KKNDALL : OBSKHVATIONS ON THK (JLACIEH LAKKS. 
41 
a transverse watershed. This was the significant feature upon 
which I based my conclusion, and for its explanation it is necessary 
to recall the conditions attending the cutting of a lake channel 
parallel to the ice-front. Let us imagine a lake held u]) by an 
ice-barrier, and draining across a col into another lake at a lower 
Fig. II. 
level. 8up])ose the section across the spur at the col to be semi- 
circular (Fig. II.). The watershed will at first be on the summit of 
the semicircle, but as the overflowing stream wears a channel for 
itself, the watershed will gradually advance towards the higher 
lake. Now reverse the flow, and the watershed will letreat 
towards the centre, and peiliaps ))ast it. as indicated by the 
broken lines in the figure. This is the ])ositi()n of the watershed 
in the little channel referred to. 
(2) Relative Duration of the Stage of Retreat of the Ice-sheet. — 
Another subject upon which I think it desirable to add some 
observations is the evidence afforded through the district of the 
relative duration of the stages of letreat of the ice and of a geneial 
but brief readvance. 
The relative rate of retreat and the duration of the lialts may 
be judged by the magnitude of the moraines, when present, and 
of the overflow channels. Moraines are very imperfectly de- 
veloj^ed in the area, but the small patch of gravel on Gallows Hill, 
in advance of the Wykeham moraine (Q.J.G.S., Vol. LVIII., 
p. 559). seems to indicate that the lobe of ice which entered the 
Vale of Pickering stood for a short time at, or beyond, that 
point, then retreated to make a long halt at Wykeham and 
throw down the great moraine mass upon which the Abbey 
stands. 
The evidence of the overflow channels is much clearer and 
more decisive. 
