Pervasive deformation of subglacial fill has been suggested as the primary mechanism of motion and sediment transport for many modern glaciers and portions of Pleistocene ice sheets. Unfortunately, standard field observations of the geologic record are insufficient to determine whether basal fills have experienced the high strains required of this model;To establish field criteria for identifying highly sheared till, a ring-shear device is used to study both clast-fabric development and mixing between lithologically-distinct till units as a function of shear strain. In some experiments, the relation between the alignment of prolate clasts and shear strain is explored. The results indicate that a strong fabric develops in the direction of shearing at low strains and remains strong with continued deformation. Thus, clast fabric can be used to distinguish otherwise featureless tills that have been sheared from those that have not. Other experiments are performed to investigate if the degree of mixing between subglacial till layers provides a means of estimating bed shear strain. Two lithologically distinct Pleistocene tills are added as discrete layers to the ring-shear device and sheared parallel to their contact. The results indicate that mixing across the contact is diffusive and proportional to the cumulative shear strain. Using diffusion theory to model this mixing allows the maximum bed shear strain to be estimated from mixing profiles measured across the contact between these two tiffs in the field;These laboratory observations were used, together with other data, to help determine the primary flow mechanism of the Des Moines Lobe. Reconstructions of the lobe from geomorphic data reveal that motion occurred at its base, and results of consolidation tests on basal till samples indicate that the lobe was near floatation. Field measurements of fabric and mixing in the basal fill of the lobe indicate that bed deformation was not significant. This is supported by the results of a theoretical model of plowing and sliding that predicts, albeit with some uncertainty, that motion should have occurred at the bed surface, rather than at depth in the bed.