502 MR. G. BEXTHAM Off CLASSIFICATION AND 



petal-like perianths of Coronariae, and the Juncacese never show 

 the spikelets of the Glumales. On the other hand, there are a 

 few species forming exceptions to the general character of the 

 Grlumales, although their general affinities require the including 

 them in the alliance. Thus in Centrolepidese there are often seve- 

 ral flowers within each glume, or the glumal arrangement is en- 

 tirely disturbed ; and in a very few Restiacese (chiefly Anariliria 

 and Lepyrodia) the glumes are shorter than the perianths and 

 scarcely or not at all imbricate. These exceptions will be further 

 considered after a few preliminary observations on terminology. 



Great confusion has arisen from the varied use made by different 

 botanists of the names given to the organs enclosing the essential, 

 parts of the flowers of Grlumales. These organs are sometimes 

 perianth-segments, but more frequently bracts or bracteoles, 

 performing the functions of very much reduced or absent peri- 

 anths. And there would be no absolute error in the uniform em- 

 ployment of the three terms, bracts, bracteoles, and perianth- 

 segments, whenever the distinction between the three is clear, and 

 of the more general terms scales or setse when it is as yet doubtful 

 to which of the three categories the organ belongs. But in ac- 

 cordance with a very general rule applicable to a variety of organs, 

 as the biological functions of these bracts or bracteoles increase 

 in importance, so also do their characters acquire fixity ; and there 

 is therefore great convenience, at least in the three large orders 

 Restiaceze, Cyperacese, and Graminese, in the general introduction 

 of the term glume for the primary bracts of the spikelet, and of 

 some special ones, such as perigynium, palea, lodicule, &c, for 

 special modifications of the secondary scales in particular groups. 

 It is, however, very essential for a correct understanding of the 

 floral structure, that, whenever possible, a definite meaning should 

 be given to each term used. 



In the three typical glumal orders the term bract might be used 

 only for those which subtend the spikelets or the branches of the 

 inflorescence when compound. These bracts are very general, 

 but not universal, in Restiacese and Cyperacese, very rare but 

 occasionally developed (as, for example, in Sesleria) in Graminece. 



Glumes, in their proper sense, are the imbricate scale-like bracts 

 inserted on the axis of the spikelet ; and, following out the views 

 of Mohl and others of the most distinguished among recent organo- 

 logists, the term should, in these three orders, be applied to all the 

 bracts so inserted, irrespectively of any modifications of form or 



