TWO NEW CniKESE GRASSES'. 295 



ones of the latter genus.* Noi- is ^-iochn/s is much 1 etter (listinp;uishe(l 

 from Eritmthiis by its cartilaginous glumes and the muticous lower 

 glumella of the fertile floret ; the frei^nent suppression of the awn in 

 the closely allied genus Sorghum shows the slight value of this appen- 

 dage. The Andropogoneous genera have received special attention 

 from several eminent agrostologists ; Trinius,f Grisebaeh, J aadMunro,.§ 

 having each given analytical claves of their distinctive chaiacters. 

 Though differing inter se in details, and even in the piecise circum- 

 scription of the group, it is impossible, I think, for an unprejudiced 

 student of these valuable and instructive synopses not to see on what 

 excessively slight giounds many gen-era are still retained, differences- 

 of texture in the glumes and absence or presence of an arista being' 

 assigned too high a value. In Saccharnm apontaneum, Linn, (which I 

 cannot agree with Trinius as considering identical with 8. cBgypticuum, 

 Willd.), whereof I have examined specimens recently detected in 

 Kvvaug tung province, the lower tliird of the glume is quite as cartila- 

 ginous as in ^>;ocAry.si.s. |1 Trinius, indeed, combine-; Erianthus with 

 Saccharum, remarking — " Erianthus haud magis diff'ert a Saccharo 

 quara Andropogon ab Amtthero. Acicula adeo abbreviata in S. arundi- 

 naceo ut spiculai niutica) videantnr." tirisebach keeps up botli Erianthm 

 and Saccharnm, but then he also maintains Anatherum, distinguishing 

 it solely by the awn, when present, being sti^aight. Again, while the 

 continuous or jointed indorescence-rachis is employed by Trinius as the 

 special character for separating the AndropogonecB into two primary 

 divisions, Grisebach refuses to accord it even generic value. One- 

 must possess a foi robuste in order to maintain that these genera' 

 are of equal value throughout, or that the characters made use of are 

 assigned uniform importance in each particular instance ; and this I 

 regard as a very serious defect in the classification of a single group, in 

 itself undoubtedly natural. I presume no one will contest the state- 

 ment that no instance occurs in any Dicotyledonous order of dozens of 

 genera the lindts of which scarcely any two experts agree upon ; in 

 them the worst evil is that subgenera and sections are by analytical 

 bi)tanists too often raised to generic r.inlc ; whilst in (J-r.isse^ eacli par- 

 ticular author's genus in a multitude of cases is built up of " pickings 

 find stealings " {sit venia verbis) from those of another ; there are no 

 dednite or well-marked limits between them recognised by the con- 

 sensus of students of this order. 



2. PnyLLosxACHrs Nevinii, sp. nov. — Rhizomate procurrente 

 pallido 3'-4 lin. diauietro intervallis 9-12 liu. no<loso iibras crebras- 

 culmosque edenti, his tistulosis levissimis e viridido stramineis serai- 



* In E. A. Eemy's " Essai d'ane nouv. classif. des Graraioees " (Paris,. 

 1861), a somewhat pretentious but worthlei<s work, with a siugulaily arliiicial 

 arrangement of the genera, and which irretistibly suggests the notion of its 

 having been drawn up exclusively from books, and not from the actual study of 

 Grasses themselves, Eriochrysis is stationed between Pwtatheriim and Potta- 

 po'jon ! 



t Op. jam citat., 2'i3. 



+ Nachricht. k. Gotting-. Gesellsch., Feb., 1868, 88. 



I Op. cit., 431. 



II I find this had already been noted by Kunth in his careful analytical 

 description (Supplcm. Agrostogr. Synopt., 38o). 



