PITTIER—MIDDLE AMERICAN SPECIES OF LONCHOCARPUS., 45 
affinities. But it has also marked deficiencies. In the first place, 
the modification of the vexillar margin of the legume, which is funda- 
mental to the distinction of the two subgenera, seems to be only a 
secondary character, not bound to any other important modification 
in the structure of the flowers or of the leaves, so that it is hardly 
possible to place a species in the right subgenus in the absence of the 
fruit. In a general way, however, it can be said that all species with 
small leaflets belong to Neuroscapha, and all species with pellucid- 
punctate leaflets are included in Eulonchocarpus. 
The above system could easily be extended so as to include the 
other American species, by elevating the sections Paniculati and Fasci- 
culati to subgeneric rank.!. The species of the section Laxiflori would 
be included under Eulonchocarpus. 
DISTRIBUTION OF THE SPECIES. 
DISPERSION OUTSIDE OF MIDDLE AMERICA, 
The genus Lonchocarpus is distributed over tropical America as 
well as tropical and austral Africa. A few species are found also in 
Madagascar and Australia. Lonchocarpus sericeus (Poir.) H. B. K. is 
the only species common to America and the West Coast of Africa. It 
seems to be also the only representative of the subgenus Neuroscapha 
reaching beyond the limits of America. Eulonchocarpus is entirely 
neotropical; all the African and Australian species but the one 
mentioned above belong either to the section Paniculati or the section 
Fasciculati, or to recently established groups. 
1 The section Fasciculati Benth. forms a natural group, characterized by having 
the flowers fasciculate on thick peduncles branching from the common rachis, by the 
peculiar shape of the wings, often reflexed and shrivelled at the later stages of the 
flowers, and by the leaflets, the costa and veins of which are neatly reticulate and 
very prominent beneath. These characters appear sufficient to justify the eleva- 
tion of the section to the rank of a subgenus, for which the name Phacelanthus is here 
proposed. 
In all species the pistil is also slender, with a moderately arcuate style, which is either 
glabrous (L. nitidulus, L. glabrescens) or more or less hairy. Among the species I have 
seen, the ovary is 2-ovulate in L. negrensis and L. floribundus and 8 to 10-ovulate in 
L. spruceanus and L. glabrescens. Lonchocarpus costatus is said to have 6 to 8 ovules, 
while L. variflorus and L. nitidulus are species with 2-ovulate ovaries. Lonchocarpus 
boliviensis, the fruit of which is not known, while closely related to several of the above- 
named species, constitutes a remarkable exception in the genus on account of its hav- 
ing 16 to 18 ovules. It is described for the first time in the present paper (p. 93). 
The Mexican tree or shrub described by Bentham as L. eriophyllus has only 11 or 12 
ovules, but it differs from Lonchocarpus by several other important characters (p. 85). 
We have seen that in the Middle American species the maximum number of ovules 
is9. In South America, besides the 2 species mentioned above, LZ. denudatus Benth.., 
of the section Laxiflori, is also known to have 10 ovules, all the other species showing 
a smaller number. 
79566—17 2 
