in many Compositae, Umbelliferae, etc.,--in the structure of the central flower in some plants,--in the two kinds of flowers produced by cleistogene species,--and in several other such cases. These instances clearly prove that the flowers on the same plant have often varied independently of one another in many important respects, such variations having been fixed, like those on distinct plants during the development of species. It was therefore necessary to ascertain by experiment what would be the effect of intercrossing flowers on the same plant, in comparison with fertilising them with their own pollen or crossing them with pollen from a distinct plant. Trials were carefully made on five genera belonging to four families; and in only one case, namely, Digitalis, did the offspring from a cross between the flowers on the same plant receive any benefit, and the benefit here was small compared with that derived from a cross between distinct plants. In the chapter on Fertility, when we consider the effects of cross-fertilisation and self-fertilisation on the productiveness of the parent-plants we shall arrive at nearly the same result, namely, that a cross between the flowers on the same plant does not at all increase the number of the seeds, or only occasionally and to a slight degree. I will now give an abstract of the results of the five trials which were made. 1. Digitalis purpurea. Seedlings raised from intercrossed flowers on the same plant, and others from flowers fertilised with their own pollen, were grown in the usual manner in competition with one another on the opposite sides of ten pots. In this and the four following cases, the details may be found under the head of each species. In eight pots, in which the plants did not grow much crowded, the flower-stems on sixteen intercrossed plants were in height to those on sixteen self-fertilised plants, as 100 to 94. In the two other pots on which the plants grew much crowded, the flower-stems on nine intercrossed plants were in height to those on nine self-fertilised plants, as 100 to 90. That the intercrossed plants in these two latter pots had a real advantage over their self-fertilised opponents, was well shown by their relative weights when cut down, which was as 100 to 78. The mean height of the flower-stems on the twenty-five intercrossed plants in the ten pots taken together, was to that of the flower-stems on the twenty-five self-fertilised plants, as 100 to 92. Thus the intercrossed plants were certainly superior to the self-fertilised in some degree; but their superiority was small compared with that of the offspring from a cross between distinct plants over the self-fertilised, this being in the ratio of 100 to 70 in height. Nor does this latter ratio show at all fairly the great superiority of the plants derived from a cross between distinct individuals over the self-fertilised, as the former produced more than twice as many