The First Ascent of Roraima. 29 
accessible portion of this savannah slope and of the 
forest belt above, even as far as the base of the cliff, 
partly in order to examine the vegetation but chiefly in 
order, when a few rare hours of clear weather admitted 
of this, from every available point to study the ledge up 
which we hoped to ascend. Sometimes it looked possi- 
ble, sometimes impossible. To make clear the nature 
of the position some further account of the contour of 
this aspect of the mountain must be given. 
See Fig. 3. 
The gradual savannah slope has already been distin- 
guished from the much more abrupt forest slope. But 
in this latter again three regions or belts may be quite 
clearly distinguished. First, and immediately above the 
upper edge ot the savanah slope, is a belt of very dense 
wood in which the trees are small but stand very closely 
together, great quantities of the Geonoma already de- 
scribed and a few curiously dwarfed manicole palms 
(Euterpe edulis) occurring in it, the whole being 
much matted together by the long winding stems of the 
small bamboo. The path upward through this is steep 
and slippery, but there are few boulders. Next comes abelt 
of bush in which the vegetation is chiefly low and bush-like, 
averaging not more than from six to eight feet in height, 
and indeed consisting in great part of Brocchinia cordy- 
linoides. The ground here is almost completely covered 
by boulders, though these are not often of any very great 
size. Next comes a belt of rock and tree, where the 
boulders, many and large, often tower overhead, and the 
trees few, stunted, gnarled and twisted, grow round, over 
