216 Rhodora [DECEMBER 
and that the uppermost, lemma is sterile or empty — in which group 
both Festuca and Bromus would be found. Those in which two or 
more imperfect florets exist might be further subdivided into 
(a) Lemmas 1-3-nerved, including Eragrostis, Sphenopholis, Koeleria, 
and Catabrosa. 3 
(b) Lemmas 3-5-many-nerved, including Diarrhena and Melica, as 
well as the South American Anthochloa, the African Harpachne, 
and the Australian Ectrosia and Heterachne. 
It must not be overlooked that occasionally specimens of Festuca or 
Bromus are found with more than one empty lemma; but these are 
the exception rather than the rule. Tribal and generic differences 
must be based on prevailing rather than universal characters, and there 
can be no doubt that the single empty lemma predominates in Festuca 
and Bromus, just as in Melica and its allies the existence of more than 
one imperfect floret is fairly constant. 
If these characters be taken as the basis of our dichotomy, the sub- 
genus Bromelica may still remain in Melica, because it agrees in having 
uniformly more than one imperfect floret; and the membranous tex- 
ture and more remote arrangement of the upper lemmas become 
characters of subgeneric rank. The nearest ally of Melica among 
American grasses would then not be Bromus or Festuca, but Diarrhena, 
from which it would be easily distinguished by its three stamens 
(Diarrhena having 2 or rarely 1) and 1 lodicule (Diarrhena having 2). 
Melica would further be distinguished from the other genera with 
more than one imperfect floret as follows: Heterachne and Harpachne 
have lemmas and glumes keeled, Anthochloa has the lemmas fan-shaped, 
Lophatherum and Ectrosia have the sterile lemmas awned. 
But the attempt to distinguish Melica on texture and arrangement 
of the upper lemmas alone will, as Mr. Farwell has pointed out, never 
be satisfactory. I am inclined to think, however, that the problem is 
not to be solved by a segregation of the genus, but by an attempt to 
find a different set of characters on which to base the delimitation; 
and this cannot be done until we consider the genus in wider relations 
than those afforded by its North American allies. 
SALEM, OREGON. 
