514 D. MILNE HOME ON HIGH-WATER MARKS ON THE 
the River Somme in France, a river apparently about the same size as the Earn 
and Teith. In the Somme valley, there are beds of fluviatile gravel, at the 
height of about 80 feet above the present channel of the river ; and these beds, 
Mr TyLor maintains, were deposited by the river in its present channel rising 
to that height when in flood. 
It occurred to me, that some test of the correctness of these views might be 
obtained by an examination of similar high terraces visible in the valley of the 
Tweed. I had a good opportunity of making this survey, having long had my 
residence on the north bank of the river, and being therefore well acquainted 
with the locality. 
I shall also in this memoir describe some other things in districts adjoining 
the Tweed, which seem to be more or less connected with the formation of the 
terraces. 
I divide my paper as follows :— 
1st, Notice of the River Tweed, between Melrose and Berwick. 
2d, Notice of districts adjoining the River Tweed. 
3d, Theoretical explanations. 
4th, Views held by other persons. 
I—Riwer Tweed and its Banks. 
1. The channel of the Tweed at Melrose, is about 270 feet above the sea, 
and about 30 miles distant from the sea. 
The towns of Kelso and Coldstream divide the course of the river into 
three parts, each part being about 10 miles long. The levels of Kelso and Cold- 
stream, above the sea, show that between Melrose and Kelso, the gradient or fall 
of the river is at the rate on an average of 17 feet per mile ; between Kelso and 
Coldstream, of 7 feet per mile; between Coldstream and Berwick, of 3 feet 
per mile. The diagram fig. 1 represents these several gradients. 
M 

Fig. 1. 
B, Mouth of Tweed at Berwick; C, Tweed at Coldstream; K, Tweed at Kelso ; M, Tweed at Melrose. 
That rate, however, is on the assumption that the river runs from one point 
to another in a straight line. If in consequence of its windings, 60 miles in- 
stead of 30 be assumed as the length of its actual channel, the rate of fall 
per mile would be very much less. 

