Geological Papers. 89 



Geike, Davis and Salisbury look on drumlins as analogous 

 to the sand-bars of streams. 



In speaking of the origin of drumlins, Upham in substance 

 writes:' "The till forming drumlins invariably exhibits the 

 characteristic features of subglacial drift or ground-moraine, 

 excepting its superficial portion which was englacial and sub- 

 glacial when the ice-sheet melted away." 



His hypothesis of the drumlin accumulation is somewhat as 

 follows : "In the central area of the ice-sheet the currents of 

 its upper and lower parts probably moved outward with nearly 

 equal rates, the upper movement being slightly faster than at 

 the base. Upon a belt extending many miles back from the 

 margin, however, where the slope of the ice surface had more 

 descent, the upper currents of the ice, unsupported on the 

 outer side, would move faster than its lower currents, which 

 were impeded by friction on land. There would be accord- 

 ingly in this belt a strong tendency of the ice to flow outward 

 with somew^hat curved currents, tending, first, to carry the 

 onwardly moving drift gradually upward into the ice-sheet, 

 and, later, to bear it downward and deposit it partly beneath 

 the edge of the ice and partly along the ice boundary. And 

 as the ice at the time of its melting was making halts and 

 occasionally readvancing, if the ice had nearly a constant posi- 

 tion during several years, its border became marked by ter- 

 minal moraines close to the glacial boundary. Whether it 

 halted and ever readvanced, or merely its retreat was much 

 slackened, the upper part of the ice must have descended over 

 the lower part. This differential and shearing movement 

 gathered the stratum of englacial drift into the great len- 

 ticular masses or sometimes longer ridges of the drumlins, 

 thinly underlain by ice and overridden by the upper ice flow- 

 ing downward to the boundary and bringing with it the for- 

 merly higher part of the drift stratum to be added to these 

 ground-drift accumulations. The courses of the glacial cur- 

 rents and their convergences to the places unoccupied by 

 drumlins were apparently not determined so much by the 

 topography of the underlying land as by the contour of the 

 ice surface, w^hich, under its ablation, had become sculptured 

 into valleys, hills, ridges, and peaks, the isolation of the ele- 

 vations by deep intervening hollows being doubtless most con- 

 spicuous near the ice margin. And again, when the boundary 



4. See Upham, Glacial Lake, Agassiz, G. S. Monograph xxv, and Ice-sheet Moraine, 

 Geol. Surv. Minn., 1880. 



