conclusion of the water and ice interaction, the underwater part of the iceberg assumes the form of 

 a truncated cone with its base at the bottom (this phenomenon partially explains the creation of 

 podsovs* and a vertical gradient of salinity is created in the water (due to the action of temperature 

 differences). Let us now assume that the underwater part of the hummock is a cone with its point 

 down and that the hummock is floating in water of uniform temperature and salinity. When the pro- 

 cess is concluded, the conical form of the underwater part of the hummock will be preserved, the 

 temperature will decrease to the temperature of freezing, and the salinity will again prove to be 

 lower in the upper layers than in the deeper ones (due to the action of the masses). 



In the examples investigated above, we have assumed that the temperature of the ice is 0°. 

 Actually, even during the summer, it is somewhat lower than the freezing temperature of the water 

 in which the ice is floating. As a result of this, the underwater part of the ice melts chiefly be- 

 cause of the heat accumulated by the water during the summer in a given region or from that ac- 

 cumulated in more southerly regions. Inasmuch as, according to formula (6) (other conditions being 

 equal) the mass of the melted ice is directly proportional to the mass of the water in contact with 

 the ice, it is natural that the melting of the part of the ice which projects beneath the level surface 

 of the ice field occurs with particular intensity if the water and the ice are in motion. This hap- 

 pens, for instance, when there are sea currents under fast ice, or during wind drift of ice. Such 

 erosion of the lower projecting parts of the ice has been noted by many observers and has a deci- 

 sive importance in isostatic phenomena (see Section 103), particularly in destroying hummocks and 

 creating level fields (Section 49). This same phenomenon explains the rapid waste of separate 

 floating ice floes during the summer in high seas. 



LITERATURE: 77. 



Section 63. Thermal Expansion 



For pure ice the coefficint of volumetric thermal expansion has an average value of many 

 measurements 



P = -^ = 0.000165, 

 '^ vdt 



and therefore, the linear coefficient of expansion is 



a = ^ = -|- = 0.000055, 



where u is the volume, i is the length, t is the temperature. 



In deriving the formula for the thermal expansion of sea ice with the same assumptions as 

 when deriving the formula for the specific heat of sea ice, Malmgren considered that the coeffi- 

 cient for the expansion of sea ice is equal to the coefficient of expansion of pure ice plus the cor- 

 rection for change in volume depending upon the formation or melting of an additional layer of ice in 

 connection with a change in temperature. 



Thus, in accordance with Malmgren's assumptions we obtain 



(1) 



where u_ is the coefficient of thermal expansion related to one gram and one degree. 



"^0.92 ^dz\^ Sj' 



*Podsou- an underwater projection of ice. 



158 



