Principles of training Rivers through Tidal Estuaries. 523 



near the outlet ; and a bar appeared in front of the outlet outside. 

 The breakwater also, extending across part of the outlet, favoured 

 deposits both inside and outside the estuary, by producing slack water 

 in the sheltered recesses. 



The second class of trained channel was designed to profit by the 

 scour at the concave face of bends, so clearly exhibited at the first 

 bend of all the charts, and to continue the depth thus obtained by 

 restricting the width between the bends, on the principle adopted for 

 winding non- tidal rivers. Experiment, however, did not bear out the 

 advantages anticipated from this system, probably owing to the variable 

 drection of the flood tide at different heights of tide, its being checked 

 in its progress by the winding course, and not acting in unison with 

 the ebb from the difference in its direction and the width of the trained 

 channel near the outlet. The main stream in a non-tidal winding 

 river always follows a tolerably definite course; whereas the flood 

 tide tends gradually, as it rises, to assume as direct a course as 

 possible. The difference, therefore, in the conditions of a non-tidal 

 and tidal river,. in this respect, is considerable. 



The third class of trained channel afforded a wide, tolerably uniform 

 channel in the experiments ; the flood tide was less impeded in its 

 progress than with the other forms of training walls, and appeared to 

 act more in concert with the ebb. 



The experiments, accordingly, indicate that the only satisfactory 

 principle for training rivers, through wide estuaries with silt-bearing 

 currents, is to give the trained channel a gradually expanding form, 

 with as direct a course as possible to the outlet. The rate of increase 

 of width between the training walls must be determined by the special 

 conditions of the estuary. If the outlet is very wide, and the gradual 

 expansion in width cannot be commenced a considerable distance up 

 an estuary, some restriction in width at the outlet may be expedient 

 to avoid a too rapid expansion. It is evident that the widening out 

 adopted in the last experiment (Plate 4, fig. 3) was carried to its 

 utmost limits, from the continuance of sandbanks inside the trained 

 channel, and that, regarding merely the improvement of the 

 channel, it might have been preferable to restrict its width at the 

 outlet as effected in Scheme C (Plate 3, fig. 1). At the same time, it 

 must not be inferred, from the existence of these sandbanks, that the 

 distance apart of the training walls was much too great in the last 

 experiment; for the width apart of the training walls necessitated 

 the inclusion of a greater extent of sandbanks within the trained 

 channel at the outset, and also rendered the rate of improvement in 

 the channel more gradual, so that the improvement in the channel 

 both in direction and depth was still progressing at the close of the 

 experiment, and the sandbanks in the channel were in process of 

 removal, and not being formed. The choice in such cases, where the 



