(This paper in addition to its main theme gives observational evidence of plasticity and faster flow in the lower strata of a glacier under pressure. The conditions it describes provide additional Support for the Extrusion Flow hypothesis. Ed.)

Roches moutonnées, the rounded rock hummocks found in glaciated regions, vary somewhat in shape according to the nature of the rock, its stratification or cleavage planes and the action of the ice. Usually they have a smooth, rounded back pointing uphill, while their downhill face is rough and often steep.

There are many theories to account for their development. Some authorities attribute them to highly resistant rock masses which have persisted after the erosion of the rest of the glacier bed; others believe them to have been rugged protuberances of pre-glacial times which the glacier has not been able to remove entirely.

I believed that controversy could perhaps be stilled by direct observations in the bed of a living glacier. In the years 1940,1941 and 1942 I several times found it possible to make my way from 20 to 50 m. down into the interior of both the Upper and Lower Grindelwald Glaciers by means of the marginal crevasses between the ice and the glacier banks. The photographs in Figs. 1 and 2, PI. II, were taken on the Upper Glacier. In 1941 a System of subglacial caves led back from B (Fig. 1, PI. II) to a point below the camera. The arrow in Fig. 2, PI. II, marks the entrance to another series of caverns. By making my way into these I was able to reach the bed of the ice stream and there to watch the physical state of the ice and its action upon the rock.Footnote *

Fig. 7. Diagrammatic representation of a roche moutonnée forming under a living glacier. The hatching indicates the area of semi-fluid conditions

N pressure of superincumbent ice upon eroding stone

dd, hydrostatic pressure

n, reduced pressure upon stone

R, frictional resistance

r, reduced frictional resistance

Vn, normal speed of ice-flow

Under a pressure equivalent to a depth of 50 m. or less the glacier ice is rigid and brittle; it fractures easily on bending (see Fig. 3, PI. II). If it flows over a large rock it does not immediately re-conform to the shape of the glacier bed after it has passed the obstruction (see Figs. 4 and 5, PI. II). Instead a cavity is formed through which the observer may succeed in penetrating. It is more difficult to force access to the uphill side, but here too, by unusually good fortune, I was able to make a few observations and even measurements.

The problem was: did the upper layers of ice, which I found to be moving at the rate of 36.8 cm. a day against the obstructing rock, glide over the lower layers arrested by it? I found that this was not the case. The lower layers, squeezed by the narrowing of the vertical section, lost their rigidity and became plastic (see Fig. 6, PI. II, also Fig. 8, p. 59). They assumed, so to speak, the character of a semi-viscous fluid. I measured their speed. The result was striking. The maximum rate of flow was 71.8 cm. a day—double the normal (see Fig. 8, p. 59).

Plate I Fig. 1 (top left). Sublimation in presence of air on a surface of Al deposited by evaporation in vacuo; an "old surface" prepared twelve days beforehand and kept in a desiccator. Frost point — 52.1°C. First ice crystals appeared at — 54.o°C. Photograph taken 15 minutes afterwards at -54.5° C. (x

Fig. 2 (top right). Same experiment 5 minutes later. When surface was cooled to — 58.0° C. supercooled water drops covered it, leaving free spaces round previously deposited ice crystals, owing to the decrease of water vapour pressure in their vicinity. To compare with Fig. 1 this photograph should be

Fig. 3 (left centre). Sublimation in absence of air. The surfaces of a thick layer of Cd and "thin" layer of Zn. When the evaporation of Cd was completed half the surface was screened longitudinally and a thin layer of Zn was deposited on the other half .Frost point — 38-0" C. Temperature of test plate -39.1° C. On the Cd side water vapour is subliming vigorously none on the Zn. Natural

Fig. 4 (right bottom). Sublimation in absence of air on Cd (left) and Zn (right) polished surfaces. Frost point of air -15.0°C.± o-1° C. Temperature of test plate when first crystal appeared -16.7° C. Ice crystals of various shapes are shown (1) and (2) similar to those observed by Rau and classified by him as cubic

Plate II Fig. I. Marginal crevasse. A C measures 160 m.

Fig. 2. Taken from B in Fig. i. Ice 40 m. thick

Fig. 3- Roll of expressed plastic ice subsequently hardened. Figs. 3-6 Flashlight

Fig. 4. The rock mass (centre) has moulded its contour on the ice (right). Camera is pointing uphill

Fig. 5. The jagged downhill -ide of a roche moutonnee. Note grooves in ice. Rule extended 20 cm. 50 m. helow surface

Fig. 6. Ice flowing over a rock. Camera pointing downhill. Layer (I) has been hressed out between an obstructing rock and the 20 m.-thick layer (II) above it

It was not surprising to find that water exuded from countless capillaries and that pieces of ice hacked out had almost the consistency of cheese. These two very different conditions of the ice—the plastic, through being dammed up on the uphill side of the obstruction, and the rigid, under normal conditions—must have a determining influence in the fashioning of rock forms.

Let us imagine that a stone the size of a man’s fist is frozen into the ice moving parallel to the direction of flow over the glacier bed. In the unyielding ice, where each grain is frozen firmly to its neighbours, it is pressed on to the rock bed by the vertical (static) pressure (N, Fig. 7, p. 58). The observations I made showed that under these conditions it can scratch striae 4 mm. deep and 20 mm. wide. But as soon as the region in front of the obstruction is reached the increased pressure causes the surfaces of the ice grains to melt. The result is that the plastic mixture of ice and water flow round the stone; the pressure is no longer transmitted downwards from ice grain to ice grain, but is communicated to the fluid which now permeates the ice. The stone is therefore subjected to what amounts to hydrostatic pressure (dd, Fig. 7) from all sides and becomes practically a floating body. The pressure upon it is therefore much reduced and with it its power to erode.

Fig. 8 Longitudinal section of Fig. 6, Pl II, showing the measured speeds of layers I and II

The sequence of processes can be likened first to heavy scratching by a strong hand followed by a more gentle bumishing touch. In this way differential erosion acts upon an uneven rock surface and intensifies any irregularities irrespective of whether these existed preglacially or were caused by contemporaneous subglacial action (breaking and washing out).

As soon as the mixture of water and ice has passed the region of pressure it becomes rigid at once, as pointed out above, and retains the shape impressed upon it by the hummock. In the subglacial hollow so formed the temperature is always close to freezing-point. It seems probable, therefore, that the melt water penetrates through crevices and cleavage planes on the downhill side of the rock, where, the pressure being released, it refreezes and bursts off fragments.

I suggest therefore that reduced erosive power on the uphill side of the rock followed by thaw-freeze processes on the downhill face are the agencies which mould the typical shape of roches moutonnees. Differential resistance of the rocks forming the glacier bed is, in my view, of secondary importance in that it only appears to modify the individual forms.