The bearing of this partial parallelism of the crystals on shearing and foliation is supposed to reside in the fact that a crystal of ice is made up of a series of plates arranged at right angles to the principal axis of the crystal. These plates may be likened to a pile of cards, the principal axis being represented by a line vertical to them. If a cube be cut from a large crystal of ice, it will behave much like a cube cut from the pile of cards. If the cube be so placed that its plates are horizontal (Fig. 291_a_), and if it be rested on supports at two edges and heavily weighted in the middle, it will sag, the plates sliding slightly over one another so as to give oblique ends, but in this case the cube offers considerable resistance to deformation. If the cube be so placed that the plates stand on edge, each reaching from support to support (Fig. 291_b_), it will offer very great resistance to deformation; but if the plates be vertical and transverse to the line joining the supports, as in Fig. 291_c_, the middle portion will sag under very moderate weighting by the sliding of the plates on one another, and in a comparatively short time the middle portion may be pushed entirely out, dividing the cube. These properties have been demonstrated by McConnel[135] and Mügge, and they appear to throw light on certain phases of the action of glaciers that are most pronounced in their basal parts, and are best illustrated in arctic glaciers. _The Probable Fundamental Element in Glacial Motion._