PAIRED SPIKELETS 85 



the spikelets removed (Fig. 77, C) and note that the 

 rachis joints (the thicker parts) form a central axis 

 with the pedicels borne on alternate sides, just as they 

 are in Holcus (Fig. 73, C). In this raceme we have 

 just the reverse of the arrangement of Manisuris. In 

 that, the fertile spikelet fits into a cavity formed 

 by the rachis joint and pedicel w^hile in Rytilix the 

 rachis joint and pedicel fit into a cavity formed by 

 the fertile spikelet. (See Fig. 77, D, showing the 

 inside face of the first glmne of the perfect spikelet, 

 and examine the two views of the joints, A and B, 

 and the two views of the raceme, E and F). The 

 sterile spikelet is well developed but usually contains 

 no flower. The perfect spikelet with its peculiar 

 gray, ridged and pitted, subglobose first glume sug- 

 gests the achene of some species of Scleria (a sedge). 

 The whole inflorescence is very unlike that of any 

 other known grass. 



Return to Fig. 73, A, and compare with, it Fig. 

 78, A {Rhaphis pauciflora), which shows a raceme 

 reduced to a single joint consisting of the sessile 

 perfect spikelet and two pediceled sterile spikelets, 

 the rachis joint (found in Holcus and the others 

 studied) often being replaced by a second pedicel and 

 sterile spikelet. This second sterile spikelet is not 

 always developed. In Rhaphis this 1 -jointed raceme 

 has a sharp hairy callus at the base. Fig, 78, B (the 

 base of a raceme and the summit of a branch, the 

 hairs removed), shows the source of this callus. The 

 racemes are borne on the long slender branches of a 



