34 ANNALS NEW YORK ACADEMY OF SCIENCES 



of Kio Amazonas, taken along with the fact that the Amazon flows in 

 one channel only in two places (just below the month of Eio Madeira 

 and near Obidos), also strongly indicate that this is the region of the 

 previous watershed. Is it not to be expected that some trace of the 

 original watershed would still remain, if the reversal of the Amazon took 

 place during the Miocene? There certainly is not the slightest sug- 

 gestion of such a divide between Santerem and the month of the Amazon, 

 because the Amazon is so wide and swampy in all of this region, and 

 this is especially true in the region of Rio Parii. Below Santerem, the 

 Amazon has many channels and many islands; between Santerem and 

 Manaos, there are two places where the Amazon flows in one channel, and 

 above Manaos Eio Solomoes (Amazon), there are again many channels 

 and many islands. 



In view of all this, it appears that the original watershed must have- 

 been somewhere between the mouths of Rios Madeira and Tapajos, and 

 I think that Obidos is near the actual point of reversal. 



3. I also do not believe that there is any evidence of a huge lake 

 which extended eastward from Rio Nauta to the old watershed. This 

 old view of Tertiary lakes has been ably combatted by Hatcher, Matthew 

 and others, but it will perhaps be worth while to consider the formation 

 of highland deposits in order that the fresh-water deposits of Erere of 

 the lower Amazon as well as the other supposed evidence for a colossal 

 Amazonian lake will have an explanation. In the following brief con- 

 sideration, I have chosen the headwaters of Rio Guapore, but those of 

 Rio Paraguay are equally as instructive. 



Many streams, headwaters of Rio Guapore, dash down the more or 

 less perpendicular faces of the so-called sierras from the flat surfaces of 

 the dissected Piano Alto into a large semicircular valley. This valley 

 includes the extensive Campos de Matto Grosso. Farther down this- 

 valley, the campos are replaced by gigantic forests which encroach upon 

 the Guapore so much that its channel is almost stopped up in several 

 places below the Villa de Matto Grosso. 



In this same region, the Serras de Ricardo Franco and de Parecis 

 (made by the Guapore dissecting the Piano Alto) are much nearer to 

 the river than in the region of the semicircular valley. This serves as 

 8 block to the exit of the water which falls above this point. The dense 

 forest, fallen trees and water plants naturally assist in storing the heavy 

 rains which fall in the semicircular valley during the rainy season 

 (November to April). 



In the middle course of the Guapore, its channel widens very much. 

 Tremendous sand bars are encountered in each bend of the river. This 



