CHAP. xvm. FLOATING OF BOULDERS ON ICE. 361 



c, and the intervening valleys, the hj-pothesis of glaciers is 

 out of the question. I conceive, therefore, that the erratics 

 Avere conveyed to the places they now occupy by coast-ice, 

 when the country was submerged beneath the waters of a 

 sea cooled by icebergs coming annually from arctic regions. 



Fiff. 53. 



d, e, Masses of floating ice carrj'ing fragments of rock. 



Suppose the highest peaks of the ridges A, b, c, in the an- 

 nexed diagram, to be alone above water, forming islands, and 

 d e to be masses of floating ice, which drifted across the 

 Canaan and Richmond valleys at a time when they were 

 marine channels, separating islands, orrather chains of islands, 

 having a N.JST.E. and S.S.W. direction. A fragment of ice such 

 as d, freighted with a block from a, might run aground, and 

 add to the heap of erratics at the N.W. base of the island (now 

 ridge) b, or, passing through a sound between b and the next 

 island of the same group, might float on till it reached the 

 channel between b and c. Year after year two such exposed 

 cliifs in the Canaan range as d and e of the map, fig. 50, 

 p. 357, undermined by the waves, might serve as the points 

 of departure of blocks, composing the trains Nos. 5 and 6, 

 It may be objected that oceanic currents could not always 

 have had the same direction; this maj'^ be true, but during a 

 short season of the year when the ice was breaking up the 

 prevailing current may have always run S.E. 



If it be asked why the blocks of each train are not more 

 scattered, especially when far from their soui'ce, it may 

 be observed that, after passing through sounds separating 

 islands, they issued again from a new and narrow stai'ting- 



