140 THE JOURNAL OE INDIAN BOTANY.
show that the last named species is extremely variable in this res- pect, every intermediate form being found from large conical heads with short reflexed bracts to small flat heads with long bracts. It is clear therefore that, at least with this species, the relative length of the involucral bracts is of varietal rank only. But if this be so then perhaps it is also with other species ; and the peculiar Burmese plant E. Martianum Wall, should be considered a variety of E. quin- guangulare, not as either a distinct species or a proliferous of state {cp. Hooker (1) p. 582.) A similar but less pronounced lengthening of the involucral tracts appears in two other species, in other respects widely separated, E. gracile Mart, and E. cuspidatum Dalz. At the same time extreme variation of this kind should probably be consi- dered specific— e. g. in E. xeranthenum Mart. -and E. roseum, as also of course when accompanied by differences in the flowers {E. Ediuardii). But it is clear that the form and length of the bracts is not in itself of ' sectional ' ranks as Koerniche supposed.
The colour of the involucral bracts is some guide, as in the separa- tion of what I take to be E. atratnm Koern. from E. subcaulescens Hook, f., but it is an uncertain one, the black colour fading often out of these bracts with age. On the other hand the scarious, usually straw- coloured, floral bracts of some species are easily distinguished from the more common black with white hairs. The difference is definite, with no merging, and I take it to be of ' sectional ' value. Black bracts are usually hairy and scarious bracts glabrous, but there are exceptions.
Hairs on the receptacle are a constant feature of most species, but are partly or altogether absent from others ; and Hooker made use of this in the F.B.I. This difference accompanies others of ' sectional ' rank, but may occur inside the section and is I believe of only secon- dary importance. Another character to which Hooker gave some im- portance is the length of the pedicel (stipes.) I am not able to follow him in this. The length appears to me to depend on the age of the individual flower, and to be therefore of no importance.
The Flower , its Petals and Stamens.
It is in the flower itself that the greatest differences are found. In the male the chief distinguishing characters are white or yellow instead of black anthers, and one corolla-lobe being so much longer than the rest or to protrude beyond the floral bracts and cover them. The first gives an absolute line, not in any way bridged over by an occasional species with olive-green colour. Euhland stated (lc. p. 16) that the colour is constant in the species, but used this difference more than once in his clavis as if of only minor importance. Hooker
