14 ME. C. B. CLAEKE ON INDIAN SPECIES OF CTPEETJS. 



the glume, come to that view) a point of little or no value in 

 defining species. The spikes (as explained above under G. pu- 

 milus, and as shown by the discussion of fig. 27) pass so insensibly, 

 in the same species and in the same plant, from simple to com- 

 pound, that the presence of this bract is even in species like G. 

 inundatus, Roxb., where it has been fixed on as an important 

 character, highly uncertain. 



The characters of the spike itself are of high differential im- 

 portance, and also in high degree constitute the features of the 

 whole umbel. We have first the spikes of few spikelets, as in 

 C. elegans, the spikelets strictly digitate. In some of this set (as 

 G. dijfums, "Willd., G. multistriatus, Boeck.) we have " pedicelled 

 spikelets," which I consider, however, always as spikelets sessile 

 in spikes of one spikelet. Erom this we have " stellate spikes;" 

 and so by degrees the rhachis more elongate, until we have cylindric 

 spikes with arhachis 1-5 centim. long: in these the spikelets maybe 

 densely crowded or distant, but many species (as G. exaltatus) are 

 very variable in this respect. When the spikes are long, bowing 

 with subremote spikelets, they are often described as racemose, 

 which view is somewhat supported sometimes by the remote empty 

 (or deciduous) lower glumes ; but I think in all these cases the 

 spikelets should be considered sessile and the ultimate divisions 

 of the inflorescence " spikes." 



While the inflorescence is thus flexible in character, it must 

 remain a chief character in all species ; but exactly the Bame kind 

 of inflorescence meets us in various (and remote) sections of the 

 genus, so that reliance on external general character of the inflo- 

 rescence, without also examination of the nut and stigmas, has 

 led botanists into numerous errors for the last century. C. cepha- 

 lotes, G. pygmcBus, and C. dubius all have a single capitulum, the 

 leaves and bracts much alike ; they were variously confounded, 

 then mixed, then reseparated under new names by Roxburgh, 

 Wallich, and others. Boeckeler has united with G. pygmceus a\ao 

 Isolepis Micheliana, Roem. et Sch., which I agree with Kunth 

 (as attempted to be proved below) to be of a different genus. C. 

 procerus and G. puncticulatus, than which no two species are more 

 essentially distinct, are confounded in the description of Roxburgh 

 and in the collections of Wallich, and the confusion has (by the 

 issue of specimens correctly matched from K.ew) been widely dis- 

 seminated. Roxburgh, after figuring 0, amoenus, Heyne MS., 

 guessed, on comparing RottboelPs figure of 0. alopecuroides, that 



