That many flowers have been rendered conspicuous for the sake of guiding insects to them is highly probable or almost certain; but it may be asked, have other flowers been rendered inconspicuous so that they may not be frequently visited, or have they merely retained a former and primitive condition?  If a plant were much reduced in size, so probably would be the flowers through correlated growth, and this may possibly account for some cases; but the size and colour of the corolla are both extremely variable characters, and it can hardly be doubted that if large and brightly-coloured flowers were advantageous to any species, these could be acquired through natural selection within a moderate lapse of time, as indeed we see with most alpine plants.  Papilionaceous flowers are manifestly constructed in relation to the visits of insects, and it seems improbable, from the usual character of the group, that the progenitors of the genera Vicia and Trifolium produced such minute and unattractive flowers as those of V. hirsuta and T. procumbens.  We are thus led to infer that some plants either have not had their flowers increased in size, or have actually had them reduced and purposely rendered inconspicuous, so that they are now but little visited by insects.  In either case they must also have acquired or retained a high degree of self-fertility.

If it became from any cause advantageous to a species to have its capacity for self-fertilisation increased, there is little difficulty in believing that this could readily be effected; for three cases of plants varying in such a manner as to be more fertile with their own pollen than they originally were, occurred in the course of my few experiments, namely, with Mimulus, Ipomoea, and Nicotiana.  Nor is there any reason to doubt that many kinds of plants are capable under favourable circumstances of propagating themselves for very many generations by self-fertilisation.  This is the case with the varieties of Pisum sativum and of Lathyrus odoratus which are cultivated in England, and with Ophrys apifera and some other plants in a state of nature.  Nevertheless, most or all of these plants retain structures in an efficient state which cannot be of the least use excepting for cross-fertilisation.  We have also seen reason to suspect that self-fertilisation is in some peculiar manner beneficial to certain plants; but if this be really the case, the benefit thus derived is far more than counter-balanced by a cross with a fresh stock or with a slightly different variety.

Notwithstanding the several considerations just advanced, it seems to me highly improbable that plants bearing small and inconspicuous flowers have been or should continue to be subjected to self-fertilisation for a long series of generations.  I think so, not from the evil which manifestly follows from self-fertilisation, in many cases even in the first generation, as with Viola tricolor, Sarothamnus, Nemophila, Cyclamen,