ME. 0. BENTHAM ON GRAMINEJi. 33 



The tribes composing the series of Panicacese run much into 

 each other, and have been very variously extended or reduced. 

 We have adopted the following six, as having appeared to us to 

 be rather better defined than the smaller or larger ones that have 

 been proposed. 



Tribe I. PAiiTiCEiE. 



The principal character of the Paniceae, considered as a tribe 

 of Panicacese, consists in the hardening of the fruiting glume. 

 In several of the smaller genera, however, and even in some 

 species of Panicum itself, it is membranous, but usually larger 

 than the outer ones, and forming the chief covering of the fruit, 

 never hyaline or much reduced as in Andropogonese. Orymopsis, 

 Milium, and their allies, which were formerly included in Panicese, 

 have been transferred to Agrostidese on account of the persistent 

 lower glumes below the articulation. Among the other general 

 characters of the tribe, the inarticulate rhachis of inflorescence is 

 constant except in Stenotaphrum jwherejlaowever, the articulation 

 is very tardy and not constant, so that it has often been denied. 

 The flowering glume never bears the twisted awn, so general in 

 Andropogoneee and Tristeginess, although in UriocMoa and a very 

 few species of Panicum its obtuse apex has a short, erect, almost 

 dorsal point ; the awns of Oplismenus, Ghcetaria, the section 

 Ijchinochloa of Panioum, &c. are straight and terminate one or 

 more of the empty glumes only. The fertile flower terminating 

 the spikelet is, in the normal genera, either perfectly hermaphro- 

 dite, or, at any rate, as far as I have observed, has staminodia 

 round the pistil. It is only in a few of the abnormal genera 

 added to the tribe that there are strictly female spikelets. 



The normal genera of the tribe may be distributed in four 

 rather distinct groups, though scarcely marked enough to be 

 raised to the rank of subtribes ; and to these we would add a few 

 more or less abnormal genera, but little connected with each 

 other, but all apparently more nearly allied to Paniceje than to 

 any other tribe. 



In the first group, or Paniaece proper, we have distinguished 

 eleven genera — a number somewhat arbitrary ; for much might be 

 said in favour either of uniting the whole into one vast genus 

 Panicum, or of dividing them still further, as some have proposed, 

 into about twice as many as those here adopted, the distinctive 

 characters being often either very uncertain, or such as are not 

 universally recognized as generic in the Order. 



