DAVIS : RIVER TERRACES IN NEW ENGLAND. 299 



the general processes of weathering may lower the flood plain about as 

 fast as the river sinks to a fainter and fainter slope. The rate of degra- 

 dation of a terracing river in a drift-filled valley, resulting from a 

 climatic change or from a land movement, may therefore be allowed a 

 decidedly higher value than that of an ageing river in a normal, undis- 

 turbed cycle. 



(2) It is well understood that a graded stream may work actively in 

 wearing or building its banks laterally, however slowly it aggrades or 

 degrades its valley floor. 



(3) There are abundant examples of rivers whose lateral oscillation 

 or swinging carries them from side to side of a flood plain that is 

 much wider than their meander belt. The Mississippi is a noted case 

 of this kind. Its meander belt is six or eight miles wide in a flood 

 plain whose enclosing bluffs are from twenty to sixty miles apart. A 

 similar relation may be seen in many meadow flood plains, drained by 

 small brooks. 



(4) The fourth postulate involves a principle of river action which 

 may be familiar to hydrographers, but which is nevertheless not com- 

 monly stated. Imagine a stream in a broad flood plain passing from a 

 straight stretch or tangent to a well-defined curve. On the tangent A B, 

 Figure 7, the thread of fastest current might, as far as local control is con- 

 cerned, lie along the middle of the channel, or indifferently on one side 

 or other of the middle line. On entering the curve the fastest current 

 is gradually shifted, B C, towards the outer bank of the channel, and 

 there flowing steadily all around the curve C D it determines the line of 

 greatest depth. On passing from the curve the thread of fastest cur- 

 rent is necessarily delivered to the next down-stream tangent, D E, on 

 the down-valley side of the channel, and only after flowing for a signifi- 

 cant distance will the fastest current gain a path near mid-channel. If 

 the curves or meanders are close set, so that one curve passes directly 

 into the next one with no intervening tangent, then the thread of 

 fastest current must, on passing the point of inflexion, enter the up- 

 stream end of the next curve near its inner or convex bank, and only 

 gradually be displaced towards the outer bank as inertia has time to 

 bring about its usual effect. 



As a result of this systematic displacement of the fastest current 

 from the mid-channel line the bank that it approaches will be worn 

 away, while deposition will take place along the bank from which the 

 fast current is withdrawn. The stream will therefore tend to wear 

 away the bank on the outer side of its curves (but perhaps failing to 



