As far as my dissections show, the functionally female flowers usually develop first and for this reason the length given for the filaments in the male flowers may sometimes be too short owing to their immaturity. The glomerule or cymule usually (or at least often) consists of one female flower and several male ones.The taxonomy of the species of this genus is unusually difficult as nearly all the species appear to hybridize (though I know of no experimental evidence of this), the leaflets are very variable in shape and indumentum and the flowers provide few characters of value in classification. It is difficult to avoid using the character of “branched” or “unbranched” inflorescences in the key and in good specimens this is usually not difficult to determine; but poor or depauperated specimens, probably normally “branched”, may not show the character, while species with normally “unbranched” inflorescences may occasionally show a small branch. The inflorescences are never truly simple as the flowers are grouped in small subsessile glomerules or shortly pedunculate cymules along the axis or axes and this is technically branching.The classification here proposed is still far from satisfactory but only experimental work can disentangle the taxonomy. The fruits may well provide useful characters but so relatively few fruiting specimens have been collected that there is little point in using fruit-characters in the key.