1899] THE DEVELOPMENT OF RIVERS 283 



Ken net. And it is the bigger now ; for the Kennet system was the 

 first to be attacked from outside. 



But before detailing the incidents of that attack it may be 

 remarked that, on the evidence of the Cotteswold breached escarp- 

 ment, the Mid Wales or ultra-Severn drainage had been collected into 

 two main channels — the rivers which cut the upper valley at Stroud 

 and at Cheltenham. The latter, the original Coin, seems to have been 

 the most important; by developing subsequent streams it beheaded 

 the Wiudrush on the north at Winchcomb, 1 and the Churn on the 

 south at Gloucester, thus gaining parts of the Teme, Lug, and Leadon. 



The Middle Wye-[Frome] 2 does not seem to have made any im- 

 portant captures ; but it is surmisable that it gained the Upper Wye, 

 while the Middle Wye and the original Coin lost their head-waters to the 

 Upper Severn extension of the Evenlode. Thus, in what may be called 

 the second phase of the Thames system, the west and north-west 

 drainage had been collected into five principal channels just east of the 

 present Severn — (1) The valley between Dundry and the Mendips ; (2) 

 the valley north of Dundry; (3) the Stroud valley; (4) the Chelten- 

 ham valley; and (5) the Moreton valley. Of the streams in these 

 valleys the first two joined to make the Kennet, and the other three 

 joined, two by capture, to make the Evenlode - Upper Thames. 

 Then came the attack from outside — the headward growth of a Bristol 

 Channel stream draining to the south-west. 



IY. The Growth of the Severn. — Not to go too far back, it is prob- 

 able that the Jurassic rocks which stretched from north-east Somerset 

 to Glamorganshire had much to do in determining the position of the 

 Severn. Possibly a western extension of the Carboniferous Limestone 

 of the Mendip axis — represented by the Holmes — strengthening the 

 Jurassic rocks, prevented or rather delayed headward growth of what 

 may have been a small river with a strongly tidal estuary ; and this 

 might have caused the water to be turned south-eastwards to destroy 

 the Jurassic rocks of Sedgemoor and Bridgewater. 3 



But the barrier was broken down, and the Taff and Bhymney were 

 captured. Next the Usk system was obtained. 



The largest and most important obsequent stream ought to be found 

 on the south side of Dundry ; but it is not — the Avon on the north 

 side is so. That is to say, the longest obsequent has been developed in 

 the valley of the later beheaded consequent, where there has been the 



1 This subsequent stream was developed along an anticline. 



- The river-name in brackets indicates the predecessor which occupied the valley ami 

 flowed in the other direction. 



3 Probably the Dorset Stour represents the direction of the drainage just south of the 

 Mendips, and it originally headed in Glamorganshire, crossing the non-existent Bristol 

 Channel. Then the Parret, the Yeo, etc., are the obsequent streams developed when the 

 Channel, or perhaps it was then only a south-west flowing stream, headed back towards 

 Burnham. It will be seen that the obsequents have had time to become as long as the 

 consequent Stour. 



