210 



back, nerveless between the fine, minutely scabrid keels, upper 

 boat-shaped, 3-nerved, shortly aristulate, margins minutely ciliate 

 upwards ; lower valve almost lanceolate, acute, 6 mm. long, 

 hyaline, 1-nerved, minutely ciliate, upper lanceolate, 5 mm. long, 

 deeply bifid, 1-nerved, awn about 15 mm. long, kneed at or below 

 the middle, glabrous ; palea ovate, nerveless, 1*5 mm. long, glab- 

 rous ; anthers 4 mm. long; stigmas 2*5 mm. long. Pedicelled 

 spikelets male, linear-lanceolate, 7 mm. long, glabrous, purplish ; 

 glumes thin, lower shortly aristulate with 3 stronger and 3-1 very 

 fine nerves, ciliate, upper hyaline, 3-nerved, ciliate, 6 mm. long ; 

 valves hyaline, 1-nerved, 6-5 mm. long. 



Vern. Burr (Appleton, Drake-Brockman). 



Without precise locality, Drake- Brock man, 43, 44. " Dur : grows 

 to b or 8 feet in great feathery bunches. Very plentiful in the 

 Hand and at Bohotle. A valuable camel feed."— Appleton. To 

 this species refers possibly, at least in part, another note, written 



i.- ° 1 ' A PP leton against a specimen of Sporobolus fruticulosus, 

 which by mistake had been mounted with his specimen of " Durr." 

 It runs "Grows thickly in most parts of Italian Somaliland 

 (between Obbia and Bohotle), also less freely in the Haud and 

 British Somaliland. After rain it forms large feathery bunches 

 ot green In the dry season the leaves and young shoots fall off, 

 leaving the stubble on which our horses grazed until the rain 

 came. The best grass in Somaliland." The specimen of the 

 bporobolus referred to resembles a dwarfed "Durr," but the 

 anatomy of the leaves and the presence of a few withered 

 inflorescences are sufficient to recognise it. " Tall Durr " is also 

 mentioned in Aylmer's map (.see p. 205) as occurring in patches in 

 the Haud, east of Toyo. • 



Hackel placed the closely allied A. Kelleri in the section 

 bcnizachyrmm, where it was supposed to occupy an isolated 

 position. He was probably led to do so by the solitary racemes, 

 lhe same condition is characteristic of .4. Bentii, from Socotra, 

 which I describe at the end of the paper. The close affinity of 

 the three species A. cyrtocladus, A. Kelleri and A. Bentii is, 

 however evident from their general appearance as well as from 

 the structure of their racemes and spikelets. The latter refers 

 them unmistakeably to the section Arthrolonhis, within which 

 mey torm a small group of well-marked habit, approaching in 

 meir technical characters more the American A. glaucescens, 

 iiunth, and A. mcanus, Hack., than any of their African con- 

 geners. Among the latter A. appendiculatus, Nees, has probably 

 most in common with them. 



fi n,v;i 8Chae T m ' h Var ' W somalensis, Map/, var. nov. ; a typo 

 Er Pa ? , mi ™ ribn8 (3 mm. longis), articulis 2 mm. longis, 

 spicuiis pedicellatis valva superiore destitutis neutris distinctus. 



JUt\ B "™°™ d T U PP er Sheikh, grows two feet high, and is a 

 good feed."— Appleton. B ' 



The specimens consist only of a few culm top* and inflorescences, 

 lhe racemes are on the whole so similar to those of A. Ischaemum 

 (var .genumnm, Hack.) that I hesitate to base a new species on 

 the differences which I have indicated. Yet it is quite possible 

 mat it will have to be treated as specifically distinct when 



