible daub. It is the _type_ of the colour-pattern that must be attained, and in nature this is attained very slowly, step by step, spot after spot, and therefore, obviously, no correct stroke once attained will be given up again, since, in combination with the rest, it secures the proper type of colour-pattern. Only thus, it seems to me, can we understand how apparently meaningless lines, like the figures 1840 on the under surface of _Vanessa atalanta_, could have become a constant characteristic of the species. To sum up briefly, we may say that sexual selection is a much more powerful factor in transformation than we should at first be inclined to believe. It cannot, of course, have been operative in the case of plants, nor can it be taken into consideration in regard to the lower animals, for these, like the plants, do not pair, or, at any rate, do so without any possibility of choice. Animals which live on the sea-floor, or which are attached there, must simply liberate their reproductive cells into the water, and cannot secure that they unite with those of this or that individual. This is the case among sponges, corals, and Hydroid polyps. In some other classes the sense organs are too poorly developed, and the eyes in particular too imperfect to be excited in different degrees by any peculiarities in the appearance or behaviour of the males. This is what Darwin meant when he ascribed to these animals 'too imperfect senses and much too low intelligence' 'to estimate the beauty or other attractive points of the opposite sex, or to feel anything like rivalry.' Accordingly, in the Protozoa, Echinoderms, Medusæ, and Ctenophores, secondary sexual characters are entirely absent, as pairing also is. In those worms that pair we first meet with secondary sexual characters, and from this level upwards they are never quite absent from any large group, and gradually play an increasingly important rôle. But the significance of sexual selection lies, as we have seen, not only in the fact that one sex of a species, usually the male, is modified, but in the possibility of the transference of this modification to the females, and further, in the fact that the process of variation may start afresh at any time, and thus one variation may be developed upon or alongside of another. In this way we can explain certain complex and often fantastic forms and colourings which we could not otherwise understand; thus the extraordinary number of nearly related species in some animal groups, such as butterflies and birds, in which the differences mainly concern the colour-patterns.