THE COMMELTNACEiE OF BE^GAL. 447 



entirely exserted from the bracts at the branching of the pa- 

 nicle. But Mr. Knrz has lately collected in Birma a remark- 

 able Anilema with two perfect stamens (and moniliforra hairs on 

 the filaments), in which the bracts are subcucullate and the pe- 

 dicels subincluded therein. Perhaps the only absolute distinc- 

 tion remaining between the two genera is that in Anilema the 

 stamens are either two or, if three, then the middle one has not 

 larger anthers than the others. However, the only Gommelyna 

 (known to me) that has a regular trilocular capsule, viz. Comm, 

 obliqua^ differs widely in its bracts and inflorescence from all the 

 3-8eeded Anilemw. 



In the 3-stanieued species of Anilema there are often only 

 three, sometimes only two, sterile dijQformed stamens ; but in larger 

 specimens of the same species sometimes four sterile stamens are 

 seen. So in the 3-stamened species the sterile stamens are often 

 two or three on the same panicle. The pains often taken in 

 describing these sterile stamens seems to me ill-bestowed. 



The number of perfect stamens in each species is tolerably 

 constant. Thus A, lierhaceum has generally three perfect sta- 

 mens ; but occasionally it has the middle one of these three 

 smaller than the other two, and sometimes this middle stamen is 

 altogether absent. So, in A. protensuin^ the fertile stamens 

 normally three, but not infrequently two only. 



In specific distinction, stress is laid on " filamenta barbata " as 

 opposed to "filamenta imberbia." These moniliform hairs often 

 are very difficult to find in the bud, being developed greatly at 

 the time of expansion of the flower. But, further, though they 

 are generally characteristic of A. herlaceum^ I have found fully 

 opened flowers of A. herhaceum without a trace of these hairs. 

 So R, Brown assigns to A, vaginatum bearded filaments (as 1 

 have always seen them), whereas Wight found them beardless, 

 though it is clear from his figure that he had got the true plant. 



On the whole, I find that (as in Commelijna) the ovules and 

 seeds supply the most constant characters. 



Sect. I. 



1. Anilema scapiflorum, Wight, Icones, t. 2073, — Commelyna sca- 

 piflora, Roxb. Flor. Ind. i. p. 1/5, & t. 1521, Calc. Herb, — Anilema 

 tuberosum, Hamilt. in Wall. Cat, n. 5207. — Murdanuia scapilioiu, 

 Royle, lllust, t. 95. 



Foha ensiformia, subglabra, omnia radicalia. Scapa radlcalis ; panienlii 

 robusta. Stamina 3 perfccta. Seraina 15-21, angulata, dorso laevia. 



Uab, in sylvis Mudhopoor ad Daccam. 



The root is more or less bulbous, whence Hamilton's name. 



2 II 2 



