506 ME. G. BENTHAM OK CLASSIFICATION AND 



homologues in Junceae. In a few genera or species, however, the 

 typical arrangement is apparently at fault- Thus in the two 

 above-mentioned small genera Anarthria and Lepyrodia, the 

 glumes are small and not closely imbricate, and there are usually 

 two bracteoles under the perianth, bringing these genera nearer to 

 Junceae, from which, however, they are constantly distinguished 

 by their pendulous orthotropous ovules. In the females of Lep- 

 tocarpus Brownii and a few allied species, the flowers and sur- 

 rounding scales are so densely crowded in the compact compound 

 inflorescence that it is sometimes very difficult to draw the line 

 between the bracts, glumes, bracteoles, and perianth-segments, 

 whilst in the males of the same speeies the spikelets are quite 

 normal. In the monotypic genera Onychosepalum and Chcetan- 

 thusj the perianths are reduced to small hyaline scales or setae, 

 and the habit is quite that of some Cyperaceae ; but they have always 

 the leaf-sheaths slit to the base and the pendulous ovules of true 

 Restiaceae. 



In Cyperaceae the glumal arrangement is usually normal, as in 

 Kestiaceae ; but the perianth is, with few not very certain excep- 

 tions, either deficient or reduced to small hyaline scales or setae ; 

 and the presence of bracteoles or of a so-called perigynium within 

 the glumes is much more frequent than in Eestiaceae. The term 

 glume, however, is not so generally used by cyperologists as I 

 think it ought to be ; several irregularities, which seem scarcely 

 justifiable, have crept into their terminology ; and in a very few 

 genera there are real difficulties in describing properly the parts 

 of the spikelet. A few instances may here be given in illustration 

 of these irregularities in diction or in reality. 



In Kyllinga the spikelet has been described as having usually 

 four scales or glumes, of which two lower ones small and empty, 

 and two larger flowering ones enclosing one flower. Now, as in 

 the general character of the order each flowering scale is said to 

 enclose one or rarely more flowers, it was difficult to conceive how 

 a spikelet with two flowering scales should have but one flower ; 

 and on examination the fact proves to be that in most species of 

 Kyllinga the spikelet has but three glumes, two empty and one 

 flowering one. What has been termed the lowest small scale is 

 not part of the spikelet at all, but is a bract inserted on the re- 

 ceptacle or main axis of inflorescence, and subtends the spikelet ; 

 the next scale, though still small and hyaline, is rather larger or 

 sometimes considerably larger than the bract, and is the lowest 



