NOTES ON COMMELINACEX. 197 
extends north to Amurland, and Tradescantia virginica to Lake 
Wisconsin. Several species reach southwards to Monte Video and 
the Cape of Good Hope; in Australia they are hardly found so far 
from the tropic. 
e arrangement of stamens in the genus Commelina is repre- 
sented diagrammatically in fig. 8; the whorls are regularly placed, 
the three of the outer whorl opposite the sepals, the three of the 
inner whorl opposite the petals; two stamens of the outer, one of 
the inner whorl being fertile ; the fertile stamen of the inner whorl 
d 
others; the barren stamen of the outer whorl (next the outer sepal) 
any 
he stamens throughout the order follow this arrangement ; 
the first stamen to become sterile and to disappear being that 
a 
figured by Wight alternate with the sterile, and I have copied some 
of his figures in my ‘ Commelinacee Bengalenses but Bentham has 
The stamens in Aneilema are 
really as in Commelina, but by a displacement of the whorls the 
two fertile ones (which are really next the two inner sepals) are 
declinated so as to appear opposite the two inner petals. 
But Cochliostema is a remarkable exception, as shown in fig. 2; 
it has those stamens fertile which in Commelina are sterile, and 
those sterile which are in Commelina fertile. Cochliostema, so far 
as its stamens are concerned, stands in the same relation to other 
Commelinacee that Cypripedium does to other Orchids; but Cochlio- 
stema has other strongly distinctive characters. In the Com- 
ith each cell of the ovary, these 
are arranged vertically on two placental nearly coincident lines at 
the inner angle of the cell; in many cases the ovules are evidently 
one ce 
Sect: Dichaspermum seen fromwithout. Dichespermwn, founde on 
i ed genus distinct from 
both fully represented among the Aneilemas with 1-ranked seeds. 
* aracter ‘* Dich 
