All that the theory requires is, that the selective and eliminative processes do, _on an average_, secure their ends, and in the same way the theory of sexual selection does not need the assumption that every female is in a position to exercise a scrupulous choice from among a troop of males, but only that, on an average, the males more agreeable to the females are selected, and those less agreeable rejected. If this is the case, it must result in the male characters most attractive to the females gaining preponderance, and becoming more and more firmly established in the species, increasing in intensity, and finally becoming a stable possession of all the males.
When we go more into details we shall see that the _particular qualities_ of the distinctive masculine characters are exactly such as they would be if they owed their existence to processes of selection; in other words, from this point of view the phenomena of the decorative sexual characters can be understood up to a certain point. It seems to me that we are bound to accept the process of sexual selection as really operative, and instead of throwing doubt upon it, because the choice of the females can rarely be directly established, we should rather deduce from the numerous sexual characters of the males, which have a significance only in relation to courtship, that the females of the species are sensitive to these distinguishing characters, and are really capable of exercising a choice.
In my mind at least there remains no doubt that the 'sexual selection' of Darwin is an important factor in the transformation of species, even if I only take into consideration those secondary sexual characters which are related to wooing. We shall see, however, that there are others in regard to whose origin through processes of selection doubt is still less legitimate, and from which, on this account, we can argue back to the courtship characters.
The first beginning of transformation is not, even in ordinary natural selection, to be understood as due to selection, but is to be regarded as _a given variation_ (the causes of which we shall discuss later on); it is only the increase of such incipient variations in a definite direction that can depend on natural selection, and they _must_ depend on it in so far as the transformations are purposeful. Now, all secondary sexual characters can be recognized as useful, save only the decorative distinctions, although these also undoubtedly represent intensifications of originally unimportant variations. Are we then to regard these alone as the mere outcome of the internal impulsive forces of the organism, while in the case of the analogous sexual characters for tracking, catching, and holding the female, and so forth, the augmentation and the directing must be referred to processes of selection? But if there be any utility at all in the decorative sexual characters it can only lie in their greater attractiveness to the females, and it can only be of any account if the females have, in a certain sense, the power of choice. Independently, therefore, of direct observations as to the actual occurrence of choosing, we should be compelled by our chain of reasoning to assume that there was such a power of choice--and I shall immediately discuss it more precisely.