12 



THE TRUE GRASSES. 



bracts (Anthoxanthum, Oryza, Andropogonece). However, 

 it is possible to demonstrate a perfect series of stages by 

 which one-flowered spikelets with a distinctly lateral 

 flower pass over into those with terminal flowers ; con- 

 sequently a palea and flowering glume can be distin- 

 guished in the latter by analogy. 



The uppermost glumes of a spikelet are not infre- 

 quently empty, and in this case they are usually dwarfed 



h ^ 



Fig. 4.— I, Diagram of a many-flowered spikelet of Arena, Poa, Triticiun; A, axis 

 (rachilla). II, Diagram of a flower of Bambusa with flowering glume and palea; 

 Ii, posterior lodicule. Ill- VII, Diagrams of an entire spikelet; A, 1^ . etc., empty 

 glumes; d. flowering glumes; v, palea; /, anterior lodicules. Ill, Streptochceta; 

 A, B-F, small outer bracts; a-f, large inner bracts, the latter distinctly in two 

 whorls. IV, Andropogon; V, Coleanthus; VI, Oryza; VII, Anthoxanthum. 



or anomalous (Melica, Fig. 80). The empty glumes are 

 lacking (Coleanthus), or scarcely visible as rudiments 

 (Leersia), in only a few genera ; there is rarely but one, 

 most frequently there are two, and more rarely more 

 than two, present. In the latter case those following the 

 second are described by most authors as sterile flowers 

 (e.g., in Panicum, Andropogonece, and many Bambusece), 

 a plan which we shall not adopt in our descriptions. It 

 is true that in closely related species (of Panicum, for 



