Flutes are low-relief, elongate landforms that form subglacially and parallel to the glacier flow direction. They usually consist of till and have boulders at their heads. Flutes can be pervasive in the forefields of glaciers and contain information about the role of bed deformation in basal slip. There are two leading hypotheses for their formation that involve cavity formation in the presence of a deformable bed: the freeze-on hypothesis, in which till that flowed into a cavity downstream from a boulder is frozen into the glacier sole, transported downstream in ice, and then redeposited at the downstream end of the flute, and the cavity–propagation hypothesis, in which till flows into a cavity that begins in the lee of a boulder but propagates downstream in the lee of the flute as it builds downstream with time.

Five flutes were studied at two temperate, surge-type outlet glaciers in Iceland, Múlajökull and Breiðamerkurjökull, by measuring fabrics based on anisotropy of magnetic susceptibility (AMS) and till matrix densities to infer past strain patterns. To increase the robustness of AMS measurements, a new method for characterizing AMS ellipsoids, calibrated to laboratory ring-shear fabrics, was used, and the magnetic mineralogy of the till was determined with a series of geomagnetic tests, many of which have not before been applied to tills. Flute pebble fabrics from the literature were also re-analyzed.

The magnetic susceptibility of the Múlajökull and Breiðamerkurjökull tills is dominated by pseudo-single-domain titanomaghemite and magnetite, respectively, allowing for a straight-forward interpretation of AMS fabrics. When referenced to a single flute orientation, both AMS fabrics from this study and pebbles fabrics from the literature show that convergent fabrics dominate flutes. Locally, however, AMS fabric orientations are highly variable and had shapes that indicate low-to-moderate shear strains (less than, or not much in excess of, ~7). More importantly, till matrix densities, which are a proxy for past maximum effective stress, are significantly higher in the middle of flutes than at their sides. The difference in densities across the width of flutes was much larger in a parallel-sided flute than in a tapered one.

Together, these data indicate that flutes form by cavity propagation (e.g., Benn (1994b)) that requires unsteady, subglacial water pressures and sliding speeds for fluted till to strengthen sufficiently to propagate a cavity downstream. Flutes are initiated and grow through flow of weak till into cavities during periods of high subglacial water pressure and sliding speed. Newly accreted till at a flute's end is then compacted and hence strengthened during a subsequent period of lower subglacial water pressure and sliding speed when water pressure in the leeward cavity falls and the stress that ice exerts downward on the till increases. This effect accounts for high till density near the middle of flutes where cavities were longest and stress increases largest when water pressure fell, and also accounts, through strengthening of till, for the preservation of transverse fabric components in an environment dominated by flow-parallel shear. The strengthened till provides the rigid take-off point for a cavity during subsequent glacier acceleration and thereby allows the flute to grow downstream. Long, parallel-sided flutes and short, tapered flutes are likely end-members of a continuum and represent a high and low degree of till strengthening, respectively, during periods of low water pressure. Flute formation and growth may be influenced by the hydraulic diffusivity of till, and therefore by its texture, which controls the rate and magnitude of effective stress increases during decreases in basal water pressure.