ICE FORMATIONS IN MARSHES 107 



shapes. With a little imagination, I have been 

 able to distinguish a polar bear and a lamb ami- 

 ably and incongruously roaming the ice-fields to- 

 gether. These suggest the curious shapes and 

 cavernous structure seen in icebergs due to the 

 undercutting and wash of the waves. It is evi- 

 dent that undercutting by high tides is to some 

 extent responsible for the formations, but some 

 of them have not been reached by the salt-water 

 since the time they were stranded. The melting 

 of the lower, more salty layers of the ice cake 

 while the upper fresh-water and snow-ice remain, 

 are probable factors in their formation. The thin 

 ice arch shown in the illustration suggests the 

 bending of the plastic ice, but this form would 

 require lateral pressure at the two ends. It bend- 

 ing took place from gravity, a reversed arch 

 would result, as the supports are at the two ends. 

 It is probable, therefore, that the arch form is 

 due, not to bending, but to undercutting and melt- 

 ing of the lower salty ice. The arch is appar- 

 ently a later stage of the cavern. 



As the ice melts from the marsh the large cakes 

 endure the longest, and of these the ones that are 



