308 
ROBINSON. 
leaves have great similarity to those of E. carinoi, except that in E. banahaense 
the base of the narrower side of the lamina ends some distance from the petiole, 
while on the other side it is usually at least subauriculate, But there are 
intermediates to such an extent that if there were no other character, the 
species could not be held separate. That character seems to be supplied by the 
stipules, which are longer and usually wider in E. banahacnse, their length 
reaching 25 mm on Merrill 7516, and ranging from 14 to 22 mm on the College 
of Agriculture collection, while they are from 10 mm (young) to 16 mm on 
the type. The Zamboanga collections cited in the description of E. banahaense 
also show this character, as do two numbers from Mount Pulog, Benguet, 
For. Bur. 16047, 16051 Curran, Merritt, & Zsckokke, originally identified as 
E. carinoi and responsible for part of the variation recorded for that species 
with regard to this very character, although by some oversight they were not 
cited. All of these should be referred to E. banahaense. The stipules of E. 
carinoi and E. variabile are usually narrower, and vary in length from 3.5 
to 11 mm, slightly longer on the E. variabile type than on the other, but are 
usually from 4.5 to 7 mm, in contrast to a normal 15 to over 20 mm on E. 
banahaense. Taken ill conjunction with the other characters, insufficient as 
the latter alone would be, E. banahaense and E. carinoi are better held distinct. 
This amended view of the affinities of E. banahaense also brings it very close 
to E. edulc, which has wider leaves, less acuminate, a still more pronounced 
basal auricle, and stronger venation. It will not be surprising if future collec- 
tions render E. edule untenable as a separate species. As it is, the Mount 
Mayon plant cited under E. edule, is rather E. banahaense. 
Vanoverbergh 828, the type of E. variabile, was pistillate, and on its general 
appearance, no close . alliance with E '. carinoi would be suspected, but I now 
find it very difficult to keep them apart, although their union would throw 
discredit on the validity as differentiating characters of leaf-outline, peduncled 
as opposed to sessile staminate receptacles, pinnate as contrasted with the most 
definite triplinerved venation, and coarse teeth or even lobes as against entire 
margins, all in the one case. No one could fail to distinguish the two types 
of leaves; one of the difficulties is that some plants having the upper leaves 
of the narrow, pinnate-veined type of E. variabile, have others lower down on 
the same stems which approach closely to the normal type of E. carinoi. This 
is to some extent shown on plate 1 of this volume, but is more definite on 
collections recently obtained. In no case has a plant yet been obtained with leaves 
which are entirely typical of both species. Vanoverbergh 1226 is certainly E. 
variabile; the receptacles are borne on slender peduncles usually 10 to 18 mm 
long, but shorter in the case of very young receptacles; the bracts are barely 
eorniculate as in E. carinoi, the flowers tetramefous. The staminate receptacles' 
on the type sheet of E. carinoi have usually rather stout peduncles 4 to 6 
mm long, hidden on a front view by the receptacles themselves, or the young 
receptacles may be sessile. For. Bur. 4841 Curran, Merrill 4847, 7638, Bur. Sci. 
2828 Mearns, Bur. Sci. 14111, 14128 Robinson, are quite typical E. carinoi in leaf- 
characters, except that the under surface of the leaves of the first is slightly 
pubescent; all but one have staminate receptacles quite as in the type of that 
species, with short peduncles or none, the exception is pistillate only. Vanover- 
bergh 510 is also typical E. ca/rinoi, bnt Vanoverbergh 510a, collected by him 
at a different locality as the same species, while good E. carinoi as to leaves, 
except that they are smaller than usual, has the staminate receptacles on 
slender peduncles, on one plant 5 to 10 mm long, on the other only 1 to 3 
mm. In spite of the differences in the peduncles the plants could so far bo 
