8 ORAMINEJE. 



kept in view is too indefinite to serve as a practical character; but 

 combining it with that proposed by Munro of the articuhition in 

 the axis of the spikelet being below the spikelet itself (in the pedi- 

 cel) in Panicaceae, and above the lowest glume or none in Poacese, 

 the exceptional forms are reduced to the lowest possible figure. 



" Kunth entirely gave up Brown's groups and divided the order 

 into thirteen tribes, many of which were natural, fairly defined by 

 a combination of characters, and have been very generally adopted. 

 He attached too much importance to such characters as the separa- 

 tion of the sexes or the increase in the number of stamens; in the 

 general arrangement his removal of the Andropogoneae to a dis- 

 tance from the Paniceas is disapproved of; and his describing flow- 

 ers as actually existing when only theoretically imagined is some- 

 times misleading. Nees generally adopted Kunth's tribes, but 

 improved the circumscription of some of them, and added two 

 or three small ones. 



"Fries, followed by Andersson, proposed for a jjrimary division 

 of Gramineae that into Clisa7ithecB, with the flower (i.e., the flower- 

 ing glume and palea) closed and the elongated styles protruding at 

 the apex, and EuryanthecB, with the glume and palea open at the 

 time of flowering and the short styles protruding laterally. This 

 division is practically useless, as the flowers of most species open 

 only for a very short time, and in dried specimens are almost always 

 closed; besides, the styles are usually slender and fugacious. The 

 long styles, moreover, would place the majority of the subtribe 

 Seslerieae, for instance, among Panicaceae, when all their other 

 characters are those of Poaceae. 



"Fournier rejects both Brown's and Fries's primary divisions, 

 but proposes a new one founded on the position of the lowest glume 

 of the spikelet next to the main axis in Chlorideae and Hordeaceae, 

 and averted from it or external in other tribes. But this rela- 

 tive position cannot well be ascertained in loosely paniculate Gra- 

 mineae, and in one-flowered spikelets it is often uncertain which is 

 to be regarded as the lower glume. The total number of glumes in 

 the tribe Paniceae is variable, two, three, or four; the lowest in 

 Reimaria, the highest in Panicum, and medium in Paspalum. All 



