261 
odpiiaeed or spreading with short points. Palea an almost micro-~ 
scopic hyaline 2-lobed scale. Upper (fertile) flor ve Floral glume 
(valve) lanceolate, acute, almost as long as that of the lower floret, 
thin-papery, often tinged with purplish-blue, delicately 5-nerved, at 
length somewhat firmer. Palea similar in sha ape and structure to 
a floral glume, “tr Betis tc) 2-nerved. Lodicules siete atte 
"15 mm. long. Stamens 3, anthers linears, yellow, abou 
sig Ovary rAlate ; ‘styles capillary, up to "12 mm, oe 
stigmas esha from the apex of the floret, purple 0°75 mm. long. 
Grain unkno 
Distribution ie specimens examined :— 
Mascarenes. Mauritius, Bory de St. Vincent, Dupetit Thouars 
e ripe in the British Museum, from illdenow’s We Bi a 
r, Fl. Maurit. i., 42, Bouton, Ayres ; Réunion,* Balfour 
pee ee Central Madagascar, Baron, 4253 ; ‘South 
Madagascar, Fort Dauphin, “prostrate, matted, sa ndy dunes 
Scott Elliot ; without precise locality, Sere 82. 
Tonkin. Tankeuin, near Guang-yen, “croissant en touffes 
compactes ...... sur les rochers calcaires,” Balansa, 460. 
AUSTRALIA. New South Wales, “Port J ckaon.” Sieber, 
Agrostoth., 72 ; Sydney, Maiden. 
_ The affinity of the grass lies evidently with D. sanguinalis, L., 
from which it differs, however, sufficiently in its perennial duration 
and peculiar mode of growth, the fine blades, the small number of 
racemes and their slender build, the smaller spikelets and the evenly 
grooved meter of the elume of the lower floret, at least when 
To this might be added as a secondary anatomical character 
‘the smaller size of the wavy-walled cells of the glume of the fertile 
floret and their smaller number of folds (3-4) against those of D. 
sanguinalis (6-7). Although I have examined all the specimens 
mentioned above for grains, I have not found a single one, even in 
racemes which had already shed most of their spikelets and were to 
all appearances perfectly mature. 
* “ Dans les champs incultes,” Cordemoy, Flore de I’Ile de la Réunion, p. 117. 
