always its shape and general character, the general bank line in the same time caving back about 150 feet. Photograph No. 1 gives a view of this eddy pocket taken looking upstream in April, 1899, at a fairly high stage. The width of this pocket measured perperdicular to the general bank line is about 150 feet. Other and smaller but similar pockets can be seen in the bank above.

Caving of the eddy kind is specially important when we come to consider bank revetment. The lower end of each piece of contin uous revetment may, after some ordinary caving has taken place below it, act as a false point. Below such a false point an eddy is soon formed, and as a result a pocket is cut into the bank below the work. This pocket tends to enlarge on its upstream side and thus to eat into the bank behind the revetment and thus destroy it slowly from below.

Later in this paper, in the history of the revetment work that has been carried on, numerous examples will be mentioned of eddy pockets formed below the ends of revetments. In two special cases the pockets thus formed were so large and dangerous that dikes had to be built in them to break up the eddies. These cases are the Fletchers Bend revetment (160 R), where a dike was built in 1895 below the end of the 1888 work (Plate XLVII), and the Hopefield revetment (229 R), where a similar dike had to be built in 1900 at the end of the existing revetment.

If a dike projects as a prominent salient from a generally straight bank, an eddy will probably be found below it. Such dikes have in a few cases been built for the purpose of protecting the bank below them, and have done more harm than good, for the eddies below have been so large as to cut the bank away rapidly, and the dikes, instead of protecting the banks, have accelerated their destruction. Examples of this will be found in the Caruthersville dike (110 R), built in 1898, and the dike at Fletchers Bend (159 R), also built in 1898. The eddy below this latter dike was at high water nearly 1,000 feet long, and the vortices and whirls in it were so large and powerful as to be dangerous to small boats. After the decline of the flood, it was found that an eddy pocket had been cut in the bank, 800 feet long by 250 feet wide, with depths of over 40 feet where the shore line had been the year before.

There is another cause of bank destruction that has not been touched upon, namely, the action of the waves caused by winds or passing vessels. Erosion from this cause is restricted to a belt

No. 1. Shows caving bank near Caruthersville, April, 1899, looking upstream. The large eddy pocket is 150 feet wide. Smaller eddy pockets seen in background.

within a few feet of the surface of the water. During low water. from the shape of the river and the protection of its banks and th timber on them, the wind can never get a very long, continuous sweep, and the height of waves is limited, and the damage done h them being local and occasional, is relatively small. However during overflow the water surface extends to the levees. The sweep of the wind on the water toward an exposed levee is fre quently several miles long, and the season of the year when over flows usually occur is that of the strongest winds, and the damage done by them is relatively greater. Levees are at best no larger than required and damage to them from wave wash becomes of considerable importance. This matter, therefore, relates rather to levees, and under that head the subject will be discussed at greater length.

EXTENT OF CAVING

Numerous attempts have been made to determine the average amount of bank caving going on, and, as might be expected, the different estimates vary much, depending on the way the estimates are made. Some places do not cave at all. Others cave slightly at one stage and fill again at another. Other points cave at all stages. Indeed, an average of such a matter is of but little value, the aggre gate and maximum amounts being of much more importance.

It has been estimated that the total volume of bank destroyed annually by caving, below Cairo, exceeds a prism a mile square and 1,000 feet high, or several times as much as the sediment carried annually into the Gulf of Mexico.

RAPIDITY OF CAVING

As examples of the extent and rapidity of caving a few cases will be given, though there are many such on the river. A large cave at Hopefield Bend (228 R) occurred December 12, 1883, and took only a few hours to develop its maximum. As measured next day, it was 440 feet long by 175 feet wide in the middle. On December 17 it again caved at the end, extending its length to 700 feet. The bank along the locality of this cave had been revetted the previous year with woven river mats 140 feet wide. When sunk, the depth along the outer edges of these mats was about 44 feet. Immediately after the caving the depths at the same places were from 75 to 80 feet. In a single season the bank at Fletchers Bend (160 R) for a distance of over a mile receded 700 feet, and at Hopefield Bend

(228 R) a half-mile length of bank in one season cut back 750 feet. As an example of continued rapid caving the following may be mentioned: Hopefield Bend near the point receded 2,800 feet in

five years.

As examples of caving continued for a series of years we have: O'Donnell's Bend (151 R) receded 4,700 feet between 1850 and 1897, 100 feet per annum; Little Prairie Bend (107 R) receded 7,200 feet between 1821 and 1897, or 95 feet per annum; Little Cypress Bend (92 R) receded 8,100 feet between 1821 and 1891, or 115 feet per annum; Barfield Bend (142 R) receded 9,500 feet between 1837 and 1897, or 160 feet per annum.

TIME OF CAVING

It has been stated that the greater the velocity the greater the erosive power of the current, and hence, as the velocity is usually greater during high stages than during low, it would be expected that bank caving would be more rapid in high water than in low, and such is the fact. But the analogy does not go further, for though current velocities are usually, for the same gage reading, greater during rising than during falling stages, bank caving is usually more rapid during falling than during rising stages, this being due to the greater saturation of the bank and the return of seepage water.

These facts are generally true as far as the river as a whole is concerned, but are not necessarily true for any one locality. At some places caving goes on at all stages; this is usually true in sharp bends. Other places cave at certain stages only. Cases occur where some distance from a bank there lies a sand bar of comparatively coarse material. At low water this bar deflects the current against the bank and caving takes place. At high water the bar is submerged, the general direction of the current is not against the bank, and caving not only does not take place, but there may actually be deposit made. In other cases the current may impinge at all stages against a bank composed of material that resists erosion fairly well. At low water the reduced velocity is unable to cut away the bank, while the increased erosive power at high stages will cause the bank to cave.

As an illustration of the manner in which bank caving takes place, a few photographs have been inserted.

BANK AND BAR FORMATION

The banks and bars of the river are formed of sediment dropped | by the river when from any cause its ability to transport the sedi ment is reduced, and as this ability to carry sediment is due to the motion of the water, a decrease in the velocity of motion is usually followed by a fill, and in general in those portions of the river where from any cause, natural or artificial, the current of the river be less than usual, a deposit will be made, and if the water be abso lutely brought to rest all its sediment will be dropped and it will become clear.

Like bank destruction, bank building goes on continually. The river is forever scouring in one place, and therefore necessarily fill ing in another. All of this depositing is, of course, done beneath the surface of the river, and is, therefore, at the time, invisible, but, though not so apparent as the caving, it goes on just as surely. As with bank caving, bank building is usually more rapid during higher stages than during low and reaches its maximum as the river is falling after a flood.

EDDY DEPOSITS

A few pages back there was described a formation of what we called "pressure" eddies. It was then shown that the current in such eddies usually moves slowly, and sometimes in one direction and sometimes in the other. In consequence of this slow velocity and occasional halts, much of the load of sediment carried by the current is dropped, and as the water forming the eddy is being constantly renewed, and as the fresh water brings with it its own proportion of sediment which is dropped in turn, such eddies usually cause large and rapid deposits. The best examples in these districts of such eddy filling is in the growth of the Memphis sand bar (230 L), which will later be described in full.

DEPOSITS DURING FLOODS

During flood stages the banks of the river are generally submerged and the river bottom covered with overflow water. This water, escaping over the banks, carries with it its load of sediment, but as its velocity is reduced by the many obstructions and the greater resistance met with, the water becomes less and less able to