CATALOGUE. 193 
mostly ascending; leaves from oblong to lanceolate or linear; flowers 
small for the section, 1-14’ long, usually clustered in the axils of the 
upper leaves, rarely few; bracts lance-linear; calyx-lobes linear, unequal, 
usually shorter than their sometimes cleft tube; lobes of the blue corolla 
acute, plaits bifid, anthers unconnected; sessile stigmas lanceolate; seeds 
narrowly, or sometimes more broadly, winged. 
Wet, grassy places in the Rocky Mountains of Colorado and Utah. 
The numerous thin seeds are borne on the whole inner surface of the cap- 
sule, which thus not only has the function of a placenta, but really seems 
to be nothing but a thin, membranaceous expansion of the placente them- 
selves, forming a free sac within the capsule, which originates from the 
commissures of the carpels, and remains attached to them only, and is at 
last otherwise entirely unconnected with the walls of the capsule. It is 
probable that all the Pnewmonanthes with ovules from the entire inner sur- 
face of the capsule have this structure, and that in the others the ordinary 
arrangement of commissural placentz prevails. 
GeNTIANA seRRATA, Gun. FI. Norveg. 10. Gray, Synops. 117. (@. 
detonsa, Rottb. Fries, Gray’s Manual ed. 5th, 387.)—Low, simple specimens, 
a few inches high, with single flowers, 1-14’ long. Mount Graham, Ariz. 
(751), at 9,000 feet altitude. 
The Norwegian specimens of this plant in my herbarium have much 
smaller flowers than ours and much smaller seeds. The “scales” which 
roughen the surface of the seeds prove, when moistened, to be transparent 
vesicles, single protruding cells of the epidermis. In the Norwegian form, 
these vesicles are small, oblong, or cylindric; in the American specimens, 
they are much larger and mostly hemispherical ; in the allied G. crinita, I 
find them large and oblong. 
GENTIANA BARBELLATA, Engelm. Trans. Acad. St. Louis, 2, 216, pl. 
11.—Has thus far been found only in the mountains of Colorado, near the 
timber-line, and is a very distinct species, which can in no way be con- 
founded with serrata or simplex. I have already, in the first account of this 
species, given the diagnostic characters, and have also stated that it is the 
only perennial one of the section Crossopetalum in America, somewhat allied 
13 BOT 
