-i 
species free from near its base,'in others adnate up to the base 
of the lobes (an important character which has often been neg¬ 
lected) ; the outer stamens are usually placed a little higher than 
the inner ones. In the bud the filaments are always doubled up,* 
geniculate, and straighten out when the flower opens, and almost 
always become much longer than the perigon ; in a few species 
they do not exceed, the length of the lobes. The filaments are 
generally attenuated from a broader base and terminate in a thin 
point, on which they bear the large and conspicuous linear,f near¬ 
ly quadrangular, somewhat introrse, 4-celled, versatile anther, 
attached near or a little below the middle ; in A. Virginica the 
filaments are thickened upwards, almost clavate. The globose, 
or elliptic, delicately reticulated pollen-cells have, on an average, 
a diameter of 0.06 to mostly 0.08 or even o.n mm. 
The ovary consists of three carpels, opposite the outer perigo- 
nal lobes, forming three cells, in each of which two vertical rows 
of flat, horizontal, anatropous ovules spring from the central pla¬ 
centas. The stout, somewhat triangular, tubular style rises to the 
height of the anthers and sometimes above them, but its length 
is variable and does not seem to be always characteristic. The 
stigmatic part is thickened, clavate, or somewhat capitate, and 
is divided into three carinalj: lobes, which at last open some¬ 
what, or especially in the first section, expand horizontally, and 
are often emarginate or even obcdrdate ; after expansion they (at 
least in A. Virginica) exude a viscid liquid—whether stigmatic, 
or only intended to allure insects, has not been ascertained. 
The flowers of Agave—I speak particularly of A. Virginica , 
the only one I have been able to observe in its development, but 
I suspect that the same holds good in all the species—are ves¬ 
pertine or nocturnal and proterandrous. They open late in the 
afternoon or in the evening, and, while the filaments straighten 
out and elongate, the anther-cells burst and emit the large pollen 
grains, and on the following morning are found withering and 
empty. The style at this period usually does not yet exceed 
the perigon (in A. maculosa it is much shorter), and its lobes are 
* Even the short filaments of A. maculosa are thus geniculate. (See p. 301, note.) 
f The curved anthers spoken of in some descriptions can only refer to effete and with- 
ing ones. 
t See page 293, note. 
