136 Rhodora [AUGUST 
readily separable from E. pilosa; the spikelets in the first two are 
ordinarily about 1.5 mm. wide; in the last, about 1.0 mm. In such 
closely allied species the length of the spikelet and the number of 
florets in a spikelet may not be expected to furnish very tangible points 
of difference. Furthermore considerable variation is found in differ- 
ent colonies of the same species, in different individuals of the same 
colony and in spikelets of different age upon the same plant.| How- 
ever, it can readily be shown that a tendency to increase of length and 
number of florets runs from E. peregrina through E. pilosa to E. 
Purshii. The spikelets in E. peregrina are characteristically well 
under 5. mm. (and rarely if ever over that length); in E. pilosa, some- 
what tending to exceed 5. mm.; in E. Purshii, rather frequently well 
over 5. mm. A similar ratio of increase occurs in the number of 
florets: in E. peregrina commonly under 10; in E. pilosa, occasionally ` 
over 10, up to about 12; in E. Purshii, frequently over 10, at times 
as many as 15. 
It is to be maintained that the very short pedicels of E. peregrina 
constitute an excellent diagnostic character, liable to very little if 
any confusion from supposed short pedicels in E. pilosa. In di- 
mensional characters of this kind it is commonly no difficult task to 
pick out individual cases that would seem to show the given measure- 
ments to be of little value. But in the present instance it may be 
confidently asserted that with experience this character will be recog- 
nized as of the greatest service. It immediately distinguishes FE. 
peregrina from E. Purshii, which latter species commonly has quite 
long pedicels — in fact from several to many times longer. 
The character of smooth empty glumes may be reapportioned 
among the three species. They are characteristically quite smooth 
in Æ. peregrina but Professor Wiegand further notes that in E. pilosa 
they are usually but not always scabrous on the keel. This latter 
statement is to be connected with the inclusion of E. Purshw in E. 
1 The spikelets of all three species are subject to a notable reduction in length and number 
of florets in the later panicles. This seems to be most marked in E. peregrina. When 
bearing short spikelets of rather few florets E. peregrina will appear to almost assume the 
key-characters commonly used for E. Frankii. In fact, in general appearance — size of 
plant, habit of growth, many-flowered panicle with stiff, spreadiog branches —it often 
bears a greater likeness to this species than to either E. pilosa or E. Purshii. When the ripe 
grains are protruding from the spreading scales, in plants showing this marked reduction, 
E. Frankii is so strongly simulated that at a distance it is no slight task to distinguish the 
two with accuracy. Professor T. C. Porter, as shown by his own determinations, consistently 
referred the plant to E. Frankii. 
