VOL. XVI. (3) THE LOWER SEVERN 253 
suggesting former lines of streams following completely 
across the area. I do not, however, feel that the main- 
tenance of the direction of these lines, all through the 
period of denudation down to the present level, is fully 
explained. The position of the outliers at Churchdown 
and Robinswood-hill also affords matter for speculation 
which cannot be discussed here. 
In the western division, a surface-line from a point 
above Upton-on-Severn down to Westbury-on-Severn 
hardly rises anywhere, so that a continuous stream along 
it is not difficult to imagine. There is, however, a divide 
at Dobshill, from which the present drainage is carried “p 
the line of the valley to the Severn at Queenhill, fig. 7B. 
Below Dobshill the Glynch Brook flows down the line of 
the valley to join the Leadon, which in turn joins the 
Severn at Gloucester. Farther down the valley, beyond 
the water-shed of the Leadon, there is, at Tibberton, 
another divide, below which the rest of this division of 
the valley is drained into the Severn at Minsterworth and 
at Westbury by Newnham. 
To me, a continuous line of valley, even if divided by 
a pass in a range of hills, is always an evidence of denud- 
ation by a stream flowing along that line; but when, as in 
this case, streams are found flowing in opposite directions, 
authority explains the continuous valley as formed by two 
streams “cutting backwards” till they nearly meet. The 
subject has been much discussed in relation to the glens 
of Scotland (“ Scenery of Scotland,” by Sir A. Geikie), 
with the “ballocks,” as dividing-ridges between the streams 
are called. My conception of the order of events in the 
case just mentioned is as follows :—When the Severn 
and Avon united their streams and elected the middle 
channel as the permanent one, all passing down the 
eastern and western division of the valley ceased, but 
provision had to be made for the local drainage, and, in 
