v VALLEY OF THE AMAZONS. 
217 
described above, at a later period it became an 
island in the bed of the Amazons, which, di¬ 
viding in two arms, encircled it completely, and 
then, joining again to form a single stream, 
flowed onward to the sea-shore, which in those 
days lay much farther to the eastward than it 
now does. I suppose the position of the island 
of Marajo at that time to have corresponded 
very nearly to the present position of the 
island of Tupinambaranas, just at the junction 
of the Madeira with the Amazons. It is a 
question among geographers whether the To¬ 
cantins is a branch of the Amazons, or should 
be considered as forming an independent river 
system. It will be seen that, if my view is 
correct, it must formerly have borne the same 
relation to the Amazons that the Madeira River 
now does, joining it just where Marajo divided 
the main stream, as the Madeira now joins it 
at the head of the island of Tupinambaranas. 
If in countless centuries to come the ocean 
should continue to eat its way into the Valley 
of the Amazons, once more transforming the 
lower part of the basin into a gulf, as it was 
during the cretaceous period, the time might 
arrive when geographers, finding the Madeira 
10 
