60 B. S. Tarr — Origin of Terraces in Glaciated Regions. 



it into rills and rivulets even where no drainage lines previ- 

 ously existed. While this is written chiefly with reference to 

 the arid headwaters it applies almost equally to all the streams, 

 even those in the sub-humid belt. These tributaries during 

 the greater part of the year consist of a few pools, often iso- 

 lated, sometimes connected by a slowly trickling stream. These 

 pools are enclosed commonly in bars or delta bars in the stream 

 channel,, formed during flood times, and the violence of these 

 floods is attested by the presence of drift wood lodged in 

 the pecan trees many feet above the low water stage of the 

 stream. In the Colorado these are sometimes at an elevation 

 of fifty feet above the low water surface. 



If I am not mistaken, we have here all the essential condi- 

 tions which were present at the time of the formation of the 

 terrace deposits at the close of the Glacial period. There is a 

 slope so moderate that the excessive sediment load cannot be 

 transported, and the greatest excess of sediment comes at times 

 of great flood, for the sudden downpour of water upon the arid 

 plains carries along to the streams a vast bulk of sediment. 

 There are, owing to the peculiarities of rainfall, periods of 

 extreme high water and of extreme low water, and also occa- 

 sional irregular periods of moderate rise. In the glacial 

 regions the south-flowing stream had a moderate slope, proba- 

 bly less slope than at present. Yast quantities of sediment 

 were furnished not alone by the supply from the ice itself but 

 also from the beating of the rains and the washing action of 

 the melting snows upon the barren soil recently uncovered 

 from beneath the ice and as yet unclothed by vegetation. In 

 the summer, excessive floods must have been furnished by the 

 melting of the ice ; in the winter the water flow was at a 

 minimum ; and in the spring and autumn, when the melting 

 was moderate and somewhat spasmodic, floods of medium 

 height occurred. 



The parallel of conditions seems almost exact ; let us see how 

 the results coincide. In the Colorado the great floods rise 

 forty and fifty feet and spread out over broad flood-plains on 

 either side, and these flood plains are of fine silt, well strati- 

 fied. This is the upper flood-plain terrace. In this flood- 

 plain a broad channel is carved, which, in moderate floods, is 

 either partly or completely filled by the water. It is however 

 a double channel, for in it, either on one side or on the other, 

 or even in the middle, there is a smaller channel about half its 

 width and this is the ordinary channel, — the one in which 

 water is always to be found. So, rising from this inner chan- 

 nel, one comes to a terrace which is formed by the moderate 

 floods and which is always present on one side, sometimes on 

 both. It is in general more sandy than the other or upper 



