312 GRAMINEE (Stapf). 
po light in a drop of water. Where the orig conditions of the florets 
of importance, it at ie kept in mind, that many grasses are very distinctly 
r ice. th i i 
expand. Such flowers anni frequent wee ta kew to be rege whilst they were 
actually hermaphrodite. To avoid this error, young spikelets should, if possible, 
be examined beside the fully ‘developed os or the filaments which usually 
remain “eeoant the ovary s elie be sou 
{In the sequence of gene r. Stapf oe found it woah to ae ey — 
adopted in Bentham = Tooke’ s Genera Plant ie as well as e Flor 
Au straKensis, and in Sir J — Hooker’s enumeration of India n grass rece otitly 
ag ed in the Flora of B rvitish India. In the — of ‘colonia floras issued from 
it has been thought Me ee for the most part void a ve as possible 
breaking new ground and to adhere closely to a aentra joeenndt eds: sequence 
e 
ed 
comparable n the present case, however, further research has shown some 
nea - be indispensable. To these oe Joseph Hooker, who has kindly allowed 
me to consult him, now assents. In accordance with general practice, Andropo- 
gonee pret at the i of the endae. ‘wht le a different position is assigned to 
Zoysize and to Oryzeew.—W. T. T. D.] 
SERIES I, Mature pilot falling entire from their pedicels or with them 
(rarely subpersistent on t, indistinctly and tardily disarticulating rhachis: 
= etna all “ee or sir . in sex and structure ; bay t spikelets with 
s e up 
hetero 
tinued ee nd the upper 
Spikelets falling entire, er or in clusters, occur in the following genera 
belonging to the second series: Holcus, Chetobromus , Polypogon, Perotis, Tragus, 
Prive Spartina, Lamar rekia, Fingerhuthia, Urochlena, Tetrachne, 
n 
ANDROPOGONEZ. Spikelets usually in pairs, one consgan the 
othr ne ie Trachypogon), rarely 3-nate or solitary on the of 
ously arr i ron eo sili, racemes. (/wmes more or less rigid and iret 
te the valves, and the lower always longer than the florets. Valve em- 
_— often Teatins, tbat of the upper ford generally awned or see to 
wh, 
* pure ee all alike. Racemes in compound cate (except Pollinia) ; joints 
the rhachis slender 
i ec 1. SaccHarEs. 
I. Imperata.—Racemes in a spiciform silky panicle; rhachis not fragile. 
Spite, iit niess. c 
II, Saccharum.—Racemes in large much branched silky panicles; rhachis 
fragile. Spikelets senile. 
IIT. Erianthus Rianne in large much branched silky panicles ; rhachis 
fragile. Spiketets awne 
IV. Pollinia, barnes 2- -nate, digitate or approximate on a short main axis. 
¥*® The ve spikelets of each pair differing in sex and structure (in the Sou th 
African species, except in Ischemum fasciculatwm, where the difference 1s 
lianited to the lower glume). 
Subtribe 2. Iscnemex. Sessile pesca i ra in hollows. Racemes ana 
for olitary, thats », digitat e or appro on a short main axis. Low 
floret "of the sessile spikelet always g. 
V. Ischemum. The sis South African genus. 
te $ Ror £. Sessile spikelets sunk in hollows formed by th 
3 joi oint of t the rhachis 9 nd the appressed or adnate pedicel of the 
pied spikelet of the same pair. 
. Rottbellia.—All s a ets awnles 
ba Urelytrum, —Pedicelled spikelatr' long awned, 
e 
pees 
SSS Pt re 
