304 Stapf. — On the Sonenieae of Asia. 
the shorter petioles and very slender peduncles and pedicels. 
Cogniaux describes it as having 4 folia valde asymmetrical but 
that may be a slip, as they are hardly asymmetric at all, par- 
ticularly in the type-specimens of Wallich (4094). 
VII. Group of S. linearis. South-west Ceylon; Chota 
Nagpore ; Kumaon to South China and the Philippines, 
southwards to Penang. 
1. S', linearis , Hook. f. Moulmein, 900 m. 
2. S’. Guneratnei , Trim. South-west Ceylon. 
These two are characterised by very narrow one-nerved 
leaves, an almost ovoid calyx, and slender, sub-cylindric cap- 
sules. They are annuals with a thin, somewhat wiry stem. The 
former was found on Mount Gerai in Moulmein, the latter was 
discovered by Trimen in the Pasdun Corle in Ceylon. Both 
species resemble each other in so high a degree that they are, 
in my opinion, undistinguishable. It is true, Cogniaux at- 
tributes to S'. linearis opposite, and to S. Guneratnei alternate 
leaves. But I find the leaves in both as a rule in whorls 
of four, but sometimes of three, and sometimes they are 
opposite. If I keep them separate for the present, it is solely 
because the fruit of 5 . Guneratnei is not known, and it might 
be that it constitutes a differential character. 
3. S'. angustata , Triana. South-west Ceylon. 
The leaves are broader than in S'. Guneratnei , but still 
lanceolate, with coarse crenations and a distinct middle nerve 
besides two very faint side ones. The flowers are not known, 
but the capsules diverge very clearly from the linearis-ty^o. 
towards that of the zeylanica-g roup. It was indeed named 
first S'. rhombifolia v. angustata by Thwaites in sched. 
3. S', erecta , Jack. Penang to Moulmein. 
Whilst S'. Guneratnei , and still more S', angustata , point to 
a close affinity with the zeylanica-group from Ceylon, we find 
in this and the following species types, the evolution of which 
lies in a different direction, and ends, if I may say so, blind, 
without links towards any other group. S', erecta is distinctly 
different from S', linearis , but its close affinity is still clear 
