Charlksworth — Glacial Geology of North-JVest of Ireland. 235 



VII. — Eetreat of the Donegal Ice. 



"With the passing of the maximum stage of glaciation, changes occnrvetl in 

 general in the reverse order to those whicli may be supposed to have marked 

 the waxing of tlie iee-slieet. The facts are patently displayed and sliow tliat 

 at its margin tlie ice began steadily to shrink backward, and simultaneously 

 to diminish in thickness. Lower and lower ridges and hills successively 

 ajipeared above it, and the continuous sheet gradually gave place to a number 

 of large and detached lobes and valley glaciers. During this later pjiase the 

 effects produced by the ice at its maximum extension were considerably 

 modified. This modification was due to deposition rather than to erosion. 



Though it is impossible in all areas to trace a connected liistory of the 

 glacial events during this valley glacier phase, yet in most regions this can 

 be done with striking precision. Even in the other areas, where evidence is 

 less plentiful, there are clear indications of the late-glacial changes. 



The determination of the mode of the retreat during the different stages is 

 based upon the evidence supplied by moraines and lake phenomena with 

 their temporary drainage channels. The principles governing their formation 

 were described in Professor P. F. Kendall's paper on " A System of Glacier- 

 Lakes in the Cleveland Hills."^ 



Of the true character of the great majority of these overflow valleys, now 

 either quite dry or occupied by insignificant streams, no doubt can be enter- 

 tained ; their appearance, location, and non-accordance with the pre-glacial or 

 present drainage most clearly .demonstrate their mode of origin. Other valleys 

 may be due to similar marginal streams, but where differential ice erosion has 

 so strongly influenced the detailed topography as in the Donegal Highlands, 

 and where beds of different powers of resistance alternate so rapidly, it would 

 be hazardous in the extreme to select from these any which were used as 

 temporary glacial drainage cliannels. These have been omitted from the 

 following description. 



Owing to the dissected character of the country, direct overflows are 

 plentiful, though the marginal type of channel is the more common. The 

 pre-existing drainage lines were also frequently followed, and in consequence 

 considerably deepened and enlarged. 



Except in the Sperrin Mountains, there are but few occurrences of accumu- 

 lations suggestive of a beach or deltaic origin. 



Where a thick turf cover conceals the outcrop of rocks of unequal 

 hardness, it is frequently tempting to interpret these surface irregularities 

 as due to original inequalities in the deposition of the drift. This difficulty 



' Quart. Journ. Geol. Soc, vol. Iviii (1902), p. 471. 



