T. S. Ellis— Windings of Rivers. 109 



which, still exist but no longer serve for streams, and others altogether 

 effaced, have, in former times, been included in the system. I hold 

 to my belief, recorded nearly twenty-six years ago, that only by taking 

 into account the influence of tributary streams can river-windings be 

 explained, and that the subject is important as " bearing on the 

 formation of warths and the maintenance of navigation channels." ' 



The system of drainage as we see it, a principal channel with 

 tributaries flowing in on either side, is the result of a long process 

 of evolution. When the rainfall began its work of denudation, no 

 turf, nor trees, nor verdure of any kind existed : the rain fell directly 

 on to the surface. JN'ow a very large proportion is either conducted 

 gently down to and into the earth, or fails to reacli it at all, passing- 

 off by evaporation. According to E-eclus ("The Earth ") "Becquerel's 

 experiments prove that during heavy rain only -i%- of that which 

 falls reaches the ground." At any rate, a much larger quantity 

 of water would, with the same rainfall, have flowed off the surface 

 than now does ; and this, I think, fully explains dry valleys and 

 combes, even those of the Chalk hills, such as are now seen. 



Whatever the size of the early channels they would certainly be 

 numerous, and, having regard to the uneven surface of an uplifted 

 area, be diverted in different directions. Thus they would meet in 

 loops and form a network which might continue even after a well- 

 defined valley has been formed. Such loops are often seen in the 

 present day. Out of the network the principal line of stream (the 

 river) is selected. Which one this will be must depend on a number 

 of circumstances. A tributary stream coming from a lateral valley 

 would require a channel on that side. Into this, necessarily kept open, 

 streams near the middle line of the valley may flow, and these may 

 attract others, so that, finally, one might be continued in a channel 

 adapted to the needs of the principal stream and of the tributary. 

 Certainly, either the principal stream must be inclined towards the 

 tributary, or the tributary must be extended to the principal stream. 

 In fact, the two often meet in the form of a cajDital letter Y, as 

 I illustrated in the paper mentioned by the case of the Severn and 

 Avon at Tewkesbury. A diverted route of the principal stream, more 

 or less circuitous, would render those in a more direct line unnecessary-, 

 except only for the area close to them. Denudation, therefore, would 

 not go on in the same degree, and some of the channels would be 

 effaced, wholly or in part. 



The process of evolution is, in principle, the same, whether it be in ' 

 a denudation area or in alluvial soil where the streams not only 

 arrange their own channels but also build up their own banks. 

 The process is well illustrated in the sketch-map taken from an old 

 Baedeker's Guide to the Ehine, Fig. 2, p. 112. The river 111, which 

 occupies the same valley, is not shown ; it is now at the margin on 

 the left side, by the Yosges Mountains. Probably it was once part of 

 the Rhine system which is seen inclining to the right side, that of the 



1 " On some Features in the Formation of the Severn Valley," a jDaper read 

 before the School of Science Philosophical Society, at Gloucester, on February 7th, 

 1883. Printed for the Society. 



