288 SCHOMBURGK'S GUIANA PLANTS. 
has been placed upon the number of parts in each floral or 
carpellary verticil. In some cases, indeed, it is still the only 
tangible character by which very natural genera can be dis- 
tinguished, while in others it has already been admitted as — 
uncertain by all the above quoted authors. There aresome — 
instances, however, where both De Candolle and Martius — 
appear to me to have on this account only, broken up affinities — 
really very close. Martius has, it is true, united Arthrostemma 
with Chetogastra, Sagrea with Clidemia, &c.; but on the 
other hand he has separated Noterophila from the herbaceous 
Microlicie, which he expressly states are not to be distinguished 
by any other character but the number of cells of the cap- 
sule, their habit being precisely the same. m 
It will also be seen by some of the modifications proposed 
below, that I should be disposed to go even farther than 
Martius in lowering the value set upon the appendages to the- 
calyx; whether external, of the nature of hairs, as in Osbeckiew; 
or arising from the internal development of the margin of 
the sepals beyond the midrib, as in Miconiee. bi 
TRIBE I. OssECcKIEX. mnm: 
The baccate genera in this tribe, Otanthera, Melastoma, 
and Tristemma, are not American ; Aciotis being apparently 
founded on a mistake. The capsular species of the old 
world are at present included in Osbeckia, to which Lachno- 
podium has lately been added, though with some doubt. The. 
American capsular genera may be reduced to four, Pi Jeromo, é 
Tibouchina, Chetogastra, and Macairea. 
Pleroma of Don, is evidently the same genus as 
Poe 
t capsular 
