440 MR. C. B. CLAEKE OK 



On both tbe peduncles the innermost flower expands first ; 

 and the tendency of each successively expanding flower to be- 

 come male is greater than in its predecessor. Also the ten- 

 dency to become male is far greater in the first-expanding long- 

 peduncled lower raceme than in the upper' one. The innermost 

 flower on the upper raceme is always perfect, and generally the 

 next to it, sometimes more. But on the lower raceme the inner- 

 most flower even is generally male ; and this lower raceme is 

 often rudimentary, sometimes obsolete ; and these three cases 

 all occur in one species. Moreover, when the long peduncle 

 exists carrying (as nearly always) male flowers only, it falls off* at 

 its articulation before the other peduncle comes well into flower ; 

 and thus has been increased the discrepancy in the description 

 of plants so well known and well marked as G. hengalensis. In 

 the stamens a great change (as in most orders) takes place at the 

 moment of expansion : the anthers are often all yellow and sub- 

 similar till that moment ; but then the two outside stamens' 

 anthers turn blue, and the middle stamens' anthers enlarge, be- 

 come divaricate, and twisted at the base. 



I have united (vide C. dbliqua below) the most viscid hairy 

 plant of the genus with a particularly glabrous form. 



Having thus discarded several marks relied on for specific 

 distinctions in this genus, I have taken up characters derived 

 from the fruit. It has been objected to me, by the highest 

 authority, that in the Commelyna communis bundle of the Kew 

 Herbarium the capsules are as often 4-3-seeded as 5-seeded. I 

 have examined such a very large number of individuals of this S])e- 

 cies here, that I confidently state that no such amount of variation 

 in the fruit will be found in C communis as it grows in Bengal ; 

 and from the state in which the genus Commelyna usually exists 

 in herbaria, I am not disposed to attach much weight to this 

 objection. 



Systematic books usually attribute to the order Commelynaceae 

 an embryo opposite the hilum. The seeds in this order are gene- 

 rally so essentially nnsymmetrical that it is hard to say where 

 the axis of the seed is ; but in the seeds of C. oHiqua and other 

 species where the axis is definite, I find the embryo not directly 

 opposite the hilum, but placed very obliquely. 



The sections I have taken for the Bengal Commelynce repre- 

 sent the three arragements of the capsule. 



In Sect. I. a, the capsule is essentially unsymmetrical. Two 



