114 Graminece. 



split to the base, rarely entire ; infl. of spicate * racemed, 

 capitate, or panicled spikelets, formed of 3 or more alt. dis- 

 tichous bracts (glumes), two lowest glumes almost invariably 

 empty, one or more following i-fld., if more than one, 

 all are inserted on an axis (rhachilla) ; opposite to each fig. 

 glume, and often to empty ones, is a 2-veined usually mem- 

 branous scale (palea) with often inflexed sides (flaps) ; peri- 

 anth o, or of 2, rarely more, minute scales (lodicules) ; stam. 3, 

 rarely 1-2 or many, hypogynous, one at the base of the fig. 

 glume, one opposite each vein of the palea, fil. capillary, anth. 

 of 2 parallel cells, connective obsolete, pollen globose ; ov. 

 1 -celled, styles usually 2, terminated by feathery or peni- 

 cillate stigmas, ovule erect, anatropous; fr. (grain) a seed 

 enclosed in and usually adnate to a membranous pericarp 

 which is rarely loose, or coriaceous, or (in some Bambusece) 

 fleshy; seed erect, hilum posticous, punctiform orbicular or 

 linear, embryo anticous, at the base of but outside the 

 copious floury albumen, sometimes half as long as the seed, 

 cotyledon large, shield-like (scutellum), dorsally adnate to the 

 albumen, plumule and radicle small. 



The following Key to the tribes and genera of Sinhalese grasses 

 is rather a compromise adapted to the wants of a colonist than a strictly 

 scientific co-ordination. In drawing it up I have adhered pretty closely 

 to that given for the whole Order by Bentham (Gen. Plant.), as slightly 

 modified in the Flora of British India. The few important changes 

 (indicated where they occur) are chiefly due to the later revision of some 

 of the tribes by Dr. Stapf, prepared for the Flora Capensis (ined.). I 

 have not, however, adopted all the views of that able student of grasses, 

 deeming that some of them want further consideration. It must be 

 allowed, I think, that there is no more difficult problem in the classifica- 

 tion of the genera of any large order of phanerogams than that which the 

 grasses present. The primary divisions of Panicacece and Poacece is a 

 very unsatisfactory one, though founded primarily on such apparently 

 important characters as the relative position of the fertile flowers on their 

 axis of growth (the rhachilla), and that of the spikelets being articulate 

 or not at the base, whereupon depends in a great degree the dispersion 

 of the seeds. Of the recognised tribes, some are more or less artificial, 

 containing genera of doubtful affinity ; nor, as it appears to me, would 

 it be difficult to multiply the tribes indefinitely, by giving tribal value to 

 anomalous genera. On the other hand, the genera themselves are, on 

 the whole, well circumscribed. Though some are divisible into sections 



* The term ' spike ' is loosely used in this Order, or is more or less 

 conventional. In some genera with contracted infl., the pedicels are so 

 short that the spikelets are sessile or subsessile, when the term is 

 legitimate ; in others, where the spikelets are binate, a sessile and a 

 pedicelled, the terms 'spikes' and 'raceme' are interchangeable. 



