Drumlins are a ubiquitous landform in lands once overrun by glaciers, and yet after two centuries of studying them, we still aren't certain how these teardrop-shaped hills form.

Top image: A drumlin field 27 kilometers southwest of Amundsen Gulf in the Nunavut Territory of Canada, imaged by Landsat 8 on June 21, 2014. Credit: NASA


When glaciers creep over a landscape, they scour it clean of loose material, strip it of vegetation, polish the very bedrock, then rebury it all in thick layers of jumbled rock and sediment. When the glaciers retreat, they leave behind a distinctive landscape of stark erosion mixed with deposits of dense, unsorted till. Along with snaking eskers of under-glacial rivers, piled moraines of accumulated geological junk, and kettle lakes from melting chunks of ice are drumlins, teardrop-shaped hills parallel to the direction of ice flow.

The drumlin field in Clew Bay, Ireland, has been drown by the ocean, transforming the hills into egg-shaped islands. The ice flow went from left to right, with the long tail in the lee of the flow. Image credit: BrendanConway


First named by H.M. Close in 1897, the etymology of drumlin is from the Gaelic druim, a ridge or hill, but made more far more adorable as a diminutive droimnín, "littlest ridge." Drumlins can be almost-round eggs to elongated tears, but all have their the rounded snout pointing up-ice (the stross side) and the tail stretching down-ice (the lee side). Along with being the most defining characteristic of a drumlin, this shape is one of the clues used by researchers to reconstruct the complex movement of ancient ice flows.

The individual drumlins are usually 250 to 1,000 meters long, 120 to 300 meters wide, averaging just 13 meters high. They're at least 100 meters long — any smaller than that, and they're small proto-drumlin bumps lacking in the characteristic elongated shape. Part of the mystery lays in their extreme variability in composition: drumlins can be made of carved bedrock, the chaotic jumble of glacial till, the tidy sorted muck of glaciofluvial sediments, or some desperately confusing mixture of all three. In his epic 1979 literature review on the state of drumlin research, J. Menzies laments:

The internal composition of a drumlin varies from stratified sand to unstratified till to solid bedrock, with every possible permutation between. Drumlins may have cores of rock, sand, boulders or laminated clay.


While Stokes et al. make a heroic effort to classify drumlin composition and structure, their internal structure is just as frustratingly diverse: some have tectonic structures of interfingering, pseudo-folds, and faults that may be produced by pressure-squeezing from overlaying ice, while others have distinct layers and bands that may be the result of accretion over time, or stratified bedding independent of the drumlin's shape. They can be fissured and jointed, possibly from weathering or stress release.

Idealized sketch illustrating the potentially diverse internal structure of a drumlin, and the deceptive impression formed by limited natural exposures (1,2,3). Image credit: Stokes et al.


They do have one common compositional feature: most drumlins are composed almost exclusively of extremely local material, with transportation distances of a few kilometers or less. Erratics and other contributions from far-off locations are usually contained very tops of drumlins. For the few drumlins with more foreign material, the researchers they're in areas that experienced repeated, prolonged periods of glaciation where material could have been transported at any previous stage and only made a final, short hop when incorporated into the drumlins. With all this variability, it's no wonder that Saul Aronow wrote in 1959:

"[W]hen the conditions within the ice are present for making drumlins and related features they are formed, seemingly, regardless of the material available."

Drumlins commonly occur in vast fields with hundreds to thousands of hills over a thousand kilometres. The drumlins within a single field can have a huge variety in composition and orientation, aligned to reflect the ice flow of that particular patch. Drumlins can even layer on top of each other, possibly reflecting multiple periods of drumlin formation.


Drumlin field north of Milwaukee, Wisconsin. Image credit: USGS

Drumlin fields are usually near with the terminal moraine that marked the farthest edge of a glacier, associating them as a feature created by ice dynamics at the outer edge of a glacier or ice sheet. Most of the time, drumlin fields are in areas of thick glacial till, but they have also been found in scoured regions with just small, scattered patches of drift deposits.


As for how they form, that's where things get tricky. For decades, researchers have been openly acknowledging the lack of a decisive understanding of the genesis of drumlins. The acknowledgement of how downright perplexing drumlins are has moved beyond common understanding almost to a running joke acknowledged at the start of every research paper. In 2009, Clark et al. opened their paper with:

"[Drumlins] have mystified investigators for over a hundred years. A satisfactory explanation for their formation, and thus an appreciation of their glaciological significance, has remained elusive."

The main problem is that drumlins are formed under the ice, so we can't actually see what's happening in sufficient detail. Yet, their streamlined, elongated shape and parallel alignment makes it pretty clear that they're the result of flow somehow. That lack of a clear-cut answer is just as problematic now as it has been in the past, with the same two basic theories battling it out: Are they erosional features from subglacial water flow washing away everything around them, deposits infilling cavities under the ice, or some combination of both?


The exact details of the hypotheses have changed over time, but the basic ideas remain the same.

The earliest version of an erosional theory might be by Sir James Hall, who suggested in 1815 that catastrophic tidal waves carved the flutes and drumlins he saw near Edinburgh. A more recent theory for erosion is outlined in a truly delightful and metacognitive paper by John Shaw, explaining how meltwater and outwash from the glacier could carve under-ice material, leaving behind drumlins as slightly-more-resistant nubs under the ice.


Schematic of the meltwater outwash theory of drumlin formation by erosion. Image credit: John Shaw

The biggest catch to the erosional theory is that it requires a lot of water flow. This has been observed during outburst floods, or jökulhlaup, which can initiate with a broad, flowing sheet of water that can last for hours before eventually lessoning and becoming channelized into meltwater rivers.

The depositional theory is that some slight change in pressure from the ice above or variation in the material below results in increased deposition. This builds mounds under the glacier or deforms the underlaying bed material, which in turn changes the stress distribution from the ice above, leading to a positive feedback look increasing deposition and deformation over time.


Parabolic drumlins in a field near Livingstone Lake, Saskatchewan. Image credit: Shaw

Almost every drumlin we study is from at least several thousand years ago, created during the last ice age when vast continental ice sheets crept across the land. Literally thousands of academic papers puzzle over the mysteries of drumlin fields in northern Europe including England, Scotland, Wales, Switzerland, Poland, Estonia, Sweden, the Republic of Ireland, Germany, Denmark, Finland, Greenland, and Iceland; northern United States including upstate New York, eastern Massachusetts, New Hampshire, Michigan, Minnesota, Wisconsin; Canada, Russia, Patagonia, and even a few drumlins forming below the ice in West Antarctica.


All that changed when researchers in Iceland accidentally spotted the world's only active drumlin field while out on an entirely unrelated glaciological field expedition. When the Múlajökull glacier retreated, it left a fan of 110 drumlins behind. One of the drumlins was cut through by a meltwater stream, giving the researchers access to its interior. From looking at the sedimentological structure, they confirmed this was a modern, recently-created drumlin from between when the glacier surged forward in 1992 and when it retreated in 2009. The drumlins closest to the glacier were narrower and taller, while the drumlins farthest from the glacier were wider, shorter, and lower relief.

The location of currently-exposed drumlins (black outlines) line up with long crevasses where the glacier previously covered the area in this 1995 air photo. Image credit: Johnson et al.


From their field observations, the researchers suggest that the Múlajökull drumlin field formed mostly by bed deformation, lodgement, and subglacial erosion, not meltwater outwash. In their hypothesis, the fan-shaped glacier developed radiating cracks as it surged forward. The crevasses resulted in stress differences, with slightly more erosion between the cracks and slightly more till deposition directly under them at the future drumlin-locations. Once the slight hills formed, they in turn created stress in the ice sheet every time it flowed over that location, generating cracks in a positive feedback loop, increasing the size of the drumlins during each surge. In a later study, they suggested that the drumlins closest to the glacier have probably been worked over by more glacial surges, while those farthest away are least developed.

Where a 1997 seismic survey had nothing anomalous, a 2004 seismic survey picked up the reflection of a 10 meter high, 100 meter wide lump that may be a proto-drumlin growing under the ice. Image credit: Smith et al.


Yet, this is just one case study, and not enough to crown any formation theory as the triumphant winner of drumlin genesis. Between 1997 and 2004 geophysical surveys, a like a mound 10 meters high and 100 meters wide grew under the ice West Antarctica. In one of those tiny silver linings on the edge of catastrophe, the anticipated loss of the West Antarctic ice sheet might soon expose another active drumlin field, giving us a second case study of recently-formed drumlins to investigate.

A better understanding of how drumlins form could lead to better reconstructions of the glaciological history of the region. If drumlin formation is related to rapid surges forward by glaciers, they could also give us greater insight into the speed of ice flow along with the existing hints of directionality. This isn't just an academic exercise: understanding how a glacier surged and flowed would help us understand the formation and origin of glacial deposits, which gets downright interesting when we find a bit of gold and want to track down the main deposit.