138 THE JOURNAL OF INDIAN BOTANY.
to estimate if possible the relative importance from the phylogenetic stand- point of the more stable characters.
Stem and Leaf.
As said above since practically all the Eriocaulons grow either in swampy ground or submerged in water the stem and leaves of any one species vary but little, even in size. At the same time all sub- merged forms are for a like reason so alike among themselves, and the swamp forms also among themselves, that such differences as exist are of little use in separating the species. The difference between the usual disc-like stem, and an elongate branching one, which Ruhland following Koerniche used almost at the fore-front of his clavis of old-world species, though at first sight it may seem very definite, is not always a hard and fast line, and is in any case proba- bly bound up with the robustness of the species and the nature of its habitat. As instance the Nilgiri E. robust am and the Ceylon E. caulescens referred to just above. But I have found the former with root-stock over on inch in length, and poorer specimens with leaves narrow enough to be indistinguishable from those of the Ceylon plant. In Ruhland's list one is No. 74, the other No. 120.
The Head and its Involucre.
As regards the heads Koerniche used the difference of hairy and glabrous involucre, as also did Hooker and Euhland ; and without doubt this character is of sectional value. But I find the hairiness varies and "may even be absent from a plant undoubtedly for other reasons allied to hairy species. This character must not therefore bo used two rigidly, as in Ruhland's wide separation on this account of E. Brownianum Mart, from E. nilagirense Steud. (Nos. 93 & 117 respectively.) The specimens in Herb. Calc- show plainly that the type sheet of the former is of a not fully undeveloped plant and that the absence of hairs is here accidental.
Characters which give very distinctive appearance to the head and would certainly appear at first sight of at least specific value are afforded, by the form and length of the involucral bracts. Thus they are horizontal and very obtuse and slightly turned up at the end in E. sexangulare, E. luzulaefolium, E. truncatum and E. Thioaitesii ; they are acute and ultimately reflexed in E quinquangulare and E. trilobum. In E. xeranthcmum, E. roscum, E. martianum Wall, and forms of E. Dianae collected in Coorg, they are very much longer than the head, projecting beyond the general margin like the rays of a sun-flower. (PI. fig. 12 & 13.) Koerniche placed great value on this. But a comparison of a large number of collections made on the Western Ghats from Salsette to Calicut, all with the same flower,
