THE INDIAN SPECIES OF ERIOCATJLON. 141
did not notice it at all. Steudel and Koerniche were both aware of this difference, but neither paid any particular attention to it. Apart from the improbability of so very definite a change as the loss or acquirement of the black colour happening more than once, yellow anthers are so distinctive that I have no hesitation in making a ' section ' of the species which possess it. The protruding male petal is also a very distinctive character and one naturally used in any scheme of classification. It appears to be a very constant character, and nearly all the species which show it do so quite definitely. But as might be supposed indications of the habit are not wanting in other species, some of which may therefore be regarded as on the line of development. The stamens do not vary in number. They are always six, except in the one dimerous- flowered species.
In the female flower the ovary is invariably 3-Iobed (except of course in dimerous flowered species) — there is no reduction. The petals are 3 or 2, and only slight differences occur between those of dif- ferent species ; except in one direction. Asa rule they are oblanceolate, with thick terminal hairs and slender, longer, lateral ones ; but in some species there is a brush of slender filaments or hairs, which might be regarded as due either to the longitudinal splitting of the petal into many parts or to a narrowing of the petal accompanied by an increase in the number and length of the basal hairs. This latter change may well have come more than once, in different groups, I therefore do not use it as a 'sectional' characteristic.
The Sepals.
The sepals shows the most interesting variation. The simplest probably primitive, form is boat-shaped, black in colour and with a few hairs along the mid-rib or keel. In one Himalayan species the sepals are connected into a calyx similar to that of the male flower. The petals of this species differ from others in having the gland terminal and in being clawed. Ruhland has a group of Chino-Japanese species with these character, I therefore found my section CONNATO-SEPALAE to include them. In all other species the sepals are free. A development of the boat-shaped sepal is the formation of an enlargement (a thick- ening, wing, or crest) along the keel. In some species this takes the form merely of a thickening (E. Thoniasi PI. I. fig. 7) in others of a narrow wing or crest which may be lobed, or pectinate (figs. 5 & 6.) The depth of the thickening or crest appears to vary in the same species but there is usually no doubt about the crest when it is present. I therefore make a ' section ' of those species which possess this en- largement, whatever its precise form may be. Here again it may be that E. sexangulare, E. cuspidatum and E. Thomasi, are not derived
1697—19
