516 Mr. L. F. Vernon-Harconrt. 



between Honfleur and Amfard was tortuous as before, but its direc- 

 tion was different. The deep hole at the outlet, the bar beyond, and 

 the advance of the southern foreshore beyond the breakwater, 

 reappeared again with very similar features to those in the first 

 scheme, except that the sandbank did not quite reach the outside 

 face of the breakwater at low water. (Compare fig. 2 with fig. 1 in 

 Plate 2.) 



The results which followed from working the model with the 

 southern training wall prolonged to Amfard, are shown in Plate 2, 

 fig. 3. The main alteration from the former experiment naturally 

 occurred between Honfleur and Amfard in the model, a continuous 

 channel being formed along the new piece of concave training wall > 

 whilst the general depth inside the estuary was improved as far as 

 the meridian of Hoc Point. The channel, however, above Honfleur 

 was not improved, owing apparently to the want of uniformity 

 between the directions of the flood and ebb currents in the model. 

 The other features remained very similar to the former case, except 

 that the end of the sandbank beyond the breakwater was slightly 

 eroded, whilst deposit took place between the extended training wall 

 and the breakwater. (Compare fig. 3 with fig. 2 in Plate 2.) 



Scheme C. The third arrangement of training walls experimented 

 upon in the model was chosen from a design published in 1885.* It 

 consisted of an enlargement of the original trained channel below 

 Quillebeuf, by a modification of the southern training wall from 

 Quillebeuf, and of the northern training wall from Tancarville, and 

 the extension of the northern wall to Amfard and Havre, and the 

 southern training wall to Ratier, as shown on Plate 3, fig. 1. The 

 trained channel was thus given a curved, gradually enlarging form, 

 and was directed into the central channel of the model, between 

 Ratier and Amfard, the Villerville and Havre channels being 

 practically closed near low water. The effects of working the model 

 for about 6500 tides with this arrangement of training walls are 

 indicated on the chart (Plate 3, fig. 1). The main channel kept near 

 the concave southern training wall for some distance below Berville, 

 and then gradually assumed a more central course between the train- 

 ing walls towards the outlet, passing out just to the south of the, 

 Amfard bank. The channel thus formed had a good, tolerably uniform 

 depth, together with a fair width, owing apparently to the flood and 

 ebb tides produced in the model following an unimpeded and fairly 

 similar course. Deposit occurred behind the training walls on each 

 side; and the foreshore advanced in front of Trouville in the model, 

 in consequence of the shutting up of the Villerville Channel. 



Scheme D. The fourth arrangement of. training walls adopted in 



* 'La Seine Maritime et son Estuaire,' E. Lavoinne, Paris, 1885, p. 140, and 

 ' Instit. Civ. Engin. Proc.,' vol. 84 ; p. 248, and Plate 4, fig. 9, 



