NOTES ON COMMELINACES. 199 
it is indehiscent, often smaller than the others; in the third sec- 
tion it is altogether suppressed, or occasionally present in a reduced 
barren form. Some of the species with 2-celled capsules may 
possibly prove only varieties of corresponding species, with 3-cell 
capsules as Bentham suspects, which is an additional ground for 
not adopting some genera of Hasskarl that differ only in the 
number of the cells being two or three 
Tricarpellaria, having the capsule equally 3-celled, Dicarpellaria 
having the capsule 2-celled, the third dorsal cell being occasional 
one flower I have seen the connective in one narrow, in another 
very wide, in the third intermediate. 
5. As regards specific characters there is little to be said 
peculiar to this order. I cannot distinguish the species by eye 
myself, nor do I believe that Mr. Bentham himself can (at all 
safely); Wallich has pasted down nine species (belonging to three 
genera) under Commelina communis, Linn., Wall. Cat., 8978 (no 
one of the nine is, however, Commelina commums, Linn.) On the 
8 
sometimes very hairy viscid, has sometimes solitary, generally 
agglomerated spathes ; yet Hook. f. and T. Thoms., and, 1 believe, 
all the Bengal botanists, are agreed that the whole series makes up 
but one species. < Co : 
I am therefore very little satisfied by the genus Dichorisandra, in 
