629 Jour., Bom- Nat. Hist. Soc, Vol. XXXII, No. 4 [May 31, 1928. 



Arabia felici lecta, probabiliter etiam varietas est Th. Forskalii meae ; sed 

 descriptio rem dubiam relinquit, specimina authentica desiderantur. Nee 

 hucusque ullam hujus generis speciem in Arabia felici lectam vidi, etsi proba- 

 bile est, Th. Forskalii varietates in Syria Abyssiniaque crescentes etiam in Arabia 

 inveniri. Itaque nomen specificum " triandra " tamquam dubium et rem indi- 

 cans quae in hoc et plerisque Graminearum generibus nullius est momenti, 

 seposui, " Forskalii " a Kunthio datum non minus quidem dubium, sed generis 

 auctorem commemorans recepi.' 



Hooker does not agree with Hackel, ' The earliest names of this plant are 

 Themeda triandra, Forsk., and Anthistiria imberbis, Retz. Hackel has abandon- 

 ed both, substituting first Anthistiria vulgaris, and then Themeda Forskalii, 

 on the ground that A. imberbis was perhaps not Forskahl's T. triandra (why 

 then call it Forskalii ?) of which no type specimen exists, and because triandra 

 indicates a character of no individual value in grasses. In so doing he over- 

 looked Gmelin's name of T. polygama (Syst. 149). Having regard to the wide 

 range of A. imberbis, from Australia to Africa, its presence in Arabia might 

 well be anticipated; and that it is a native of that country is now proved by 

 Schweinfurth's finding Hackel's var. glauca in that country. This makes the 

 var. (which is local, and not Indian) the type of the species, and if Forskahl's 

 name of Themeda is to be retained, necessitates a re-arrangement of the varieties. 

 To me it appears most expedient to retain Retz. 's name which applies to the 

 prevalent form over the area of distribution as the specific one. ' 



Hooker's reasons against Hackel's view are certainly valid, but his own 

 Anthistiria imberbis does not rest on a firmer foundation. The fact that the 

 var. is local and not Indian should not prevent us from making it the type of 

 the species, and the other circumstance that a re-arrangement of the varieties 

 will become necessary if the name Themeda is retained, can only be a reason of 

 expediency. 



So far we come to the conclusion that Hackel and Hooker deal with the same 

 material, but that neither name is satisfactory. 



We come now to the latest publication affecting our question. Stapf I.e. has 

 adopted the name Themeda triandra, Forsk. (1775) instead of T. Forskalii 

 (1889), ' as there is no doubt that the type of Forskal's species, which apparently 

 has been lost, was one of the forms covered by the description given by Stapf ' 

 (see Schweinfurth in Bull. Herb. Boiss. ii, App. II, 16). 



But here Stapf creates a new difficulty. His description applies only to the 

 ' African share of HackePs T. Forskalii '. In order not to be open to misstate- 

 ments we quote the whole passage in which Stapf explains his position : ' The 

 species, as defined here, is, however, taken in a sense somewhat narrower than 

 Hackel's; this restriction requires a short explanation. Hackel in his mono- 

 graph of the Andropogonecs distinguishes eleven varieties and as many sub- 

 varieties or forms within his T. Forskalii, whilst other authors have at various 

 times described more than a dozen species, all of which come within the com- 

 pass. There can be no doubt as to the close affinity of these forms and the 

 question of their status is mainly one of expediency. A careful examination of 

 the large amount of material at Kew and the British Museum has led to the 

 conclusion that for the present it will be most useful to detach, firstly, those 

 forms that are fairly uniform, and at the same time exclusive, over a large area ; 

 and secondly, those that, though of a limited range, stand out from the remain- 

 der by some character or characters. This leaves a residuum much less homo- 

 geneous than any of the segregates just referred to. It consists apparently of 

 more or less fixed races, mutants, hybrids and edaphic forms which from her- 

 berium material are the less separable because they are to a high degree inde- 

 pendent of geographical areas. At the same time, however, they are all 

 African with an extension into Arabia, Syria and the south-eastern corner of 

 Asia Minor, and, taken as a whole, represent practically the African share of 

 Hackel's T. Forskalii. It is to this aggregate that the description and synonymy 

 given above apply.' 



From the above it is evident that Stapf 's T. triandra, Forsk. is not identical 

 with Hackel's T. Forskalii and Hooker's A. imberbis, Retz., as it comprises 

 only the African element including ' an extension into Arabia, Syria, and the 

 south-eastern corner of Asia Minor.' Stapf 's synonymy leads to the same con- 

 clusion, except for the inclusion of T. imberbis, T. Cooke (Fl. Bombay ii, 993). 

 The Australian element Anthistiria australis, R. Br. has been separated by the 

 same author as Themeda australis, Stapf. It seems to us that Stapf 's treatment 



[8] 



