distance between these temperatures would be less, but the water particles would keep their respective places.
When water freezes, it does not appear that this process is continuous, for many of the characters of the ice seem to show that it is intermittent; i. e. either a film of ice is formed, and then the process stops until the heat evolved by solidification has been conducted away upwards, and the next stratum of water has been sufficiently cooled to freeze in turn; or else the freezing being, so to speak, continuous, still is not continued at the same constant rate, but, as it were, by intermittent pulsations. Now it may well be, when a layer next the previously -formed ice, and containing an undue proportion of salts, has been cooled down to its required temperature for freezing (which would be below 32°), that on freezing, the congelation will pervade at once a certain thickness of the water, excluding the salts from the larger portion of ice formed, but including them as a weak solution within its interstices. The next increment of cold conducted from the ice above would freeze up these salts in the ice containing them, at the same time that a layer of pure ice was formed beneath it. Thus a layer of ice fusible at a lower temperature than the ice either above or below it might be produced; and by a repetition of the process many such layers might be formed.
It does not follow necessarily that the layers would be perfectly exact in their disposition. Very slight circumstances tending to disturb the regularity of the water-molecules would be sufficient, probably, to disturb the layers more or less. lee contains no air, and the exclusion of a minute bubble of air from the water in the act of freezing might disturb the direction and progress of the Congelation, and cause accumulation of the extra saline liquid in one spot rather than another. So might the tendency to the formation of little currents, either arising from the separation of the saline water from the forming ice, or from the elevation of temperature in different degrees at those places where the Congelation was going on at different rates.
The effect would not depend upon the quantity of salts contained in the freezing water, though its degree would. The proportion of salts necessary to be added to pure water to lower its freezing-point 1° Fahr. may be very sensible to chemical tests, but the proportion required to make the difference 1100th
