year in some of the tributaries, simply because the levee method can not be applied there. It must be admitted that the levee system affords fairly efficient protection from ordinary floods, in so far as damage from overflow in the lower valley is concerned, but its desirability is seriously impaired by the fact that the levees must be constantly replaced. Just as long as the river is allowed to swing against its banks the soft alluvial soil will continue to cave in; levees originally built back from the channel are eventually undermined, rendered useless, and then in cases of high water stages, unless a second line of levees exists, the entire region is open to devastation. Until the sapping action of the river is under control there can be no such thing as a system of levees built once for all. Even the most optimistic advocates of this plan do not claim more than twenty to thirty years for the life of any levee.
It is undeniable that the army engineers in charge of the work have accomplished much in saving large areas from annual inundation, but they have not to any extent permanently improved the river as a highway of commerce. Furthermore, they have signally failed in the attempts to stop erosion of the banks, for past experience shows that no style of revetment yet devised will offer more than partial or brief protection from that action. In fifteen years some landings have been forced back more than a mile, and at important points where careful revetting has been done the retreat is said to have exceeded 1,000 feet. The levee-revetment system as now practised may be the simplest way of protecting agricultural interests on the river flats, but it is certainly not the most economical in the end, nor does it in any way afford even a temporary solution of the great problems confronting navigation and river commerce. The heavy expense of maintenance must go on without end until the sum total of expenditure will aggregate vastly more than the cost of the proper, permanent remedy. The most serious of all the shortcomings charged against present methods, however, is that they do not strike at the root of the evils. The place to control floods is where they originate, in the tributaries, and thus protect both tributary and main valleys; as well try to fight fires by blowing away the smoke as to control the floods by levees along the lower course. To improve the river to the extent of a fourteen-foot channel to St. Louis would be a foolish waste of money as long as the levee-revetment system is the chief method of control.
Various other methods of control have been suggested from time to time, but most of them do not appear feasible or to offer the desired results. The rearrangement of tributaries, either by diversion or by addition, has been advocated, since the addition of tributaries to the Po very materially lessened the flood evil, the increased volume and velocity having caused a marked deepening and widening of the channel. In the Mississippi case, however, there are no important streams which