250 PROCEEDINGS COTTESWOLD CLUB 1909 
and rest upon strata of hard rock dipping eastward, as 
shown, the whole of the former may be removed before 
any of the latter. Then the divide will shift westward to 
F. As the interval between the divide, the zero of the 
river-system, and the headwater of the stream C is now 
too great for disposal of the rainfall without a water 
channel, one will form and join that already formed, in 
the stream C. This is a result of differential denudation ; 
but we are told that the process of working or cutting 
backwards goes on when a stream can give a quicker fall 
to a lower level than the one to which it is opposed. 
This is equivalent to saying that a steeper incline facili- 
tates the process. A case in point is given at Charlton 
Abbots, near Cheltenham. Here a road divides the head 
waters of the Isborne and the Coln, the former flowing 
into the Avon near Evesham, and the latter into the 
Thames at Lechlade. On the spot I have examined the 
streams, but I cannot find the slightest evidence or in- 
dication that the Isborne “is gradually working its way 
backwards and causing a retrogression of the Coln.” 
Neither Prof. Davis nor any of those who accept his 
teaching explain, so far as I know, how the areas towards 
which streams are supposed to be working backward 
are drained, before those subsequent streams reach them. 
To me it seems that if these areas were large enough 
to require channels to carry off the rainfall, channels 
would have formed without waiting for the development 
of subsequent streams “sent out” from the consequent. 
If the areas were not large enough to require channels to 
carry off the rainfall, subsequents would not be sent out 
to meet a need which did not exist. The cutting power 
of water acts downwards, towards the centre of the earth, 
and forwards, along the line of the stream; not back- 
wards, where the water diminishes until a channel is no 
longer necessary, and, at the dividing line which separates 
it from another watershed, is reduced to zero. The 
