42 
KENDALL : OBSERVATIONS ON THE CJLACIER LAKES. 
The maximum extension of the ice is in a large proportion 
of cases indicated by a range of very small and shallow channels, 
indicating a mere " touch and go " advance and retreat ; behind 
this there is generally to be found a large, well-developed channel, 
like that crossing Murk ]Mire Moor, above Goathland, telling of 
a prolonged halt. The next stage is a rather rapid retreat to 
a position of much greater stability when all the largest channels 
of the district were cut. This is well illustrated by the Moss 
Swang channel, near Goathland. Intimately connected with 
this stage is a slight, general readvance of the ice, shutting the 
half-formed channels and cutting others further forward, or 
deepening some previously-formed channel. The evidence upon 
wliich I conclude that there has been such a forward movement 
of the ice-front is well displayed in Fig. 10 (p. 33). Here a 
section of the long spur sloping from W. to E. down to Hayburn 
W3'ke Station shows six successive channels produced by the 
drainage of a lake on the northern side during the shrinkage of 
the ice-front from W. to E. for a distance of less than a mile and 
a half. It will be observed that the intake of the most westerly 
of these channels, Cowgate Slack, is at an altitude of 580 above 
sea-level, while the next channel eastwards, Hardhurst Slack, 
opens at 607 feet. It is manifest that, as they were both produced 
by the overflow of the same lake, the easterly one must have been 
cut before the more westerly, and this implies that, after the ice 
had retreated to the east of Hardhurst Slack, it readvanced and 
closed the channel long enough for Cowgate Slack to be cut 
below the level of the Hardhurst intake. Thus, when the ice- 
margin again retreated, Cowgate Slack remained the functional 
•drainage channel. Other cases of a similar character occur, 
which seem to me to point to a general forward movement of the 
ice-sheet round the Cleveland area. The deserted oxbow at 
Castle Hill above the Moss Swang Valley, Goathland (Plate VL), 
and those at Hardale Slack, Roxb}^ (Plate III.), Biller Howe 
Dale, Evan Howe Slack, and Castle Rigg (Plate X.), may not all 
belong to the same movement. l)ut I think that, while the Hardale 
Slack and Biller Howe Slack examples, which are clearly correlated, 
belong to an earlier and very slight readvance, those of Castle 
Hill, Evan Howe Slack, Castle Rigg, Cowgate Slack, and the 
