[cooKE] VARIATION IN THE DENSITY OF ICE 131 
where freezing mixtures of extremely low temperatures are used, than 
where these are of a mild character, such as salt and ice. 
The question very naturally presents itself now as to whether 
there is any species of ice which one would expect from «a priori reason- 
ing to be free from internal strains, and thus to have a normal density. 
In the case of the formation of a sheet of ice on a pond or river, 
the ice starts at the edges, gradually extending out until the forma- 
tion joins in the centre and the water is completely covered; all the 
ice formed subsequently is in the form of layers on the under surface 
of this sheet of ice. Now, when the ice is extending from the shores, 
before the sheet of water is completely frozen over, the formation is 
continually becoming thicker, and hence, as the heat from the water 
is being conducted away through the ice, the temperature of the ice 
is falling, and its volume is becoming less, the contraction being a 
two dimensional one. When the extending sheet of ice has com- 
pletely covered the water, the superficial dimensions of the sheet 
become permanently fixed, and so the ice cannot respond to any change 
in temperature by a corresponding change in superficial dimensions. 
As this ice sheet becomes thicker, its temperature falls, and it con- 
tracts, but only in a direction at right angles to the plane of formation. 
A little consideration will show that the contraction or expansion in this 
direction, provided it be practically uniform over the whole sheet, 
cannot set up strains between the different layers of ice; and so the 
layers which are formed after the superficial dimensions of the ice 
sheet are fixed, should possess a normal density; and also the pre- 
viously formed layers, when a uniform temperature of zero is attained 
throughout the mass, should be under a strain of compression acting 
in the plane of formation, and so should show a high density. 
As far as icicles are concerned, nothing very definite can be said. 
li the formation were always kept completely wet by the dripping 
water, then the icicle should possess a normal density; but if the 
forming mass were only partially wet, or only wet on one side, then 
its average temperature would fall below zero, and the newly forming 
ice would grow upon a contracted support, which it would tend to 
compress when a uniform temperature was established, and hence the 
density would rise above the normal value. As a general rule, icicles 
are not kept completely wet during formation. 
When dust-free water is cooled below zero and then agitated, 
exceedingly fine and translucent mixture. It seems highly probable that 
ice so formed should possess a normal density, as the formation is so 
finely distributed that the presence of any internal strains seems 
inconceivable. 
