NON-CRYSTALLINE ICE COLUMNS—MACGREGOR. 379 
eral below, the freezing point. I do not know what the exact 
variation of temperature was ; but the change was so great and 
so sudden, that while the layer of soil was still saturated with 
water, its surface must have begun to freeze. Owing to the 
expansion of the water on solidification, a continuation of the 
freezing would result in a slight elevation of the crust thus 
formed, until the thinner parts of the layer of loose earth 
had been frozen quite down to the hard ground under- 
neath. This layer would thus become divided into little patches, 
with completely frozen interspaces, and its crust would be pre- 
vented from further elevation as a whole. A continuation of the 
surface radiation, resulting in the freezing of a portion of the 
water in any of the spaces in which it must now be enclosed, 
must, owing to the expansion of the freezing water and the 
rigidity of the enclosure, raise the pressure throughout this space 
and in the frozen crust. The rise of pressure would of course 
be accompanied by a lowering of the melting point, and at those 
parts of the crust at which the pressure was greatest the melting 
point might be reduced to the actual temperature of the crust. 
At any such part, therefore, the ice would partially melt and be 
driven outwards, carrying with it the earthy portion of the 
crust, and, through the aperture thus produced, the ice would 
continue to be protruded so long as freezing continued. 
That a column of ice formed in this way would have, in gen- 
eral, a section of the same form throughout, is obvious. That the 
ice would likely be opaque, follows from the fact that the pro- 
_ trusion of a column of so brittle a substance must result in the 
formation of innumerable cracks and fissures, giving it the 
appearance of compressed snow. ‘The presence of these cracks 
enables us also to explain the upward curvature of the columns. 
For at the base of any column which is in course of protrusion, 
melting is going on at various points. At no one point will there 
be much ice melted, for a slight melting will result in reduction 
of the pressure. But here and there throughout the base melting 
will occur. The water thus formed will flow downwards into 
the little fissures ready to receive it, and filling some of them 
will then freeze, expanding during solidification. If the axis of 
