THE EQUATORIAL FORESTS 21 
extend on the ground to 15 feet from the base 
of the trunk, and the same distance up it — indeed, 
I have occasionally seen a sapopema stretch up- 
wards to a height of 50 feet before it fairly ran out. 
A slight roof of palm leaves being made to rest 
on the hypotenuses of two adjacent sapopemas, 
the intermediate space has often served me as a 
temporary hut. An idea of their size may also be 
formed from the fact that I have seen a table-top, 
in a single piece, 8 feet long by 4 feet wide, cut 
out of a sapopema, and in the Andes of Maynas 
I once saw a circular tray of the same material 
very nearly 6 feet in diameter. Sometimes they 
fork once or oftener before plunging into the 
ground, and sometimes they are free beneath to- 
wards the centre, so as to present a combination 
of arch and buttress. Not infrequently they are 
fantastically twisted, and the outer edge may be 
either straight or bulged outwards ; but in all cases 
their woody fibre is in a state of extreme tension, 
so that on striking an axe or cutlass suddenly into 
them they give out a sound like the breaking of 
a harp string. On examining attentively trees 
which have sapopemas notably developed, it will be 
found that they have no central or tap root at all, 
nor do the lateral roots dip deep under the soil. It 
is clear, indeed, that the roots of lofty trees which 
do not take deep hold of the ground must either 
run a long way on or near the surface, of which we 
have an example in the spruce fir [Abies excelsa), 
or else must extend vertically as well as horizon- 
tally, so as to perform the office of both buttresses 
and stays, as in the sapopemas of which we are 
treating. When I afterwards explored the great 
