According to some theories, subglacial deformation of sediment is the process of sediment transport most responsible for drumlin formation. If so, strain indicators in the sediment should yield deformation patterns that are systematically related to drumlin morphology. Clast fabrics have been used most commonly to make inferences about strain patterns in drumlins but with a wide range of sometimes divergent interpretations. These divergent interpretations reflect, in part, a lack of experimental control on the relationship between the state of strain and resulting fabrics.

Herein, fabrics determined from the anisotropy of magnetic susceptibility (AMS) of till within selected drumlins of the Green Bay Lobe are used to study the role of bed deformation in drumlin formation. AMS fabrics are a proxy for fabrics formed by non-equant, silt-sized, magnetite grains. Unlike past fabric studies of drumlins, laboratory deformation experiments conducted with this till provide a quantitative foundation for inferring strain magnitude, shearing direction, and shear-plane orientations from fabrics determined from principal directions of magnetic susceptibility (k₁, k₂, and k₃). Intact till samples were collected from transects in seven drumlins in Dane, Dodge, Jefferson, Waupaca, and Waushara counties of south-central Wisconsin, by both exploiting five existing outcrops and collecting 42 89 mm-diameter cores and sub-sampling them. Overall, ~2800 samples were collected for AMS analysis, and 112 AMS fabrics were computed.

Much of the till sampled (84% of fabrics) has k₁ fabric strengths weaker than the lower 95% confidence limit for till (S₁< 0.82) sheared to moderate strains (~10), suggesting the till has been deformed but to strains too small to indicate that bed deformation was the principal till transport mechanism. Three of five drumlins studied have k₁ fabric orientations that are not symmetrically disposed about the local flow direction indicated by drumlins. Rather, these fabrics are oriented 7-25° to the southeast of the drumlin orientations, consistent with reinterpreted microfabric data collected from nearby drumlins (Evenson, 1971). Furthermore, in all drumlins, orientations of shear planes inferred from principal susceptibilities deviate markedly from the local surface slopes of drumlins, with a 23.8° average difference between the poles to inferred shear planes and to local slopes. We infer that the drumlin fabric was set by basal till deformation during glacier flow to the southeast prior to drumlin formation and that drumlinization did not significantly reset the fabric. Thus, these drumlins are inferred to have been formed by differential erosion of a pre-existing till layer by processes unrelated to bed deformation.