bend, it seldom does so with anything like uniformity. The cor position of the bank, and consequently its resistance to erosion differs at different places, as does also the intensity of the current attack. Owing to these causes, caving is usually much more rapi in some parts of a bend than in others. At the locality of this rapid caving the curvature may increase and the shape of the bend change so that the current is deflected away from the concave bank below, so that for some distance it follows the middle of the river or even attacks the convex bank opposite. As it must, from the shape of the bend, return to the concave bank not far below, there are thus formed two crossings with the attendant bad navigation. Noticeable examples of this are found in the bend of Island 35 (190) (Plate LXXI) and in Cow Island Bend (250) (Plate LXXII), in both of which the channel leaves the concave bank, crosses to the convex and then returns, while between these crossings there is a bar against the concave bank.

The effect of this irregularity in the rate of caving is also sp. cially noticeable near the lower ends of bends, for on these irregu larities depend the shape of the concave bank at these places, and on these the direction and locality of the crossing and the point of maximum impingement of the current in the bend below. Slight changes in the shape of the bank at the foot of a bend may cause large changes in the other functions just mentioned, and it is largely due to this and to the consequent shifting of the crossings that the river at the crossings is usually so wide. Generally, as the bank caves away the crossing and the point of attack on the opposite bank move downstream, but if the bank at the foot of the bend resists erosion much better than the bank a little farther up, just the opposite is the result.

It has been mentioned above that straight reaches of considerable length are sometimes found on the river. In such reaches the banks do not serve as fixed guides to the channel, and it is apt to wander here and there, sometimes cutting away one bank, sometimes the other, with a resultant widening of the river. At high water such a wide place causes a reduction in velocity and a deposit of sediment, usually in the form of one or more middle bars. Among these bars, composed of fairly coarse material, the low water has to make its channel, and it is frequently deflected by these bars against and cuts away the main bank, sometimes one and sometimes the other, sometimes both. This but intensifies the trouble, making the river

still wider at high water, causing the bars to increase in size height, and so on.

Such straight reaches are apt from these causes to be the wi portions and the worst points on the river, as far as navigation. concerned. As examples of this, we have Plum Point Reach, ab which much is to be related hereafter, and the reach from Ni Madrid (70 R) to Point Pleasant (80 R). This reach is al straight for 10 miles, and in it the action above described went until, in 1898, the river was over 2 miles wide, with three large a high bars in it, dividing up the channel into four bad chutes.

In general, when we have from any cause a widening of the river there will be deposits made, and these usually take the form of mid dle bars. These middle bars are frequent in straight reaches, and at times in bends, where the caving takes place too rapidly for the making banks to keep up with it. These bars thus made may, by. change of conditions, be soon cut away again, but they may, as has been described, act so as to intensify the widening and therefor their own growth. In such cases they grow higher and higher, b come covered with willows, and eventually with timber, and becor. islands. Many of the islands of the river were probably made i this way, and others are growing up now. One of these, Elmot Ba or Island (160), has grown up since about 1870, and its growt from time to time may be traced on the maps of Plum Point React. (Plate IV and Plates LVII to LXIX.) One thing specially noties able about this island is that instead of being cut away slowly at its head and growing at its foot, as is the general rule, it has grown largely at its head, which is now much farther upstream than in the early eighties.

Islands in the river are very numerous and exist of all sizes. Most of them were probably made gradually by growth from a sand bar, somewhat as above described, and some possibly by cut-offs. Nearly all of the islands are so situated that the main current of the river is thrown rather to one side of them, and on this side is found the main channel of the river, the passage on the other side being usually only a high-water chute, dry or not navigable at low water. Into these chutes there is carried during the higher stages large quantities of sand or gravel, and usually considerable drift, and in this way many such chutes have in the past been completely filled and the islands thus joined on to the main bank of the river. Some such chutes are so situated that the current through them at very high stages is sufficient to keep them open or even scour them out

slightly, while at slightly lower stages deposits are made in th and the chute, therefore, remains for years in about the same eral condition; other chutes, after having been almost closed, been by a change of conditions scoured out and have become time the main channel of the river.

When these chutes behind islands are open below mid-stages discharge passing through them as the main river falls after at. reduces the amount passing through the main river, and, theref reduces the power of the river to cut a proper low-water char through the sand bars made during the flood. In such cases i sometimes desirable that the chutes be closed by artificial means and the problem of how to do this is, in the case of large and l chutes, one of the most difficult connected with the regulation the river.

One noticeable feature of the river bottom as shown on the mayis the large number of horseshoe-shaped lakes in the botten g These were formed by cut-offs.

As the banks in the bend of the river cave away and the point opposite grow up, it at times happens that these points will becon long and in places narrow. If the conditions are such that cavit takes place on both sides of such a neck, it will get narrower ar narrower, until in some high water it will be cut in two by surfac scour or by the washing out by hydraulic pressure of a pervio sub-stratum. The first result of a cut-off is to shorten the rive and thus increase its slope both in the immediate vicinity and for miles above and below. This increased slope means increased velo ity and increased scour and bank destruction and rapid changes of regimen. This bank destruction will usually be along the concave banks of bends and tends, therefore, to the gradual lengthening of the river and thus in time recovering the length lost by the cut-off. After a cut-off has taken place, the old bed of the river begins to silt up. This process goes on most rapidly next to the new channel, the two ends of the old river fill up first, and there is thus formed a horseshoe-shaped lake.

On account of the rapid changes in regimen due to cut-offs, they are considered objectionable, and at some places considerable work has been done by the Commission to prevent them from happening. However, as all such cases have happened outside the limits of these districts, no further reference to them will be made.

In 1876 there occurred about 75 miles above Memphis (205 miles below Cairo) a cut-off that shortened the river by about 15 miles.