For the exact determination of the direction of ice movement, recourse must be had to the striæ on the bed-rock. Were the striated rock surface perfectly plane, and were the striæ even lines, they would only tell that the ice was moving in one of two directions. But the rock surface is not usually perfectly plane, nor the striæ even lines, and between the two directions which lines alone might suggest, it is usually possible to decide. The minor prominences and depressions in the rock surface were shaped according to the same principles that govern the shaping of hills (Fig. 29) and valleys (Fig. 30); that is, the stoss sides of the minor prominences, and the distal sides of small depressions suffered the more wear. With a good compass, the direction of the striæ may be measured to within a fraction of a degree, and thus the direction of ice movement in a particular place be definitely determined. The striæ which have been determined about Baraboo are shown on Plate II. _Effect of topography on movement._--The effect of glaciation on topography has been sketched, but the topography in turn exerted an important influence on the direction of ice movement. The extreme degree of topographic influence is seen in mountain regions like the Alps, where most of the glaciers are confined strictly to the valleys. As an ice sheet invades a region, it advances first and farthest along the lines of least resistance. In a rough country with great relief, tongues or lobes of ice would push forward in the valleys, while the hills or other prominences would tend to hold back or divide the onward moving mass. The edge of an ice sheet in such a region would be irregular. The marginal lobes of ice occupying the valleys would be separated by re-entrant angles marking the sites of hills and ridges. If the ice crossed a plane surface above which rose a notable ridge or hill, the first effect of the hill would be to indent the ice. The ice would move forward on either side, and if its thickness became sufficiently great, the parts moving forward on either side would again unite beyond it. A hill thus surrounded by ice is a nunatak. Later, as the advancing mass of ice became thicker, it might completely cover the hill; but the thickness of ice passing over the hill would be less than that passing on either side by an amount equal to the height of the hill. It follows that as ice encounters an isolated elevation, three stages in its contest with the obstruction may be recognized: (1) the stage when the ridge or hill acts as a wedge, dividing the moving ice into lobes, Fig. 34; (2) the nunatak stage, when the ice has pushed forward and reunited beyond the hill, Fig. 35; (3) the stage when the ice has become sufficiently deep to cover the hill. [Illustration: Fig. 34.--Diagrammatic representation of the effect of a hill on the edge of the ice.]