338 ^'^^ Philippine Journal of Science 1913 
staminate, while in those specimens presenting a more or less 
urceolate calyx, the flowers were invariably perfect. In no case 
has the two types of flowers been found on the same specimen. 
In the determination of species either by direct comparison or 
by an examination of descriptions, the calyx-character, as to 
shape, must be used with caution. 
There is also, in most of the species, a vegetative character 
that has apparently not previously been recorded, and that is 
the presence, on one or both surfaces of the leaf, of small, 
irregularly disposed, cystoliths or cystolith-like bodies. 
The Philippines are apparently a center of distribution for 
the genus, as indicated by the great number of forms discovered 
in the Archipelago. In many respects some of the species ap- 
pear to be more or less in a stage of transition, for intermediate 
forms frequently occur which it is difficult definitely to refer 
to any particular species. Characters as to venation, color and 
persistence of the indumentum, etc., are by no means constant. 
Consequently it has been found to be somewhat of a task to 
prepare an analytical key to the species that will always work. 
In using the key presented herewith exceptions are to be 
looked for in the arrangement of the basal nerves. In most of 
the species the leaves are very definitely 3- or 5-nerved, or 3- 
or 5-plinerved, but in some apparently both types of venation 
are found. In determining the number of primary basal nerves 
some diflScuIties are encountered due to the usual presence of a 
pair of slender, sub-marginal nerves, which may be rather dis- 
tant from the margins and comparatively strongly developed. 
The indumentum, while very characteristic and very persistent 
for many species, is in others of comparatively little significance, 
due to its deciduous or even fugacious character. In one stage 
of development the leaves may be uniformly lepidote and dis- 
tinctly colored on the lower surface, and at a later stage be 
quite glabrous and green. Even the inclusion of the same 
species under two different heads, in some cases, is not entirely 
satisfactory. I cannot, however, discover any other characters 
than those utilized in the construction of the present key that 
will serve the purpose better. 
1. Leaves definitely 7-nerved, up to 30 cm long, green beneath and somewhat 
lepidote with distinct, scattered, brown scales 1. A, apoensis 
1. Leaves definitely 5-nerved or 5-plinerved, rarely with a very faint sub- 
marginal pair added. 
2. Leaves definitely 5-nerved, mostly 15 to 20 cm long, pale-green on both 
surfaces when dry, with few, scattered, distinct, brown, lepidote 
scales. 
