CRTSTALLOD ROMES. 
457 
into the glacier from the talus of some descending slope. 
I can not recall a case in which such fragments had 
the strictly angular character that belongs to a recent 
fracture. They were either complete bowlders, or par- 
tially rounded, as in the two preceding sketches. 
The influences of the berg 
as a raft in the translation 
of masses of rook, with their 
accompanying paste, may be 
inferred to some extent from 
the facts I have thus hastily 
thrown together. Of near- 
ly five thousand bergs which 
I have seen, there was, per- 
haps, not one that did not 
contain fragmentary rock. A walk over the berg 
would disclose them, either clinging partially imbed- 
ded in their slopes, or in the form of pebbles and still 
smaller fragments, penetrating in cylindrical cavities 
deep into the substance of the berg. 
This form of deposit was even more marked than it 
seems to have been in the glaciers of the Alps. The 
- constant daylight, without in- 
^ terruption of solar influence, 
and the absence of radiation 
during the night, will explain this. I have seen the 
surface of a berg completely covered, for perhaps a 
