Principles of training Rivers through Tidal Estuaries. 517 



the model was selected from the most recent design* proposed by an 

 engineer who had previously submitted schemes in 1881f and 1886. J 

 The trained channel was widened out by an alteration of the southern 

 wall from Quillebeuf, and the northern' wall from Tancarville, more 

 than treblino- the width between the training'' walls at Berville in the 



O 3 



model ; and the walls were extended in sinuous lines to Havre on the 

 northern side, and Honfleur on the southern side, as shown on 

 Plate 3, fig. 2, thus forming a winding trained channel rapidly 

 enlarging near its outlet. The model, with the.se lines o training 

 walls, was worked for about 5000 tides, with the results indicated on 

 the chart. Deep channels were scoured out close along the inner 

 concave faces of the training walls in the model ; bat shoals appeared 

 over a considerable area of the newly trained channel ; a bar stretched 

 across the deep channel where it shifted over from the south to the 

 north training wall, about half way between Berville and Honfleur ; 

 and a large sandbank, emerging above low water, occupied the centre 

 of the outlet opposite Honfleur. Deposit also occurred at the sides 

 of the estuary behind the training walls. 



As it was of importance to ascertain to what extent accidental 

 modifications in the arrangement of the sand in the preparation for 

 an experiment might affect the result, the lines of training walls 

 described above were inserted a second time in the model, after the 

 subsequent scheme E had been experimented upon, rendering it 

 necessary to replace afresh both training walls, and to remodel the 

 sand so as to represent approximately the present condition of the 

 estuary. The model was prepared for this second experiment in the 

 usual way, without any special endeavour to secure coincidence with, 

 the first experiment in the initial arrangement of sandbanks and 

 channels, The condition of the low-water channels in the model, 

 after working the model with this arrangement of training walls for 

 the second time for about 5400 tides, is shown on Plate 3, fig. 3. 

 The main features of the trained channel in the charts of the two 

 experiments exhibit a very fair resemblance, considering the modifica- 

 tions which any alterations in the initial condition might produce, 

 and .the naturally variable state of the channels in a wide outlet. 

 The deep channels reappear in the second chart at the inner concave 

 faces of the training walls, with intervening shoals ; a large sandbank 

 is again visible at low water along the north training wall, opposite 

 La Roque and Berville in the model ; and the sandbank in the centre 



* 'Deposition de M. Vauthier devant la Commission des Ports et Yoies Navi- 

 gables de la Chambre des Deputes,' Paris, 1888, p. 17, and Plate 4. 



t ' Rapport sur les Ameliorations dont sont encore susceptibles la Seine Maritime 

 et son Estuaire,' L. L. Vauthier, Rouen, 1881, p. 46, and Annex 29. 



J ' Dire a 1'Enquete ouverte sur 1'Avant-projet des Travaux d' Amelioration de 

 la Basse-Seine, 1886,' L. L. Vauthier, Paris, Plate 1. 



