MR. G. BENTHAM ON GRAMINEZE. 59 
take up Savi's. The Mexican P. elongatus, H.B. K., which is 
Presl's genus Nowodworskya (first described and figured by him 
under the name of Raspailia), has the pedicels, although clavate 
as in the rest of the genus, yet less decidedly articulate, thus 
forming some real connexion with the Agrostez. 
4. Garnorra, Brongn. (Migvelia, Nees, Berghausia, Endl.), 
which sometimes comes near to some forms of Polypogon, has, on 
the other hand, the spikelets in pairs on the inarticulate branches 
of the panicle as in Miscanthus, and thus very closely connects 
Tristeginee with Andropogones. It has, however, none of the 
long hairs on the rhachilla so common in Andropogone&, and 
cannot well be removed far from Arundinella, whilst Miscanthus 
is too near to Imperata to be rejected from Andropogone&. 
Garnotia comprises about eight species from East India, China, 
and Japan. 
5. ARUNDINELLA, Raddi,includes Goldbachia, Trin., Acratherum, 
Link, Thysanachne, Presl, and Brandtia, Kunth. It is the prin- 
cipal genus of the tribe, and comprises about twenty-four species 
spread over the tropical regions both of the New and the Old 
World, but chiefly in Asia. It is generally adopted and fairly 
characterized, though the habit and especially the inflorescence 
vary much, the panicle being sometimes long, narrow, and dense, 
or very large, loose, and spreading, with very numerous small or 
minute spikelets, whilst in a few species it is short and dense, 
forming almost an oval head with larger spikelets. The two 
sections proposed by Nees—Meliosaccharum, with a small tooth 
on the flowering glume on each side of the awn, and Acratherum, 
in which the glume is quite entire, tapering into the awn—do not 
prove to be well defined nor conformable to habit. 4. flammida, 
Trin., from Brazil and tropical Africa, has neither the habit nor 
the character of the genus, but is in every respect a Trichopteryz, 
with which it was not compared by Nees, Trinius, or Doell, 
because it was at first only known as Brazilian, and Trichopteryx 
was supposed to be exclusively African. 
6. Pu xxosPERMA, Munro, is a single Chinese species, nearly 
allied to Arundinella ; but there are three lodicules to the flower 
and no palea (unless one of the lodicules, although apparently in 
the same whorl as the others, be really a small palea), and the 
caryopsis is half exserted from the fruiting glumes as in some 
species of Sporobolus. Phaenosperma globosa, Munro, is a tall 
grass with a very large loose panicle, the slender but rigid 
