History of lake Agassiz. — Vpham. 227 
into this valley directly from the melting ice-sheet, its modified 
drift, gathered from the ice in which it had been held, continued 
to increase in depth ; but when the ice had retreated beyond the 
limits of the Minnesota basin, the water discharged here from 
lake Agassiz brought no modified drift, and was consequently a 
most powerful eroding agent. By this river Warren the valley 
drift, so recently deposited, was mostly swept away, and the chan- 
nel was excavated to a depth lower than the present river. But 
since lake Agassiz began to outflow northeastward, the Minnesota 
valley and that of the Mississippi below, carrying only a small 
fraction of their former volume of water, have become consider- 
ably filled by the alluvial gravel, sand, clay and silt, which have 
been brought in by tributaries, being spread for the most part 
somewhat evenly along these valleys by their floods, f 
Prof. J. E. Todd supplies me the approximate outline of a lake 
named b} r him lake Dakota, which occupied the valley of the 
James or Dakota river contemporaneously with the foregoing, 
reaching from Mitchell 170 miles north to Oakes and varying from 
10 to 30 miles in width. J It outflowed southward by the present 
course of the James to the Missouri. The Dakota ice-lobe, which 
had filled this valley and in its recession formed the northern shore 
of lake Dakota, was not therefore the cause of this lake in the 
same wa}- that the lake in the Blue Earth and Minnesota basin and 
lake Agassiz owed their existence to the barrier of the ice-sheet 
in its retreat. The bed of lake Dakota has a nearly uniform ele- 
vation of 1,300 feet, or is within ten feet below or above this, 
throughout its length ; and during the glacial recession it was cov- 
ered by a lake whose shores have now a height of about 1,300 to 
1,350 feet, probably ascending slighthy from south to north, as 
compared with the present sea level. Professor Todd states that 
the surface of this lacustrine area in its southern part, from 
Mitchell to Eedfield, is nearly flat till, but thence northward is 
sand and loess-like silt, while considerable tracts of the eastern 
boixler of its north part consist of low dunes. 
the chains of lakes in Martin county, Minnesota, which are apparently 
due to interglacial water-courses that were not wholly filled with drift 
in the last glacial epoch. 
f "The Minnesota Valley in the Ice Age," Proc. Am. Assoc for Adv. 
of Science, vol. xxxii, 1SS3, pp. 213-231; also in Am. Jour. Sci.. Ill, vol. 
xxvii, Jan. and Feb., 1884. 
X This lake is partially mapped by Prof. Todd in Proc. Am. Assoc for 
Adv. of Science, vol. xxxiii, 1884, p. 393. 
