NO. 17.] 



THE TEMPERATURE OF THE POLAR ICE. 



561 



The numbers for the air are the monthly means for the respective years 

 given on p. 483. As the figures for the surface are rather uncertain, I have 

 not made use of them in the following discussion. 



As it will be seen, the annual series of the temperatures for the different 

 depths is far from being homogeneous. We have some months for 3 years, 

 some for 2 years, and some for only one year. An attempt to reduce them 

 to the means of 3 years proved a failure. Generally the temperature of the 

 ice rises and falls with the temperature of the air, but I tried in vain to find 

 any rational coefficient of reduction, and in some months — for instance February 

 (and November) — the temperature of the air and the temperature of the ice 

 go in opposite directions. 



I have therefore taken the monthly means as they stand, and computed 

 the annual periods of the temperature at the different depths directly for them. 

 The want of homogeneity in the figures for the different months is thereby 

 to a certain extent eliminated, because all the 12 numbers enter into the 

 determination of the constants expressing the annual period. The weakest 

 months are August and September, but happily the mean temperatures of 

 the air for these months in 1894 and 1895 do not deviate from their general 

 mean more than 0'°7. 



The formula for the annual period which I have used, is 



t„ = M-\- a . sin {A + m) 



tm being the mean temperature for the month, M the mean annual tempera- 

 ture, a the half amplitude or range, A the phase-angle, and in the angle 



71 



