298 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1926 
The wind observations by pilot balloons confirm this result. At 
the ice the observed wind velocities were always small, undoubtedly 
on account of the great resistance offered by the rough ice, but 
above the inversion, where the warmer air was sliding over the cold 
layer, strong winds were met. 
The temperature distribution here described was always present 
in winter, independent of the direction from which the wind was 
blowing. Considering this and the uniform meteorological condi- 
tions over the Polar Sea, it seems justified to conclude that in winter 
the whole Polar Sea is covered by a thin layer of cold air which, to 
a great extent, is isolated from the atmosphere above it. Such 
conditions are possible on a frozen sea, which, disregarding the 
roughness of the ice, has the character of a vast plane. A sharp sur- 
face of discontinuity can exist over a vast plane even when the wind 
is blowing, but it can not exist over a mountainous continent because 
it would soon be broken up on account of the differences in elevation. 
Since the cover of cold air is isolated from the free atmosphere 
above it, the temperature of this cover must depend, to a great 
extent, upon the temperature of the ice-surface with which it is 
in contact. Particularly, the lowest temperatures of the air must 
correspond closely to the lowest temperatures of the surface. Dur- 
ing the six winters I have spent on or off the Siberian coast the 
minimum temperature always has been close to 50° below zero, F. 
There must be some reason why this limit is reached but not 
passed. The answer seems very simple. The surface of the ice, 
which is covered by a very thin layer of hard snow, loses heat 
by radiation to space at night. The temperature would sink to 
very low values during the long, continuous winter night if this 
loss were not compensated in some way. It is compensated. Heat 
is constantly conducted through the ice to the surface from the 
underlying sea water, which has a constant temperature of 29° 
above zero, F., the freezing point of the sea water. The amount 
of heat conducted to the surface increases when the temperature of 
the surface sinks, but the amount of heat lost by radiation decreases 
at the same time. Loss and gain, therefore, must equalize each 
other at a certain temperature, and when this limit is reached the 
temperature of the surface can not sink any further. 
We have made extensive measurements of the heat lost by radia- 
tion and the heat conducted through the ice, and have found that 
loss and gain, on the average, compensate each other at about —40° 
F. and at about —50° F. under exceptional circumstances. The con- 
ditions seem, therefore, to be actually as simple as assumed. The 
minimum temperature of the air is reached when the surface receives 
as much heat from the sea as it loses by radiation to space. 
