266 



SCIENCE 



[N. S. Vol. XXXIX. No. 



miles southeast leading across a large bend 

 of the Missouri in a southeast direction. 

 This seems to show quite clearly that the 

 Missouri occupied this channel before the 

 ice came. If this was true it is probable 

 that the river continued south of east to Ft. 

 Stevenson, where the present stream turns 

 sharply southward. Knife River joins it 

 several miles further south, and from the 

 direction of the valleys and the fact that 

 there is an unusually large valley for a 

 small stream, Snake Creek, coming in from 

 the northeast at Ft. Stevenson, it seems pos- 

 sible that the Knife and Missouri formerly 

 joined near Ft. Stevenson and ran north- 

 east to the Souris. The divide between is 

 heavily covered with drift and we know 

 nothing of the preglacial surface below, 

 but some maps show a deep notch in the 

 east front of the divide nearly north from 

 Ft. Stevenson and in line with the north- 

 east course of the Souris on the south 

 side of its great bend. The height of the 

 Missouri at Stevenson is close to 1,700 

 feet A. T. ; of the Souris, 45 miles away, 

 is about 1,500 A. T.; the height of the 

 divide between, not less than 2,100 A. T. 

 The altitude of the terrace at Ft. Stev- 

 enson is about 1,790, which would indi- 

 cate that the old channel, if there, is filled 

 at least 230 feet with drift, which is no un- 

 common thickness. 



The Souris at present is several feet be- 

 low the bottom of the drift there, and 50 to 

 75 feet below the level of the Souris plain 

 of Lacustrine origin. The surprisingly low 

 altitude of the Souris compared with that 

 of the Missouri is most reasonably ex- 

 plained by great glacial erosion of the soft 

 rocks underlying the region. The pre- 

 glacial drainage doubtless had a much 

 gentler slope than the present surface 

 would indicate. 



2. The Heart and Cannon Ball Rivers. — 

 The Heart Eiver joins the Missouri at Bis- 



marck at 1,620 A. T. It shows a conspicu- 

 ous terrace west of J\Iandan, which rises 110 

 feet higher. Southeast of Bismarck a terrace 

 and broad valley-like depression runs up 

 Apple Creek to Menoken, 15 miles east, 

 then southeast as much further, to the 

 southwest end of Long Lake. Its bottom 

 fluctuates a little above and below 1,750 

 A. T. Cannon Ball River joins the Mis- 

 souri about 30 miles south of Bismarck, op- 

 posite the mouth of Long Lake Creek, 

 which drains a valley running north-north- 

 east, in line with the lower course of Cannon 

 Ball to the west end of Long Lake, and to 

 the valley already mentioned as coming 

 from Bismarck. Long Lake occupies a val- 

 ley 25 to 30 miles long and 4 or 5 miles 

 wide, continuing in the same direction, past 

 Dawson, on the Northern Pacific railway, 

 until it becomes filled with glacial deposits 

 and untraceable. The altitude of the bot- 

 tom of the valley near Dawson is 1,738 

 A. T. A lobe of the ice of the Wisconsin 

 stage pushed up this valley as far west as 

 Sterling. We have, therefore, quite convinc- 

 ing evidence that Heart and Cannon Ball 

 rivers, at some time, most probably in pre- 

 glacial times, flowed over the present di- 

 vide between the Missouri and James rivers. 

 Their course further east has been obliter- 

 ated by the marked erosion and deposition 

 of the various ice sheets, which have suc- 

 cessively occupied the region further east. 

 3. Grand and Moreaio Rivers. — The 

 Grand River turns southeast before reach- 

 ing the Missouri, apparently to go around 

 the south end of a high ridge which lies 

 just east of the latter stream. The Mo- 

 reau probably joined it near this point 

 not far from Mobridge. Both streams have 

 several terraces along their valleys; those 

 about 200 feet above the present stream are 

 prominent and correspond to a broad val- 

 ley which runs northeast 20 to 25 miles, 

 where the surface becomes morainic and 



