344 NINETEENTH REPORT. 



tribute tliese genera to their three tribes and abolish the Chlorideae 

 altogether. It must be remembered that if this principle were carried 

 throughout the whole family the genera Paspalum and Syntherisma would 

 have to be removed from the Paniceae and placed in a separate tribe, a 

 procedure absolutely unthinkable. 



Of the remaining important tribes the Paniceae may be given first 

 mention. The spikelet here falls away as a whole. The spikelets are 

 mostly arranged in panicles (excepting Paspalum, Syntherisma, etc.). 

 The spikelet has one perfect flower, terminal to the rachilla with strongly 

 indurated lemma and palea, and one basal flower whicli is mostly either 

 staminate or represented only by an indurated lemma with or without a 

 palea. The two glumes are mostly membranaceous. We at once think 

 of the tendency towards reduction in the florets from the base upwards 

 exhibited by the Phalarideae and cannot but believe that the two tribes 

 may have had a common origin somewhere in the Festuceae, paralleling 

 some of the Agrostideae in such features as reduction of perfect florets 

 to one, induration of lemma and palea, dehiscence of rachilla below the 

 glumes, etc. 



The other tendency mentioned as shown by some genera of the 

 Agrostideae, that of having the glumes indurated and the lemma and 

 palea thin is paralleled by the Andropogoneae, a group distinct from but 

 clearly of common origin with the Paniceae and Phalarideae because of 

 the well developed terminal and often greatly reduced basal floret in 

 each spikelet. 



Tlie Maj'deae are obviously nothing but Andropogoneae modified by 

 having the monoecious or dioecious spikelets set apart in different parts 

 of the same inflorescence or in distinct inflorescences. In vegetative struc- 

 ture as well as in that of the grain the two tribes are practically indis- 

 tinguishable. Indeed, the grain structure of the Paniceae differs but 

 slightly from that of these tribes. 



The only important tribe not mentioned is the Oryzeae. The forms 

 ordinarily included in this tribe seem to be rather heterogeneous so that 

 it is well within the bounds of possibility that some of them should be 

 segregated into other groups. Taking the better known forms into con- 

 sideration we And that the flowers have usually three stigmas, more 

 often two whorls of tliree stamens each instead of the one whorl found 

 in most of the grass tribes, but one flower to a spikelet, the lemma and 

 palea usually quite similar and somewhat indurated, the glumes 

 membranaceous, often small or almost lacking. The tristigmatic pistil 

 and the two whorls of stamens show a degree of primitiveness shown only 

 by the Bambuseae. The structure of the caryopsis differs in many re- 

 spects from that of the otlier tribes so that it seems most reasonable to 



