ACROSS UNKNOWN SOUTH AMERICA 
down a small channel fifty to eighty metres wide, cut 
among the vertical rocks on the right side of the waterfall. 
The water in the channel flowed in steeply sloping cas¬ 
cades. The channel twisted abruptly in two or three 
places, and in one spot went through a rocky neck thirty- 
five metres wide, where the force of the current was so 
great that I was really perplexed as to how we could take 
the canoe down without getting her smashed to atoms. 
Providence came to my help again. In looking about 
I discovered another ancient channel, now almost dry and 
strewn with innumerable rocks, by which it might be 
possible to take the canoe overland until we could find 
a smooth place in the water below the rapid. On further 
exploring that channel, as I was quick enough in noticing 
its possibilities, I found at the end of it what the Brazilians 
called a recanto — that is to say, a backwater which the 
river had there formed, and which would be a great help 
to us in floating the canoe once more. 
This plan involved a great deal of hard work, as not 
only had we to shift many large rocks out of their position, 
but we had to construct a railway with felled trees and 
rollers upon them. We could not get perfectly horizontal 
rails, so that the effort of moving the canoe along inch by 
inch with levers was trying, especially as we had had 
insufficient food for many days and our strength was fast 
failing. 
To make matters worse, Alcides that day broke out in 
revolt. He had, like many ignorant people, the misfor¬ 
tune of believing that he knew better than anybody else. 
I had given him instructions to place the rails and rollers 
in a certain position, so that the canoe could be shifted 
over some unpleasant rocks. He, however, insisted on 
placing the rollers in the wrong place and on using the 
levers in the wrong spots, so that they not only did not act 
helpfully, but actually had the contrary effect on the canoe 
from that which we wanted to obtain. I remonstrated, 
200 
