ANTAKCTIC VOYAGE OB" THE BELGICA. 381 



and '" regenerated glaciers.'"' The idea of a glacier thus presupposes 

 the presence of a valley. This idea is a mistaken one, for it is quite 

 possible that the ice stream ma}' be wanting. Such is the case in the 

 antarctic whenever it happens that the collecting ground is sufficiently 

 near the coast for the ghicier to terminate at its greatest breadth in an 

 ice wall. In the antarctic regions perpetual snow can exist on level 

 ground in so low a latitude as 65°, so that even small islands may bear 

 a complete mantle of perpetual snow. On some small islets of less 

 than a mile in diameter we found a thick accumulation of ice entirely 

 covering the inequalities of the ground, and forming in consequence 

 convex glaciers. These ice caps ended seaward in perpendicular 

 walls, while on the surface they took the form of huge, perfectly cxen 

 sheep's backs. 



It is evident that this form of glacier will be found also on islands of 

 larger extent, whenever the relief is sufficiently uniform to make it 

 impossible for a peak to pierce through the glacial cap. As regards 

 the thickness of these caps, it is plain that it depends on the plasticity 

 of the ice and the extent of ground on which it rests. To my mind 

 the only difference which exists between these convex glaciers of the 

 antarctic and the inland ice of Greenland consists in the incomparal)Iy 

 greater extent of the latter, and in the fact that this does not reach 

 the coast, ])ut melts up into streamlets, and sends glaciers down toward 

 the sea only through the valleys. But it is possible that there may be 

 a sheet of inland ice more extensive than that of Greenland. We may 

 say that the great ice cap supposed by Croll" may quite well cover the 

 antarctic continent, since even small islands are seen to have the even 

 and convex covering of ice laid down l^y ('roll for the whole southern 

 continent.'' 



On the other hand, it may seem surprising that the glacial caps are 

 not the sole type of glacier in these regions, where the line of per- 

 petual snow is found at sea level." The reason is that most of the 



"Climate and Time, 4tii ed. (London, 1897), p. 374. 



•"Cf. Arctowski, " Les calottes glaciaires des regions antarctiques," Comjites Kendus 

 Acad. Sci. Parin, December 24, 1900. 



'The question of the level of perpetual snow in the region of Belgica Strait is a 

 very complex one. Professor Penck, who was present at an address that I delivered 

 at the "Naturforscher-Versammlung" at Aix-la-Chapelle, was tempted to suppose 

 that there might well be two lines of perpetual snow, one above the other, in that 

 region. Low-lying fogs are, in fact, very frequent there, and these protect the snow 

 from the effects of solar radiation, while, on the other hand, the clouds which most 

 frequently give rise to atmospheric precipitation likewise re.st very low. The sum- 

 mits and upper portions (jf the flanks of the mountains (1,000 feet and over) are 

 therefore subject to a climatic regime decidedly different from that which prevails at 

 sea level. The mean temperature of the air is possibly lower, but, on the other hand, 

 the amount of atmospheric precipitation is less and the effect of radiation greater. 

 This would explain the fact that the mountain slopes are sometimes bare of snow at 

 an altitude of 1,500 feet or even higher. It follows that the idea of two levels of 

 perpetual snow is quite a plausible one, 



