VIII. LEUCANTHERAE. The anthers are white or yellow ; otherwise as in lb. Many are water plants, with linear leaves. The group should probably be considered as of equal rank to the rest of the groups taken together. To -it belong E. horsley-kunda sp. nov., E. breviscapon Koern., E. rivulare Dalz., E. miscrum Koern., (incl E. mitophyllum Hook, f.), E. Huvatile Trimen., E. Siebolclianum Sieb. et Zucc. (Figs. 9, 10, 11.)
Many of the characters on which stress is laid in the published descriptions of species, such as the twisting of the scapes or the length of the pedicel, are either common to all species or dependant on age, and are therefore only an encumbrance to critical definition. Leaving these on one side and confining attention to the characters given above and their modifications, the species are found to be much more easily separable than has hitherto appeared. Following on this it is clear that some of the species have in the past been given far too wide a range of distribution, because of faulty identification and consequent confusion with other species. Thus E. luzulaefolium Mart., given in the F.B.I, and other works as occurring through- out India, is confined to the hilly tracts of Eastern Bengal and the Shan states ; E. quinquangulare L belongs only to South India, and chiefly to the eastern side ; and its place in Bengal is taken by E. trilobum, which does not occur in the Peninsula.
Several new species are described, but it is recognised that some at least of these may not be new, for I have failed to identify several described by Ruhland and others. Perhaps the strangest of the latter is E. melaleucum Mart., a species whiuh occurs in the F.B.I, and in other floras but has no named representative in the Herbarium of the Royal Botanic Gardens, or in any other Indian Herbarium. Ruhland describes the anthers as white, and my E. Horsley-kundae (fig. 9) may therefore be it, but Martius' original description is not quite definite enough and those of subsequent writers point to the probability of it having been confused with other species.
An interesting evolutionary fact comes to light in the existence of parallel development in different strains. Thus a reduction of one female sepal from a boat-shaped to a flat scale occurs as E. collinum crosses from S. India to Ceylon ; E. trilobum in Bengal differs from E. quinquangulare in S. India hardly at all, except in somewhat more pronounced a reduction ; and both these have their counterparts in E. Diana (fig. 3) in the Bombay Presidency, where every stage of this re- duction can be seen. Again a lengthening of the involucral bracts so that the head appears rayed occurs in Group 1 in E. Diana, in the stock of E. quinquangulare as var. Martiana (mihi) and as E. roseum mihi) (fig. 13) and in E. xeranthemum; in group III it occurs in var.
