William A. White
Department of Geology
Mitchell Hall 029A
University of North Carolina
Chapel Hill, North Carolina 27514


Most drumlins seem to have been carved out of pre-existing stuffs by rapid surges of water-rich stagnant ice which had long resided in water bodies.

The central New York drumlin field shows a zone of mostly small, well-streamlined drumlins between zones of larger, less streamlined drumlins up-glacier and down-glacier from it. A melt-water channel passes around the down-glacier edge of the zone of smaller drumlins to debouch into Irondequoit Bay and thence into a subaqueous valley which leads into the deepest part of Lake Ontario.

These observations can be explained by assuming that the drumlins were carved out of drift sheets, lake bed sand Paleozoic shales by repetitive surges from a stagnant remnant of the Laurentide Ice Sheet which was resting in the basin of the present Lake Ontario when it held glacial Lake Lundy, the earlier surges being grounded across the entire area of the three zones of drumlins to carve a continuous field of larger drumlins. And as the ice thinned, its later surges floated across the up-glacier zone of larger drumlins, further carved the middle zone of small drumlins, and dissipated its erosional force before it reached the down-glacier zoneof large drumlins.

Most other drumlin fields are also on ground which rises away from former proglacial water bodies.

It is suggested that ribbed or rogen moraine may have been deposited by water-rich surges which had not floated and hence still carried their basal load of rock waste.



The genesis of drumlins has long been a controversal subject. Muller (1974) givesan excellent review of the difficulties involved in explaining their distribution, external shape and internal composition. Most discussion has focused on the question whether their external shape is a product of deposition (Fairchild 1907; Charlesworth1957), erosion (Gravenor 1953) or both (Smalley and Unwin 1968).

Some drumlins unquestionably were carved out of pre extant stuffs for they include such diverse components as tills, Paleozoic bed rock, lake beds, and waterlaid, cross bedded sands and gravels. Sometimes two or three different components are in the same drumlin, as in Maclntyre Bluff on the shore of Lake Ontario near Fairhaven, New York. There wave erosion has exposed the interior of a drumlin made of two layers of till separated by a stratum of clean water-washed cross bedded sand which extends all the way across the drumlin and is abruptly trun-cated on both sides of it. Similar features are noted by Muller (1974)near Syracuse, New York.

I followed the drag line which opened apipe line ditch six feet deep from Geneva toAuburn, New York, and in the area west of Auburn where the drumlins are long and low, I noted that few of them were made of asingle lithologic component. In one instance Paleozoic shales, lake clays and till were all exposed in a single drumlin, despite the mere six-foot depth of the ditch.

I know of no drumlin whose external shape can conclusively be shown to bedepositional in origin. Anticlinal forms have been cited (Fairchild 1907) in the wave-cut bluffs where the shore of Lake Ontario transects drumlins. But the gentle undulation in the layers of drift in these bluffs are not coextensive with the shape of the drumlins in any consistent way. What seems to have been misconstrued asanticlinal structure results from slump atthe edges of the drumlins, where the layersof drift were truncated by the glacial erosion which carved them out of a continuous drift sheet. The tendency to slump is seen in numerous road cuts in clay-rich drumlins further inland where riprap has been installed to stop slumping onto the road beds in wet seasons.

In this paper I present a hypothesis which suggests that the shape of all drumlins is erosional and that they were carved out of prexistent stuffs by rapid surges of water-rich stagnant ice emanating in most cases from proglacial water bodies.

Central New York shows an unusual relation between three contiguous zones drumlins which offers insight into the factors which enable drumlins to form.

A wide zone of mostly small, narrow,highly streamlined drumlins lies immediately up-glacier (northward) from a cluster of end moraines (Fig. 1) of which the Waterloo andUnion Springs moraines are known by name (Shuxnaker 1957). Immediately down-glacier(southward) from the Waterloo end moraine,mostly at higher elevations, there is a zone of less- perfected drumlins which are larger and less streamlined. Apparently the external form of these larger more southerly drumlins is older than that of the smaller drumlins of more perfected streamlining for they antedate the Waterloo end moraine which delimits the zone of smaller drumlins. Up-glacier (northward) from the zone of smaller better streamlined drumlins, and at generally lower elevations, along the southborder of Lake Ontario there is another zone of mostly larger, less perfected drumlins which are similar in size and shape to those south of the Waterloo end moraine.

This distribution in which a zone of smaller, younger, narrower, more perfected, better streamlined druimlins lies between two zones of larger, broader, less perfected, poorly streamlined drumlins can be explained by assuming that originally there was a continous field of older, larger, imperfect, poorly streamlined drumlins which covered the entire area of the three present zones described above. This older continuous field of larger drumiins would have antedated the glacial advances which madethe Waterloo-Union Springs group of end moraines. When those advances came, they were surges of ice that was largely floating in the basin which now holds Lake Ontario. Apparently these surges of ice floated overthe older drumlins at the north leaving them unmodified, were grounded across the area now marked by the zone of smaller drumlins,and made little penetration into the zone of older drumlins down-glacier form the Waterloo end moraine. In passing over the area where it was grounded the iceapparently further eroded the older larger drumlins, smoothing them into smaller better streamlined forms, many of which have narrow ridge crests unlike the gentle slopes and broad curvature of the older larger "whaleback" drumlins to north and south.

Presumably the formerly continuous field of larger drumlins was cut from a prevously deposited drift sheet by surges of water soaked ice which were grounded across the entire area now divided between the three zones. That they are erosional remnants of a drift sheet is shown by their varied content of till, glaciofluvium, lakebeds and pre- glacial bed rock.

The basin which holds Lake Ontario is dual. A topographic saddle crosses it from south to north just west of the city of Rochester, New York, dividing it into two sub-basins. The central New York drumlin field described above lies south and east of the eastern sub-basin. Projections of the long axes of its drumlins converge up-glacier at the deepest part of the eatern sub-basin. Similarly drumlins in the province of Ontario northwest of Lake Ontario have long axes whose projections converge toward a focal point in the deepest central part of the western sub-basin.

Just as the part of the central New York drumlin field which has the more perfected drumlins is roughly bounded by the Union Springs and Waterloo group of end moraines, similarly the Oak Ridges end moraine encompasses the drumlin field which lies along the northwestern rim of thewestern sub-basin of Lake Ontario. This suggests that the drumlins of both fieldswere formed by glacial advances that werenearing the end of their travel. In both instances the ice was climbing ground whose regional slope rises gently in a down-glacier direction. In the case of the durumlins northwest of the western sub-basin of Lake Ontario, it is notable that the drumlin-making ice was moving in a direction toward the center of the Laurentide ice sheet, which suggests that those drumlins were shaped by an advance from a dissevered mass of ice centered in the western sub-basin of Lake Ontario.

Proglacial lakes are ordinarily dammed by ice on the side of the lake which is up- glacier in terms of the regional direction of ice flow during the time of the ice flood. Thus it seems instructive that the place,where these drumlins of Ontario were made by ice flowing in a direction opposite to that of the ice flood, is also the place where a proglacial lake had an infusible shore manifested now by relict shoreline features that encompass the drumlin field. This association is consistent with the idea that drumlins are carved by ice flowing out of water bodies.

Just as the down-glacier edges of these two drumlin fields are closely delimited by end moraines, so also are they closely associated with relict shorelines of proglacial lakes, apparently those of glacial Lake Lundy. Leverett and Taylor (1915, plate XIX) show that at the time Lake Lundy was extant its basin was filled with glacial ice save for narrow strips of open lake water around its northwestern and southern borders including the vicinities of the above-described end moraines which bound the two drumlin fields.

In the central New York field the more northerly drumlins of the better streamlined zone were cuffed and benched with associated spits and bars by the waves of glacial Lake Iroquois (the latest and Lowest- level proglacial lake of the Lake Ontario basin). But no drumlins of this same zone show any shoreline features of proglacial lakes whose surface levels were higher than that of glacial Lake Iroquois. This is consistent with the idea that the drumlins of the smaller, better streamlined group were given their final shape by ice which covered them during the time these higher lakes were extant.

These observations offer the possibility that the glacial advances which formed both the smaller drumlins and their associated end moraines were surges whose flow was facilitated by a thick basal part of the ice being saturated with liquid water which caused it to lose its rigidity.

Contained in a topographic basin with some 260 meters closure at the time of glacial Lake Lundy, when the ice thined to a thickness lightly greater than the depth of the impounded water it began to float. Since the basin is ellipsoidal in form, shallowing toward its edge, and since the mass of ice was probably dome-shaped, thinner peripheral parts of the ice might have floated in shallower water at about the same time that the thicker central part of the ice floated in deeper water. Thus partially levitated, most of the restraint caused by basal friction would cease and surging would be facilitated.

Support for the idea that the ice which shaped the younger, better streamlined drumlins of central New York was floating can be had from a subaqueous valley in the bottom of Lake Ontario. Beginning as Irondequoit Bay on the east side of the city of Rochester, New York, this valley extends well down into the eastern sub-basin of Lake Ontario. A subaerial valley extends the subaqueous valley landward from the head of Trondeouott Bay. This subaerial extension curves southward and eastward in a headward direction aroung the down-glacier edge of the part of the central New York drumlin field which has the smaller, better streamlined drumlins. It can be traced along the valley of the Seneca River in a relict down-stream direction but presently up-stream direction from the north end of Lake Cayuga to the north end of Lake Seneca, thence through a broad valley now occupied by small streams for about five miles to the valley of Canandaigua Outlet about two miles east of Phelps. Thence upstream along the Canandaigua Outlet Valley to Manchester, thence up the valley of Black Brook, through Bozzey Swamp and through a part of the valley that is drained by Ganargua Creek to the head of Irondequoit Creek which debouches into Irondequoit Bay. This route can be traced on the following Th minute topographic maps listed in sequence as they show the valley in relict down- stream direction curving from southeast to northwest: Seneca Falls, Geneva North, Phelps, Clifton Springs, Canandaigua, Victor, Fairport, Webster, and Rochester East. Apparently the subaerial part of the valley was cut by water which was deflected by the front of the ice that did the shapeing of the better-streamlined drumlins. Much of the water was probably melt water from the ice itself. Since the ice had done considerable erosive work in reducing the size and improving the streamlining of the drumlins, its melt water might have been muddy enough to form a turbidity current in Lake Lundy flowing around the grounded southern edge of the ice and, farther north, underneath the floating central part of the ice which occupied the eastern sub-basin of Lake Ontario.

A cross valley, now dry, connects the valley of the Genesee River with the valley described here, but it seems improbable that the Genesee River could have flowed through it to cut the sub-aqueous valley in the bottom of lake Ontario because the Genesee River would have been carrying only the discharge of contemporaneous precipitation rather than glacial meltwater. And there is little reason to think it should have been any more turbid than it is now because it is still extending its gorge headward into the shales of the Alleghany plateau but is rarely turbid. Nor are other maladjusted streams which have steep profiles where they flow down the walls of the valleys which hold the Finger Lakes: such acutely incised ravines as Watkins Glen, Taghannock Gorge, Enfield Glen, etc.

It also seems significant that there is no suggestion of a subaqueous valley off the mouth of the Genesee River in the bathymetry of Lake Ontario. Further argument to this point may be had from analogy with the Dundas Valley at the western end of Lake Ontario. it too has a sub-aqueous extension, into the western sub-basin of Lake Ontario but has no appreciable present surface drainage entering it. It extends the Dundas Valley well down into the western sub-basin of Lake Ontario, and seems to bear a similar relationship to the drumlins northwest of Lake Ontario as that which the subaqueous valley off irondequoit Bay bears to the zone of better streamlined drumlins of central New York.

There is no consistent subaerial valley around the edge of the drumlin field northwest of Lake Ontario save for the headwaters of the Grand River which now debouches to Lake Erie, but the slope may not have been southward toward Lake Erie while the drumlin-forming ice was present. Chapman and Putnam (1973) show a plexus of spillways in the part of the drumlin field near Guelph, Ontario, which converge toward a sand plain that plausibly was a delta in a proglacial lake. It also may be significant that a major end moraine which parallels the shore of Lake Ontario (as shown by the maps of Chapman and Putnam) changes its character from ice-laid till moraine to water-laid Kame moraine where it turns up the Dundas Valley. Successive end moraines which are convex northwestward concentric with the northwestern shorelines of Lake Ontario suggesting that several surges of glacial ice entered the area northwest of Lake Ontario from the southeast. And the absence of any appreciable present drainage into the Dundas Valley suggests that it was cut by melt water flowing out of a mass of glacial ice which, in the light of the subaqueous extension of the Dundas Valley, should plausibly have been floating in the western sub-basin of Lake Ontario.

The Union Springs end moraine is unusual in that it extends farthest down-glacier where it climbs the divide between Lakes Cauyga and Seneca in a broad southward- extending loop. This differs diametrically from an older end moraine which has long, digitate apophyses extending down- glacier into the valleys that hold the Finger Lakes, and long intervening reentrants into the ice front where the ice failed to climb the divides between the Finger Lakes. This older moraine is shown on th glacial map of North America (Flint 1945) but is not reflectd on the later map. The end moraines of the Waterloo-Union Springs group are quite different in plan. The glacial advances which formed them seem to have been little deflected by local topographic features, but rather flowed equitably into the broad, regional, scoop€shaped valley which holds the younger, better streamlined, smaller drumlins. Where the ice which made the Union Springs moraine climbed the divide between lakes Cayuga and Seneca it was on the axis of this regional scoop-shaped valley. It extends no long apophyses up any of the Finger Lake valleys but cuts rather equitably across both lake basins and inter-lake divides.

This ability of the Union Springs end moraine to climb the divide between two lakes in a broad, down-glacier loop suggests unusual momentum and speed of flow in the glacial advance which made it. These observations court speculation that such ice flows as those which produced the waterloo moraine and consummated the shaping of the smaller drumlins were incoherent because they were charged with interstitial liquid water which allowed small particles of the ice to move independently of their immediate neighbors. The idea that the ice was disaggregated, water-saturated or "rotten" is made more plausible by the fact that the ice which shaped the drumlins northwest of the western sub-basin of Lake Ontario flowed northward, back toward the area which was the center of the Laurentide ice sheet. This suggests that it had been stagnating in the Lake Ontario (Lake Lundy) basin for some time before it surged northwestward to carve the drumlins northwest of Lake Ontario.

Such incoherent consistency of the ice might result from a glacial remnant residing a long time in the deep undrained basin that held a continuous succession of high level proglacial lakes such as Lakes Warren, Lundy, and Wayne. The depth of the basin would be significant because it would influence the minimum thickness of basal ice which would be saturated with liquid water. This thickness would be augmented by the height of the water table in the ice above the level of the encompassing proglacial lake. When the openings in the water- saturated ice became sufficiently large and numerous its coherence would be destroyed and allow a rapid surge to occur. The force driving the surge would I derive not only from the weight of the I superincumbent rigid ice, but also from the upward slope of the base of the ice from the deeper central part of the ellipsoidal basin toward its periphery. This would allow the surge of disaggregated basal ice to float diagonally upward under the overlying rigid ice. The surge would move rapidly because the ice would be floating in the lake basin, encountering little resistance to basal slip. As the surge progressed the ice involved in it would become more disaggregated with distance traveled. When the surge ran aground it might be a thick slurry comprising bits of ice suspended in liquid water. This could trvel much faster than typical coherent rigid glacial ice. Such gross fluidity would facilitate change of flow direction around obstacles, thus accounting for the short horizontal radius of curvature in the shape of many of the drumlins carved out of the drift where the surge ran aground in shallow water.

After the surge a new zone of ice would sink to levels lower than that of the proglacial lake in which it was floating and become fluidized in turn to prepare for another surge. From the multiplicity of the Waterloo-Union Springs group of end moraines it looks as though surges may have been repetitive or perhaps fluctuant in successive waves.

The medium which emplaced the Union Springs end moraine and those to the south of it were not deflected by topographic obstacles. The medium which emplaced them climbed the divide between lakes Cayuga and Seneca and passed over a number of the southern group of larger drumlins without being deflected from its regional direction of flow (Fig. 1) and without any presently perceptible erosional effect on them. This suggests that toward the end of their travel, the surges which made these moraines had been so disaggregated that they moved more like a rapid flow of liquid water than a coherent glacial mass. Perhaps these moraines were emplaced more like the swash marks on a beach than like a moraine formed at the edge of an advance of coherent glacial ice. A textural study of the Union Springs and its associated moraines might be helpful. The best evidence I have found for the hypothesis of drumlin-genesis presented here comes from the area affected by ice emanating from the basin of Lake Ontario, but support is found in the locations of other drumlin fields down-glacier from water bodies. Drumlin fields are associated with the basins of all the Great Lakes. Ice moving south-easterward out of the area around Lake Erie made drumlins were it climbed the slope of the Alleghany plateau. The interlake area between the south end of Georgian Bay, the southern part of the main basin of Lake Huron and the west end of Lake Ontario supports three intersecting drumlin fields. Each of these fields comprises drumlins whose axes converge toward the adjacent, engendering basin. In the lower peninsula of Michigan drumlins of a field southwest of Saginaw Bay have axes that converge toward the bay. They persist up the regional slope of the drainage area of the Shiawassee River but do not extend over the divide into the drainage area of the opposing Grand River where the regional slope is in a down-glacier direction. Drumlins on the east side of the Great Kame Moraine west of Lake Michigan have axes which suggest that they were shaped by ice emanating from adjacent parts of Lake Michigan. Southwest of Green Bay are drumlins whose axes converge toward it. Similarly drumlins around Grand Traverse Bay have axes which converge toward it. And drumlins of the field southwest of Duluth, Minnesota, have axes which converge northeastward toward the southwestern end of Lake Superior.

It is notable that no drumlins occur in glaciated areas where the regional slope declines down-glacier as it does in the glaciated parts of the drainage areas of the Mississippi, Susquehanna arid Delaware River systems. This is consistent with the idea that ice moving through impounded water is necessary for drumlin genesis. Drumlins in western Washington seem to have been shaped by ice emanating from Puget Sound. And those northeast of the Baraboo Mountains of Wisconsin are in an area that lay beneath the water of former proglacial Lake Wisconsin. In the Great Plains of Canada the drumlins may have been formed by ice moving up the regional slope to impound such great proglacial water bodies as that of glacial Lake McConnell.

In Europe drumlin fields are associated with the great proglacial water bodies of the arc of exhumation (White, 1972) which includes the basins of the present Baltic Sea and lakes Ladoga and Onega. In contrast to these favorable observations I find little evidence that drumlins of Ireland or New England were genetically associated with proglacial water bodies.

Howerver both these areas are in places were topographic depression caused by glacial load might have let the sea enter to levitate the ice. Charlesworth (1956, p. 394) notes that drumlins are overlain by marine clays of the Champlain and Yoldia seas and the 100-foot beach of Scotland.

Ribbed or rogen moraine (Lundqvist, 1969; Carl, 1978) may be a deposit made by early surges from the same mass of ice that later carved drumlins from it. These early surges might have been able to deposit rather than erode because they occurred while the ice was still too thick to float in the proglacial water body and still held its basal load of rock waste to build the succession of ribbed end moraines. Later when the ice began to float it would lose its original basal load and would not be in contact with the lake bottom to pick up a new one.

This idea might explain the relation between ribbed moraine and drumlins. It would also explain why both drumlins and ribbed moraine occur on ground whicha rises down-glacier from water bodies. Most drumlin fields show little evidence of having been associated with ribbed moraine. This suggests that there may have been little if any basal load in the ice which carved them. Plausibly ribbed moraine was made by water-rich ice which surged while in contact with the ground. Later after it had lost its basal load by levitation in the water body it entered an erosional mode rather than a depositional one and carved some of the ribbed moraine into drumlinoid forms. Most of these drumlinoid forms associated with ribbed moraine are poorly streamlined. Even if the ice were not in a topographic basin or proglacial water body it might be more water-rich in its lower part because it rested on ground which rose down€glacier away from the ice front and hence it would not drain well.


Carl, J. B., 1978. Ribbed moraine-drumlin transition belt, St. Lawrence Valley, New York, Geology v. 6 p. 562- 566.

Chapman, L. J., and Putnam, D. F., 1973. The physiography of southern Ontario. Ontario Research Foundation. University of Toronto Press. 386 p., 7 maps.

Charlesworth, J. K., 1957. The quaternary era with special reference to its glaciation. Edward Arnold, London. 2 vol. 1700 p.

Fairchild, H. L., 1907. Drumlins of central western New York. New York State Museum Bull, Ill; p. 391-443.

Flint, R. F., 1945. Editor, Glacial map of North America, G. S. A.

Gravenor, C. P., 1953. The origin of drumlins: Am. Jour. Sci. v. 251, p. 674-681.

Leverett, F., and Taylor, F. B. 1915. The Pleistocene of Indiana and Michigan and the history of the Great Lakes. U.S. Geological Survey. Mon. 53, 529 p.

Lundqvist, Jan 1969. Problems of the so-called rogen moraine; Sveriges Geol. Undersoknung Ser. C., 648, Arsbok 64, p. l-32.

Muller, E. H. 1974. Origin of drumlins. In Glacial Geomorphology; Publications in geomorphology, State University of New York, Binghamton, New York. Donald R. Coates Editor; p. 187€204.

Shumaker, R. C. 1957. Till texture variations and the Pleistocene deposits of the Union Springs and Scipio Guadrangles, Cayuga County, New York. M. S. thesis Cornell University, Ithaca, New York.

Smalley, I. J., and Unwin, D. J. 1968. The formation and shape of drumlins, and their distribution and orientation in drumlin fields. Jour. of Glaciology., v. 7, p. 377-390.

White, W. A. 1972. Deep erosion by continental ice sheets. G.S. A. Bull. v. 83, p. 1037-1056.

Manuscript received July 24, 1985; accepted October 10, 1985.

Search billions of records on Ancestry.com