20 On some Phenomena observed on Glaciers. 



rock. It is enough that its surface, during the whole or 

 greater part of the year, should be covered with ice, to bring 

 down the mean annual temperature of its interior materially 

 below the temperature due to its elevation, and which it 

 tvonld have were it not so covered. Conceive, now, a mountain 

 whose summit is in this predicament, viz. constantly main- 

 tained at a mean temperature below that due to its elevation. 

 This intense cold will not break off at the level of the line of 

 perpetual snow, which is determined by the mean tempera- 

 ture of the atmosphere due to elevation, but will be propa- 

 gated downwards in the interior of its mass. Hence, if, at a 

 short distance below the line of perpetual snow, where the 

 mean diurnal temperature of the exposed part, taken at a few 

 feet or a few yards deep in the soil or rock, is a little above 

 freezing, we drive an adit, or take advantage of a natural fis- 

 sure, to obtain the internal temperature at a much greater 

 depth from the surface ; we ought to find it below 32°, and 

 ice ought constantly to form in such cavities. 



But even when the summit of a hill is not covered with ice, 

 and when, therefore, this particular principle does not apply, 

 it is easy to see, on the same general grounds, that something 

 of the same kind may obtain. It is obvious, that whenever 

 a change of temperature on the surface of a solid takes place, 

 a wave of heat or cold, as the case may be, will be propagated 

 through its substance ; and if the changes be regularly peri- 

 odic, the waves will be also. Moreover, it is clear that the 

 longer the periods of the external fluctuations are supposed, 

 the greater will be the interval of the waves, so as to make 

 the time taken for the propagated heat to run over them pre- 

 cisely equal to the period of fluctuation. Now the rapidity 

 with which successive waves of heat and cold destroy each 

 other is inversely as the intervals, and thus the fluctuations 

 of temperature, depending on long periods of external change, 

 will be propagated to greater depths than those arising from 

 shorter periods, nearly in the ratio of the lengths of the pe- 

 riods. Thus the depths at which the annual fluctuations of 

 temperature cease to be sensible will be between 300 and 400 

 times greater than those at which the diurnal ones are neu- 

 tralized. Now it may happen, from the slowness of propaga- 



