O. D. von Engeln — Studies on Ice Structure. 463 



only one-third that of the first. In the first case the load is 

 borne by the columns of basal plates of the ice crystals, in the 

 second it is transmitted through the mass along the adjacent 

 parallel planes of the crystal prism faces. That the inter- 

 locking, irregularly oriented, crystals of glacier ice supported 

 loads of intermediate value lends further confirmation to this 

 deduction. In this connection it is interesting to consider the 

 experiment with the pond ice tinder compression in the 

 unyielding cylinder which resulted in the complete recrystal- 

 lization of the ice with principal axes normal to the pressure 

 direction, involving a rotation of 90°. If it be assumed that 

 films of water were produced by non-uniform pressure or 

 shear melting, in the initial adjustment of the mass to the load, 

 then, since ice is of less density than water, it would follow that 

 the stable condition under pressure would be such that the 

 load was carried by the water, and that the water be of minimum 

 volume and maximum bearing surface. This condition seems to 

 have demanded the reorientation of the ice crystals with 

 principal axes normal to the pressure direction. 



Ice exists under natural conditions at temperatures near its 

 melting point, hence change of phase due to increasing 

 pressure or temperature is readily initiated and the velocity 

 of transformation is quite high. Furthermore, as suggested 

 in an earlier paragraph, residual water molecules may coexist 

 with the polymerized ice molecules at temperatures near the 

 freezing point. These relations would appear to have an 

 important bearing on the plastic yield obtained in our experi- 

 ments when cubes of pond ice with principal axes parallel to 

 the pressure direction, and cubes of glacier ice, were subjected 

 to pressures approximating the crushing strength, applied 

 slowly. At the air temperatures at which the experiments were 

 made these pressures were not great enough to induce melting 

 by uniform pressure (■0072°C. per atm.) but would induce 

 internal melting by non-uniform pressure, the effect of which 

 is 12 times as great according to Johnston and Adams. As our 

 bearing surfaces could not have been ideally plane-parallel, 

 some portions of the ice must have borne a considerably 

 higher load per unit of area than was registered by the scale of 

 the press. Hence it seems quite probable that the yield 

 achieved was due to internal liquefaction of the ice. The 

 continuance of the yield, once started, with slightly diminished 

 pressure and increased surface area would be ascribable to the 

 fact that deformation of the cube would tend successively to 

 exaggerate the localization of the whole load on certain small 

 units of the bearing surface. Flow once started would also be 

 facilitated by gliding movement between the basal plates of 

 the ice crystals. On release of pressure the dissociated ice 



