482 MR. P, F, KENDALL ON A SYSTEM OF fAug. 1902, 
flowing off the ice itself. The overflows, four in number, which cut 
through the Northern Cleveland watershed above the village of 
Egton are typical of this arrangement. 
(2) Severed spurs.—I have applied this name to those cases 
where the drainage of minor valleys has been carried across spurs 
of a main watershed. Such cases arise where the ice-front, being 
opposed to a main watershed, has impounded the waters of valleys 
descending towards it at an elevation such that a direct overflow 
is not available. The waters of the lake then discharge, either 
across a depression of the spur, or by a channel cut on its outer face 
along the edge of the ice. The former case differs in no essentials 
from a direct overflow, but the latter constitutes my third type. 
(3) Marginal overflows.—These are at first merely shelves 
cut in the hillside, and developed subsequently into actual gorges, 
Some marginal overflows cease to operate before reaching the 
‘corge stage,’ and in the Cleveland area every stage is represented. 
In some instances a marginal overflow had its iceward side com- 
posed of moraine, and in other instances entirely of ice. The marginal 
overflows and severed spurs often are arranged in series, which may 
be of two kinds—the aligned sequence and the parallel 
sequence. 
An aligned sequence is one in which several valleys may be 
obstructed, and drain from one to another by severed spurs, until 
free escape is offered either laterally or directly. These aligned 
series will all have their fall in the same direction, and the lake- 
levels will be successively lower along the series until the main 
escape is reached. When considering the evidence of any direct 
overflow, I have only in rare instances admitted the validity of the 
proof where no related severed spurs could be found to corroborate 
the available data. 
In studying the severed spurs, with their variants, in an aligned 
sequence, no surprise need be felt if two spurs are severed, while 
an intervening one of equal prominence remains intact. The ice- 
margin has commonly been very sinuous, and even lobate, and a 
lakelet might stretch across two valleys ormore. Very often it has 
happened that an aligned sequence has been modified by a differential 
retreat of the ice-margin, causing one lake to persist longer than 
another. Thus a retreat of the ice-front near the lower end of a series 
may throw some intermediate channel out of operation, while its 
neighbours, above and below, continue to fulfil their functions and 
to cut lower. 
Upon a comparison of levels, we find that the intermediate 
channel is at a higher altitude than its neighbour up stream. The 
same statement may apply to a marginal overflow that has an ice- 
wall in the middle part of its course. A retreat of the ice up to 
this point may allow of a‘ lateral escape’ being formed, and the 
upper segment of the overflow may cut deeper than the lower. 
An erople of this is afforded by the Murk-Mire-Moor channel. 
