514 Mr. L. F. Vernon-Harcourt. 



work. Moreover, the schemes exhibit sufficient variety to admit of 

 their being taken as types of schemes for throwing light upon the 

 principles on which training works should be designed in estuaries. 

 Accordingly, the third stage in the investigation consisted in extend- 

 ing the training walls in the model, in accordance with the lines of 

 some of the schemes proposed ; and, after working the model for some 

 time with each of the extensions successively, the several results were 

 recorded, as shown in Plates 2 and 3, and Plate 4, figs. 1 and 2. The 

 lines of training walls experimented on in the model were taken, with 

 one exception, from five out of the seven most recent schemes pro- 

 posed, as these five schemes are, I believe, the only ones which are 

 still put forward for adoption. The lines shown on Plate 4, fig. 3, 

 represent merely a theoretical arrangement of training walls, inserted 

 for a final experiment in the model, to ascertain the effect of the most 

 gradual enlargement of the trained channel which the physical con- 

 ditions of the estuary would have admitted of at the outset, whilst 

 maintaining the full width at the mouth. 



Scheme A. The first arrangement of extended training walls intro- 

 duced into the model was taken from a scheme, some of the main 

 features of which were proposed in an earlier scheme in 1859,* and 

 which was put forward in an amended form in 1886. f The design, 

 as inserted in the model, consisted of an extension of the parallel 

 training walls from Berville down to Honfleur, and the formation of 

 a breakwater across the outlet, from Villerville Point on the southern 

 shore of the estuary, out to the Amfard bank, thus restricting the 

 inouth to the channel between Amfard bank and Havre. The lines of 

 these works were formed in the model with strips of tin, as shown 

 on Plate 2, fig. 1 ; the northern training wal was kept low, and the 

 southern wall was raised to the level representing high water of neap 

 tides ; whilst the strip representing the breakwater was raised above 

 the highest tide level, thus forcing all the flood and ebb. water to pass 

 through the Havre Channel. The results obtained in the model with 

 these arrangements, after working it for about 6000 tides, are indi- 

 cated on the first chart (Plate 2, fig. 1). The channel between the 

 prolonged training walls had a fair depth throughout, partly owing 

 to the concentration of the fresh-water discharge between the walls, 

 and partly from the retention of some additional water in the channel 

 at low water, by the hindrance to its outflow offered by a sandbank 

 which formed in front of the ends of the training walls. A deep hole 

 was soon scoured out in the narrowed outlet by the rapid flow of the 

 water filling and emptying the estuary at every tide. The absence, 

 however, of connexion between the direction of the flood tide current 



* ' La Seine comme Yoie de Communication Maritime et Fluviale,' J. de Coene, 

 1883, p. 11, and Plate 7. 

 f ' Projet des Travaux a faire a 1'Embouchure de la Seine,' L. Partiot, Paris, 1886. 



