Ice in the Sea 249 



Petroleum is particularly suitable as an immersion liquid because it cannot penetrate 

 into the small air-filled channels of the ice pieces. The results of these determinations 

 are summarized in Table 93, The conspicuous result is the very small variation in the 

 density of young ice, in spite of the strongly varying salinity of the samples and of the 

 equally variable depth from which they were taken, as well as of the changing thickness 

 of the floes. The smallest values (0-914 and 0-916) were given by two thin and highly 

 saline young ice floes which had been formed at very low temperatures. The rapid 

 freezing must of course have trapped a large number of air bubbles, probably more 

 than normal. The uniformity of the values between autumn and winter disappears 

 gradually in spring as melting becomes more and more eff"ective. There is a progressive 

 fall in density in late spring, and this decrease becomes stronger as the disintegration 

 of the ice proceeds during the summer. Values less than 0-90 show by the large number 

 of enclosed air bubbles that the ice must have been exposed to the sun ("gesommert"). 

 The lowest value in density was found at the top peak of a large floe. During the pre- 

 ceding summer the salinity in this ice had been completely removed, and in winter the 

 melting water of it could be used for drinking water. 



(c) Thermal Properties of Sea Ice and the Temperature in the Interior of Ice Flow 



It is characteristic of sea ice that its thermal properties such as specific heat, latent 

 heat of melting and thermal expansion behave quite abnormally. During investiga- 

 tions of the heat expansion of sea ice Pettersson (1883) found that highly saline sea 

 ice expanded with decreasing temperature down to — 20°C, though for ice of lower 

 salinity this temperature was considerably higher. Malmgren showed by investiga- 

 tions during the "Maud" Expedition that Kriimmels' assumption, that this was due 

 to the salt solution enclosed in the ice, was correct. This abnormal behaviour rela- 

 tive to the specific heat, latent heat of melting and thermal expansion is thus also a 

 consequence of the formation and melting of pure ice occurring in the interior of 

 sea ice. At a temperature r, 1 g of sea ice of salinity l%o will contain a-r g of pure ice 

 and (1 — At) g of salt solution. If the specific heat of sea ice at the temperature t 

 is Ct then this quantity of ice for a temperature change dr will require a quantity of 

 heat Crdr. It is made up essentially of: (1) the rise in temperature Orcdr of pure ice 

 (with specific heat c); (2) the rise in temperature of the salt solution (1 — a-^Kdr 

 (with specific heat k); and (3) the heat Kda^ required to melt da-r g of ice (with latent 

 heat of melting A^). This gives the equation 



Cr = a-rC + (1 — a-)K -f Xr -^ . 



dr 

 Since the second term is small and as a first approximation a^ = 1 then 



Ct = c + A, — -. 

 dr 



For sea ice of salinity 5'%o the variable amount of ice is Sda-r and therefore one ob- 

 tains for it the relation: 



c.^c^-Sx/--^. 



dr 



