The bars at the mouths of deltaic rivers cannot usually be kept down by dredging except at great expense. The usual method of dealing with them is to run out two parallel jetties, in continuation of the river banks, so as to bring the mouth of the river out to the bar. The river then scours a channel through the bar and, if the walls are not too far apart, the depth will probably become as great as in the river and sufficient for navigation. The river, however, tends to at once form a new bar further out. The rapidity with which the new bar forms will be greater or less as the specific gravity of the materials carried by the river is greater or less, and as the strength of any littoral current is less or greater. Clay is spread far out while sand quickly sinks. All deposits are, however, swept away if there is a strong littoral current. The steeper the slope of the bed of the sea away from the bar the longer the new deposit will take in forming a fresh bar. Also the less the discharge of the river the less the deposit will be. The branch of a deltaic river selected for improvement by having the bar at its mouth removed, should be one which has a small discharge and whose mouth is in a position where there is a strong littoral current. In the case of the Rhone, the branch selected was the eastern one, whose mouth was not exposed to any littoral current. Moreover, the other branches of the river were closed, and this increased the discharge of the branch which was left open. The work did not succeed. In other cases, the parallel jetty method has succeeded, and notably in the case of the Mississippi. In this case willow mattresses weighted with stones were used. The question of keeping down the discharge does not, however, appear to have always received sufficient attention. In the case of the Mississippi the “South pass” was selected for improvement. In order to remove a shoal its upper end was narrowed and its discharge reduced. The upper ends of the other “passes” were then obstructed so as to restore the discharges of all the passes to their former amounts. The wisdom of this step is questionable. It is desirable to keep down the discharge of the branch which is to be improved to the lowest limit consistent with free navigation. If the width of the river near its mouth is greater than is desirable for the width between the jetties, the latter are sometimes made to converge though their outer ends are made parallel. In the case of the Mississippi the jetties were made with a slight curve to the right. It would seem desirable always to make the jetties with quite a considerable curve. The jetty which was convex to the channel could then probably be shortened. In a case where there is a littoral current, say to the right, the curve of the jetties could be to the right, so that the stream on issuing would tend to merge into the current and assist it.