224 BG. GRAMINA. 



loliacea. The diagnosis fonnerly attempted to be drawn, 

 by contrasting tlie branched inflorescence (panicle) of F. 

 pratensis against the simple inflorescence (" spiked ra- 

 ceme ") of F. loliacea, is utterly useless ; because every 

 the most gradual step from the one form to the other, 

 may be traced between them in many different localities ; 

 and I have seen the two forms of inflorescence distinctly 

 characterized, along with intermediate and less decided 

 links, on one single root. But against the evidence of 

 identity derived from this transition of characters, it has 

 been said that the outer glume of F. loHacea is 5 -nerved, 

 while that of F. pratensis is 3 -nerved. Any supposed 

 force in this sort of argument is met and neutralized by 

 one of the intermediate Imks above mentioned, and now 

 in my herbarium. Coimting downwards from the top it 

 has thirteen nearly sessile spikelets, on short simple pe- 

 dicels, the fom-teenth is represented by one similar almost 

 sessile spilielet and a second elongated pedicel bearing 

 two spikelets. The fifteenth, which is also the lowest, is 

 again a single and shortly stalked spikelet. The glumes 

 on the simple pedicels are j'i^-e-nerved ; on the branched 

 pedicel they are three-nevved ; that of the lowest solitary 

 spikelet is spuriously 4-nerved, in consequence of one of 

 the lateral nerves being almost obsolete. This individual 

 specimen was taken from a wild root of F. loliacea trans- 

 planted into my garden, and which afterwards produced 

 stems with both panicled and racemed inflorescence, si- 

 multaneously with the intermediate link thus particularly 

 mentioned. Even on one of the racemed stems some 

 glumes are 3-nerved, others 5-nerved. 



