PITTIER—MIDDLE AMERICAN SPECIES OF LONCHOCARPUS. 39 
There is some measure of uncertainty as to the_value of the charac- 
ters of the base of the staminal tube and the freedom or coalescence 
of the vexillar stamen in separating Lonchocarpus from the mentioned 
genera of the Galegeae. As indicated above, Kunth considered the 
stamens of the former as diadelphous, a view which is hardly justified 
by the facts. The elder de Candolle! has already shown that the 
stamens are often monadelphous, a statement which can now be 
generalized. The extreme basal ring of the tube is continuous up 
to the time when the fecundated ovary begins to swell. On the 
upper side of the tube, just above the basal ring, there are two open- 
ings, oval in most species, V-shaped in a few, and separated by the 
vexillar stamen, which thus appears to be free for a very short dis- 
tance. Farther toward the anthers the tube is completely closed, 
all the filaments becoming free at about the same distance from the 
anthers. The basal openings, or “‘fenestrelle,” have a distinctly 
thickened, callous margin, and the corresponding part of the tube is 
more or less dilated laterally. 
In the flowers of Robinia and Gliricidia also, the staminal tube is 
often found to be fenestrate, but the margins of the openings seem 
to be always thin and the adherence of the vexillar stamen above the 
latter less marked in Robinia, the tube being almost always split, 
and wanting in Gliricidia, where that stamen is entirely freed at an 
early stage. In Lonchocarpus eriophyllus Benth., the generic posi- 
tion of which will remain in doubt until the fruit is found, the thick- 
ening of the margins of the fenestrelle is hardly marked, and the 
tube is either split or closed. But in other cases, as in Lonchocarpus 
meistophyllus Donn. Smith, which I have transferred to Gliricidia, 
the tube is thin everywhere and the vexillar stamen either completely 
or partly free. 
In Lonchocarpus, as shown by de Candolle in the memoir cited, 
the style is usually glabrous. This is also the case in Gliricidia, but 
in Robinia we find it covered with stiff hairs. Considered alone, 
this is consequently a character of little or no value. 
The number of ovules varies in Lonchocarpus from 1 to 9. In 
only 3 of the Middle American species is the ovary 8 or 9-ovulate; 
36 species have from 3 to 8 ovules, and 2 others 1 or 2. These 
numbers are not exceeded in the other species of the genus, and this 
is another reason why L. eriophyllus, with its 11 or 12 ovules, should 
be held generically doubtful. There is also variation in the adher- 
ence of the wings to the keel, and of the carinal petals to each other, 
which is also far from being an absolute generic character, since it 
is found to a greater or less degree in several arborescent or shrubby 
genera of both the Dalbergieae and Galegeae. 
1 DC. Mém. Legum. 278. 1825. 
