45 



*{? 



LXXXIX. GKAMINEjE. 507 



{Spike cylindrical or slightly interrupted. Flowering glumes white and mem- 

 branous - ... 40. Kceleria. 

 Clusters one-sided, in an irregular spike or clos,e panicle. Glumes herbaceous 

 and rigid 32. Dactylis. 



4 o/ Outer glumes enclosing the flowers or nearly so 44 



\ Outer glumes shorter than the flowers 45 



{Glumes obtuse, coloured. Two flowers with a wedge-shaped rudimentary 

 glume 38. Melica. 

 Outer glumes very acute. About 3 flowers in the spikelet, all perfect, with 3- 

 toothed glumes 39. Triodta. 

 Glumes all acute. Two male flowers and 1 smaller perfect one in the spikelet. 

 4. HlEROCHLOE. 



'First outer glume very small, second broadly truncate at the top, and often 



coloured. Flowers 2 in the spikelet 36. Catabrosa. 



Both outer glumes pointed or obtuse, not truncate. Flowers usually 3 or 



more 46 



( Spikelets broadly ovate or orbicular. Glumes closely packed and very spread- 



46 -j ing . . 34. BRIZA. 



( Spikelets oblong or linear 47 



' Flowering glumes rounded on the back, at least at the base . . . .48 



. Flowering glumes keeled on the back 49 



48 / Flowering glumes obtuse or rather acute 35. Poa. 



\ Flowering glumes very pointed or shortly awned . . . 31. Festuca. 



{Flowering glumes very pointed. Flowers about 3, with a bristle-like continua- 

 tion of the axil 37. Molinia. 

 Flowering glumes obtuse or acute. Flowers 3 or more, the last terminal 35. Poa. 



The limits of the numerous genera into which Grasses are divided are 

 as yet far from being definitely fixed. Some are by no means natural, 

 and those which are so have not always any definite characters. They 

 have also been variously distributed into tribes, according to the special 

 views of their structure entertained by different botanists. Amongst 

 those which appeared to be the most generally adopted, eight tribes 

 were, in the first editions of this work, distinguished as being re- 

 presented in Britain. Since then a detailed study of a considerable 

 part of the Order, in connection with the Australian Flora, has sug- 

 gested some modifications in the general grouping of the genera, but 

 as the new tribes cannot yet be considered as definitely settled, the 

 sequence adopted in the first edition is here left undisturbed. A general 

 sketch, however, of the proposed arrangement as affecting British 

 genera is now added, observing at the same time that here and there 

 an exceptional species may not be in strict conformity with the tribal 

 character; but fewer, it is believed, than in the case of any other 

 method hitherto proposed. 



* Pedicel articulate below the outer glumes. Spikelet with 1 terminal fertile flower, 

 the male or barren ftoiver (if any) or empty glumes below it. (Panicace^j.) 



1. Panice^e. Flowering glume of a firmer texture than the empty ones below 

 it:— 3. Panicum. 



2. ANDROPOGONEiE. Flowering glume much thinner and smaller than the empty 

 one below it, and often bearing a twisted awn : — 12. Polypogon. 



The above two tribes, so sparingly represented in Britain, comprise a very large 



proportion of tropical GraminEjE. 



** Articulation, when present, under the flowering glumes, but above the tivo outer 

 ones. Spikelets with one or more flowers and frequently a male flower or empty 

 glume above them, the male flower rarely below them. (Poace^j.) 



3. Phalaride^. Glumes normally 6, 2 below the articulation, 2 immediately 

 above it, usually small and empty or with a male flower in each, 2 upper nearly 

 equal, enclosing a single perfect flower. No true palea (the lower glumes occa- 

 sionally deficient) :— 1. Leersia ; 9. Alopecurus ; 6. Phalaris ; 7. Digraphis ; 



5. ANTHOXANTHUM ; 4. HlEROCHLOE. 



4. Streptatherji. Spikelets with 1, 2, or rarely more perfect flowers. Flowering 



