74 Bvlletin American Museum of Natural History. [Vol. XXXVI, 



Under the second-named condition, the continuity of the forest is broken 

 by such an arid pocket as we found at Caldas, or Miller and Boyle encount- 

 ered above Dabeiba. 



At the northeast, on the right side of the lower Atrato, the Pacific coast 

 forest apparently connects with that of the Cauca-Magdalena, and this is 

 the only connection of which we definitely Imow between any of the forested 

 areas of the Tropical Zone. Mr. Douglas Fyfe, an American engineer, sit- 

 uated in northern Colombia, writes me of the country at the border of the 

 Colombian-Pacific forests and the Caribbean Savannas: "The Sinfl is a 

 wide, sluggish stream meandering through a broad plain of very deep allu- 

 vium. Along its banks are situated beautiful grass-covered cattle ranges. 

 The river is carrying a vast amount of sediment and gradually setting up a 

 large area at its mouth forming deltas and reed swamps. Numerous water- 

 fowl seem to inhabit these swamps the year around. The country lying 

 east of the Sinu for about twenty miles is low-lying and under water part 

 of the year; cienagas in fact. Beyond this lie the savannas. The country 

 west of the Sinu is entirely forested to the Pacific coast, the forest beginning 

 at the Atlantic sea shore and extending inland without a break to the 

 interior." 



The differences between the forest of the lowland and that of the foot- 

 hills are well described by Allen in the narrative of his journey with Miller 

 just alluded to. In connection with the change in gradient, character of 

 the soU, etc., we have here two widely varying types of environment which 

 doubtless account for the fact that some species appear to be confined to 

 the bottomlands. 



The Cauca-Magdalena forest is contained chiefly in the Department of 

 Antioquia. Miller's exploration and inquiries show that it extends from 

 the northeastern border of the Pacific coast forest northward to the Carib- 

 bean coast and thence eastward to the shore of the Rio Sinu, when it meets 

 the western border of the marshes which pass into the Caribbean savannas. 

 Thence it extends southward up the lower Cauca, and doubtless also the 

 Neche, to about Lat. 7°, and up the Magdalena Valley to La Dorada on the 

 Magdalena River, where, on the valley floor, it terminates abruptly and is 

 succeeded by the Savannas of the upper Magdalena. Along the foothills 

 of both the Eastern and Western Andes, the forest extends south of La 

 Dorada. Northward, in the Magdalena Valley, it is bounded on the west 

 by the San Jorge and on the east by the foothills of the Eastern Andes, 

 while its northern limit lies near Banco. Eastward of this point, it may 

 occupy the foothills of the Sierra de Motillones and Sierra del Valle de Upar, 

 at the northern end of the Eastern Andes, but I have no definite information 

 in regard to this region. Whether the Cauca-Magdalena forest is connected 

 with that of the Maracaibo Basin, remains therefore undetermined. 



