36 METEOROLOGY 



be seen low rounded clumps of pure white needle-like ice crystals of plano-convex 

 shape, with the convexity uppermost. These masses were from a few inches to a foot 

 or so in diameter, and from a few inches to about 6 inches thick. Their appearance 

 was somewhat as follows : 



jof^. 



Fig. G. moss ICE OR SPIRACLE ICE 



Formed by crystallisation of vapour, from sea water at bottom of crevasse, 



transpired throngli snow lid of crevasse 



We crossed the Drygalski Barrier early in December, and this spiracle ice was 

 obviously of quite recent growth. The crystals evidently developed at spots where the 

 lid of the crevasse was so thin as to be capable of being used as a respirator. Air would 

 alternately filter into or filter out of the crevasses through these " respirators," 

 beautiful vertical needles of ice growing upwards from the upper side and downwards 

 from the lower sides of the respirator. The probable explanation of these tufts of 

 moss ice is that on this part of the Drygalski Barrier the lower jjortion of the crevasse 

 may reach down to sea-level, and even be filled with sea water, or, at all events, have 

 a temperature near the freezing-point of sea water. Thus the relatively warm vapour 

 would rise from below and be chilled higher up in the crevasse with dejsosition of ice 

 needles on the walls and at respirators. At " Priestley Shaft " at the Blue Lake, Cape 

 Royds, a similar phenomenon was observed. The vapour from next the bottom of 

 the lake formed ice crystals on the walls of the upper part of the shaft. 



In part this moss ice may be due to convection currents under the lid of the 

 crevasse, due to differential solar heating ; but the former hypothesis seems more 

 probable. 



That an appreciable amount of vapour is present in Antarctic crevasses is 

 proved by the fact that in summer, at all events, from near Mount Nansen 



