A NEW TRIBE OF STERCULIACEAE. 257 
unfairly be esteemed subgeneric. When to differences that may 
be rated as subgeneric we add a third, affecting yet another 
whorl of floral organs, we are induced to conclude that the sum 
of these differences is of generic value. 
The leading points of agreement between Triplochiton and 
Mansonia are the deciduous calyx, the contorted-imbrieate petals, 
the hypogynous insertion of the free filaments at the apex of 
a pronounced gynandrophore, the existence of a whorl of 
hypogynous free staminodes between the filaments and the 
gynecium, and the free pluriovulate earpels, some of which sub- 
sequently develop into dry one-seeded mericarps samaroidly 
winged on the back. 
In Triplochiton the calyx does not split to the base between 
the lobes before the organ as a whole separates from the torus; 
it therefore slips down and for a time persists as a loose collar 
on the pedicel. In Mansonia the calyx splits to the base along 
one side before the organ as a whole separates from the torus ; 
it therefore at length falls quite away. The different appearance 
that results is thus accidental; in both genera the calyx as a 
whole ultimately separates from the torus. The scars on the 
torus which mark the points of insertion of the petals are small 
and nearly circular in Triplochiton, but elongated aud oblique in 
Mansonia; this difference, however, is again accidental, and is 
due to the petals being clawed in the former genus, sessile in the 
latter ; the essential feature is that the petals in both are con- 
torted-imbricate. As regards both calyx and corolla, therefore, 
the two genera fall naturally within the cohort Malvales, without 
any clear indication as to which natural order of the cohort best 
accommodates them. The pronounced gynandrophore or column 
which intervenes between the petals and the stamens makes it, 
however, advisable to exclude the genera from Malvaceae, in spite 
of the fact that, at all events in JMansonia, the anthers are 
l-locular. They are said by Schumann to be so in one species of 
Triplochiton as well, and although in another species of Triplo- 
chiton they have beenfound by C. H. Wright* to be very peculiarly 
2-locular, the writer cannot find, in a third species of Triplochiton 
from Lagos which he has examined, that the anthers have two 
cells. But, although the presence of 1-celled anthers in a 
Malval as a rule indicates that the plant is Malvaceous and not 
Sterculiaceous or Tiliaceous, the character is subject to too many 
* Hooker, Icon. Plant. t. 2758. 
