It is a well-known yet remarkable fact that the circumferential flowers of many of the foregoing plants have both their male and female reproductive organs aborted, as with the Hydrangea, Viburnum and certain Compositae; or the male organs alone are aborted, as in many Compositae. Between the sexless, female and hermaphrodite states of these latter flowers, the finest gradations may be traced, as Hildebrand has shown. (Introduction/10. ‘Ueber die Geschlechtsverhaltnisse bei den Compositen’ 1869 pages 78-91.) He also shows that there is a close relation between the size of the corolla in the ray- florets and the degree of abortion in their reproductive organs. As we have good reason to believe that these florets are highly serviceable to the plants which possess them, more especially by rendering the flower-heads conspicuous to insects, it is a natural inference that their corollas have been increased in size for this special purpose; and that their development has subsequently led, through the principle of compensation or balancement, to the more or less complete reduction of the reproductive organs. But an opposite view may be maintained, namely, that the reproductive organs first began to fail, as often happens under cultivation, and, as a consequence, the corolla became, through compensation, more highly developed. (Introduction/11. I have discussed this subject in my ‘Variation of Animals and Plants under Domestication’ chapter 18 2nd edition volume 2 pages 152, 156.) This view, however, is not probable, for when hermaphrodite plants become dioecious or gyno-dioecious--that is, are converted into hermaphrodites and females--the corolla of the female seems to be almost invariably reduced in size in consequence of the abortion of the male organs. The difference in the result in these two classes of cases, may perhaps be accounted for by the matter saved through the abortion of the male organs in the females of gyno-dioecious and dioecious plants being directed (as we shall see in a future chapter) to the formation of an increased supply of seeds; whilst in the case of the exterior florets and flowers of the plants which we are here considering, such matter is expended in the development of a conspicuous corolla. Whether in the present class of cases the corolla was first affected, as seems to me the more probable view, or the reproductive organs first failed, their states of development are now firmly correlated. We see this well-illustrated in Hydrangea and Viburnum; for when these plants are cultivated, the corollas of both the interior and exterior flowers become largely developed, and their reproductive organs are aborted.