extended to the deepest water, for out to that line erosion has recently been active, even if it be not now actually going on, but if the subaqueous slope is gentle it may not be practicable to extend the revetment to the greatest depth. Generally, this point can be reached with mats extending 300 feet from the low-water line, and in cases where the deepest water lies farther out the bottom is at that distance usually quite flat, and such cases are not found along the most rapidly caving banks. But even if the mat be carried to or beyond the greatest depths some scouring along the outer edge may be expected. If the bank be a rapidly caving one, and such banks are most in need of revetment, the growth of the bar opposite will probably be somewhat behind the caving, and such a bar will continue to grow at least for some years after the construction of the revetment. The increase of cross section by the caving of the bank being stopped, the decrease from bar growth will probably be felt in an increased scouring action along the bottom; that is, along the outside edge of the mattress. To prevent damage from this cause, the mattress must be so flexible that it can follow down this scour as it occurs, without being itself injured. Action of this kind can not proceed far, and if the mattress be wide enough, so that its outer edge can reach to a sufficient depth without the slope becoming too steep, all that is required has been accomplished, and it would appear that for this purpose mats reaching 300 feet or so from the low-water line are of ample width, and in many localities narrower mats will suffice. Of course, future experience may show that in some places still wider mats are needed, but there is nothing in the construction of the fascine mattress to prevent its being made of any width desired by merely using additional mooring and mat barges and more and possibly stronger mooring cables.

BANK SLOUGHING

As regards sliding or sloughing, the river bank is like any other mass of earth, there being an angle of steepness beyond which the bank is in an unstable condition and liable, upon being disturbed, to slide. This angle varies greatly with the character of the bank and its degree of saturation, and depends also upon whether or not it has recently been disturbed, and its value in any case can only be determined by experience. As the river banks are subject to overflow, and at such times become saturated, all plans for insuring their permanency may be based on the assumption of complete saturation.

Beyond the angle of repose the bank can only be permanently held by some form of retaining wall, and as a retaining wall a revetment can not ordinarily be expected to act, and hence the sloughing or sliding of the bank can only be prevented by grading the bank before revetment to a proper slope. As far as the part of the bank above low water is concerned, this grading is an easy matter, but should the bank below the low-water line be too steep no method for grading it has been as yet tried in these districts. Various methods of grading by scraping have been suggested, but so far as is known they have not been used with success.

Experience on this river, both in bank revetment and levee construction, has shown that, in general, if the upper bank be cut back to a slope of 1 vertical to about 3 horizontal it will be stable even when saturated, though in special cases the soil may be such as to require a more gentle slope. Below low water the slope may be steeper; some banks have been successfully held where for several feet the slope was as steep as 1 to 12, while the depth 100 feet from the shore was as great as 45 feet.

In revetting steep subaqueous slopes the best method would seem to be to grade the upper bank well back, and thus remove its weight, and if the bank be not already cracked or shattered it may be revetted with probable success. If, however, the bank be already cracked and shattered, further settlement may be expected to take place, and such settlement will probably do more or less damage to the revetment, but the revetment will, where still intact, prevent erosion; the settlement must in time cease and by constant watchfulness and quickly repairing any damage done the bank can probably be held in spite of its shattered condition, unless there be complications due to seepage.

SEEPAGE

A large amount of the trouble with banks can be laid to this cause. The banks of the river are, in general, built up of layers of clay, sand, and gravel. The clays are generally impervious, while the sand and gravel strata are quite pervious, and when the river rises, as in floods, there is a heavy water pressure against the bank and the pervious strata become saturated. Then as the river falls the pressure against the bank decreases, the water in the pervious strata returns to the river, and as it does this it tends to take with it the material of the pervious stratum, and if this material be fine sand, or what is known as quicksand, some of it will flow out from

tear the bank surfaces, its outcrop will assume a much flatter slope, thus partially undermining the overlying strata and allowing them to break off and fall in.

When the pervious stratum outcrops above low water the trouble is not so bad and can usually be remedied with comparative ease, but when the stratum lies below the surface and material is very ine considerable trouble may be expected. The mattress laid on the exposed face of this stratum, of course, will diminish the outflow of the material, but if the sand be fine enough the returning water will be able to carry it through even a fascine mattress, and a settlement or sloughing of the bank will eventually take place in spite of the mattress. This slide will probably do some damage to the reretment, but enough of the work will still probably be left to retard, if it does not prevent, erosion. However, the material brought down by the slide will probably cover the face of the fine stratum and stop the outflow through it, and a restoration of the revetment should then hold the bank. Further settlement may, of course, go on for some time, but it will probably be slight, and with care and quick repairs the bank can probably be held.

This trouble from seepage is, of course, intensified by the existence of lakes or streams having a connection with the fine and perrious stratum.

TYPES OF REVETMENT

As far as its continuity is concerned, three types of revetment have been tried, the continuous type, the interrupted method, and dikes, the first alone with success.

The continuous type needs little comment. If properly constructed it must prove successful. In general, the more regular the bank along which it is placed the easier and cheaper will its construction be and the less it will cost to maintain. Every projection in such bank is not only of itself subject to a more severe attack and therefore more liable to give way, but by the whirls and eddies it causes it disturbs the regularity of the river regimen and may threaten the work just below it.

Above low water such projecting points can by grading be easily removed, or at least their saliency can be reduced, but below the water surface such grading is not practical, and where such irregularities exist they must, in general, be incorporated in the revetment, special care being taken with the work in their vicinity.

In this connection one point is important; as the end of any re

vetment left exposed during high water is certain to be left by caving on a projecting point, it is desirable when a long and regular bank is to be revetted that as much as possible be revetted in one season and in one continuous piece without gaps.

Interrupted Method. This method aims to reduce the cost of the work by leaving gaps in the revetment. It was tried but once, at Fletchers Bend in 1888. Here the attack of the current was quite light for some years, so that no very great caving took place in the gaps, but when the current became stronger these pockets enlarged rapidly, destroying the exposed ends of the revetment, and large. stone groins had to be constructed to prevent further damage, and it is believed that had the current been as strong here as at other places that even these groins would have proved insufficient. Indeed, judging from the experience in other places and the rapidity with which small faults are enlarged when the current attack is very strong, no work of the interrupted type will hold unless the gaps be very small and the protecting works be very large, and in this case the advantage sought in this kind of work, cheapness, would be lost.

Dikes. This subject has been fully discussed, so it will be but briefly reviewed here.

The experience on this river with dikes used as bank revetment has shown them not to be suited to that purpose. In every case where a single dike has been built so as to make a projection from the bank it has not only failed to protect the bank below it, but has by its eddy caused or accelerated bank destruction. Where dikes have been used in sets and have been subjected to severe attack they have failed until additional work had been placed in the intervals and the set of dikes converted into practically a continuous revetment.

One of the results expected from the use of dikes was the causing of deposit between them and the building out of the bank. This they have clearly failed to do and the gaps between the dikes have not been filled up, but have been cut into the bank. Were the gaps made very small and the dikes very large, they might possibly serve, but at the expense of economy, and, in general, experience shows clearly that when the caving bank is fairly regular it is both better and cheaper to use continuous revetment.

In certain cases, however, dikes have been found very useful; namely, as eddy breakers in pockets. The formation of eddies in pockets and the caving due to them have been fully explained. By building a dike near the center of an eddy pocket the eddy is broken

Two, its action interfered with, the velocity of the eddy current is atly reduced, and, if the dike be properly planned, scouring is oped and a deposit is caused. In pockets and similar localities any dikes have been built and with about uniform success. Hower, if the pocket be large and the eddy in it strong, it is advisable irst floor the pocket completely over with a mattress and then ild the dike, and in addition a portion of the sides of the pocket with advantage be paved. Indeed, at Hopefield Bend, as will e noticed, it became necessary to use a dike to break up an eddy in pocket that had been completely revetted, the eddy current being strong as to make the revetment of the pocket difficult to main

tain.

Indeed, as a final conclusion, it may be said that it has been arly shown that for use on this river only the continuous type of rk should, in general, be used; that such continuous work be conructed of the fascine subaqueous mattress of proper width, with an upper bank paving of stone and spalls on a properly graded ope, or of some equivalent construction.

With this type of work and with proper care in its construction and maintenance it is believed that any of the banks of the river an be successfully protected.

The materials used in this revetment are brush and stone, of hich the supply has so far been ample. Some years ago it was ared that the supply of brush would prove insufficient if much work had to be done, but from appearances at present it would em that the work will always be stopped by exhaustion of funds long before the brush is used up. Other material may, of course, be substituted, and, in fact, experiments are now in progress in the Fourth district with concrete as a substitute for rock and rough timber for brush.

CONSTRUCTION OF THE PRESENT STANDARD TYPE OF BANK REVETMENT

As the method of constructing this type of revetment has not been fully described at any one place in this paper, the description of different parts being scattered through it as improvements have from time to time been made, there will be added here a compilation of these different details, the plant used in constructing the revetment, the material of which it is composed, and the method of doing the work, and for further illustration see photographs printed herein and plates in accompanying atlas.

PLANT

The floating plant needed by a revetment party consists of moor