461 
THE FORMATION OF THE FALLS. 
which thus rent the bed of the stream, no change of level 
took place in the two parts of the river bed thus rent, so 
that at Garden Island the water suddenly disappears, and 
on the opposite side of the reft can be seen the grass and 
trees growing where once the river ran, on the same level 
as the present bed of the stream. The first crack is a little 
over I860 yards long, a little more than the width of the. 
river. The river here runs nearly north and south, and the 
reft is nearly east and west. From the edge of the reft the 
depth of the chasm was measured by a line let down, with 
a weight attached. After paying out 310 feet, the weight 
rested on a projection, about fifty feet above the water. 
The reft was found in its narrowest part, at Garden Island, 
to be eighty yards wide. Into this chasm the river, full a 
mile wide, rolls with a deafening roar. 
Looking from Garden Island into the chasm, down which 
the river has poured, the water is seen to divide, one half 
of it continuing its course through a narrow channel, 
twenty or thirty yards wide, directly at a right angle with 
its former course ; the other half, which fell over the eastern 
portion of the Falls, flows toward our right, and these two 
streams unite, midway, making a fearful whirlpool, and 
finding an outlet in a crack which runs at right angles with 
the reft forming the falls. This outlet is about 1 170 yards 
from the western end of the chasm, and about 600 from its 
eastern end ; the whirlpool is at its commencement. For 
about 400 feet, the Zambesi thus confined in a channel of 
only twenty or thirty yards in width, surges south, and 
then enters another chasm, nearly parallel with the first, 
and somewhat deeper. Then turning sharply to the west, 
the river flows round a promontory 1170 yards long and 
416 broad at its base. Then turning round another pro- 
montory, the river flows east in a third chasm, and again 
turning another, it flows west; and this it repeats still again, 
forming thus a zigzag channel for itself, the rocky bounda- 
ries of which are all so Rharply cut and angular, that it is 
evident the channel must have been formed by some fore* 
