and sunk. The upper bank was graded by hand and paved for a length of 1,610 feet, the paving being carried to different elevations, usually to from 23 to 37 feet above low water, though in one place where the toe of the levee was quite near to the top bank line the paving was extended all the way to the top of the bank, or to the 40-foot stage, for a length of about 400 feet. A crib dike was placed at the extreme lower end of the work for the purpose of breaking up a pocket eddy. This dike extended from the zero stage across a depression to the levee above, a distance of 350 feet; it had a height of 6 feet, a volume of 907 cubic yards, and was paved with stone. In addition to the above, a brush crib paved with stone was placed on Dike No. 4 to repair damage by decay and flood; and repairs, but of stone only, were made to Dikes Nos. 2 and 3.

The mat which broke during the process of sinking was 1,057 feet long and from 280 to 300 feet wide. It was of the usual fascine mat construction, with five-sixteenth-inch bottom and one-fourth-inch sewing strands, and had in addition four one-half-inch and one fivesixteenth-inch wire strands run on top of its outer portion its entire length to give extra strength for sinking in a current then observed to be 81⁄2 feet per second along the outside of the mat. The mat-head was also made stronger than usual, being of hardwood poles and about 3 feet in diameter. The shore anchorages from the head were twelve steel cables of 1-inch and 11g-inch diameter, the larger being at the outer portion. The Helena gage read 17 feet on the day of sinking. The head of the mat was located a short distance below Dike No. 4, only 25 feet separating the dike from the mooring barges, and against these a mass of drift had accumulated. Some of this drift had lodged under the barges before the driftcatching chains were placed, and it was probable that some of it had passed under the barges and had lodged under the mat. In sinking the mat it went down as usual, though with extremely heavy strains on the lines, and it was assumed that about 350 feet had reached the bottom when the mat suddenly tore in two 430 feet below the head, the lower part drifting down the river and the upper part going to the bottom apparently in good shape. The combined breaking strength of the wire strand running through the mat longitudinally was about 50 tons, not including the sewing strand, and this strength was considerably increased by the longitudinal poles, which were placed on the mat at intervals of 8 feet and were well wire-spliced together and to the mat. The accident showed the mat to be deficient in strength for sinking in such a strong cur

ent, so the strength of the next mat was considerably increased. This mat was 700 feet long by 280 feet wide and was constructed during a moderately high stage with swift current, though sunk Then the river had fallen to the 10-foot stage, and the current had somewhat decreased. In the construction of this mat additional strands of five-sixteenth-inch and one-half-inch diameter were run from the head at each shackle strap through the entire length of the at, making the combined breaking strength of all these straight aire strands fully double that of the broken mat.

The high water of 1897 damaged the work slightly in several places. There was some caving along the two upper unrevetted spaces between the dikes, a slide of the bank between Dikes Nos. 3 and 4, and a settling of the shore end of Dike No. 4.

SEASON OF 1898

The high water of 1898 increased the damage considerably, and when the water fell the condition was about as follows: The cave between Dikes Nos. 1 and 2 had a length of about 200 feet and a maximum width into the bank of 60 feet, and the cave between Dikes Nos. 2 and 3 was of practically the same dimensions. The shore end of Dike No. 4 had been flanked by the river, which had eut into the bank and had also undermined the end of the dike, allowing it to settle considerably. A fault through the dike at the extreme lower end of the 1896 work was produced by the settling of the lower part of the bank for a vertical distance of about 5 feet, this fault extending for some distance along the paved slope above. The bank below the end of this revetment had also caved somewhat. The work this year consisted in repairing the damaged work and extending the revetment downstream.

In repairing the old the following work was done: Revetting the interval between Dikes Nos. 1 and 2 with a mat 190 feet long by 175 feet wide, on which was placed a crib dike 180 feet long by 10 feet maximum height. The interval between Dikes Nos. 2 and 3 was similarly treated with a mat 216 feet long by 175 feet wide and a crib dike 190 feet long by 10 feet maximum height. The shore end of Dike No. 4 was repaired with a crib 100 feet long by 11.6 feet maximum height. The settled portion of Dike No. 6, which is the dike at the end of the 1896 work, was built up with a crib 130 feet long by 10.5 feet maximum height. For reinforcing the revetment a new dike was placed across a depression about 550 feet above Dike No. 6. This dike is called No. 51⁄2 and is 177 feet long

by 16 feet maximum height. The exposed ends of all dikes were well paved over with stone to prevent decay of the brush, and the paving along the 1896 revetment was repaired at a number of places.

The new work consisted in extending the revetment downstream with a fascine mattress 976 feet long by 250 feet wide, lapping the old work about 50 feet. To join this mat with the shore two connecting mats were built, one along the middle and one at the lower end. The upstream half of the mattress was placed along a ridge formed by a large bank slide. This ridge had an elevation of about the 14-foot stage, and back of this was a depression and then a gradual slope to the top of the bank. The paving here was carried to the top of the ridge, and two spur dikes were built to connect the ridge with the top of the bank to prevent the increase of the depression by scour. These cribs are known as Dikes Nos. 7 and 8 and are, respectively, 110 feet long by 6 feet high and 125 feet long by 5 feet high. The bank along the lower third of the mat was bluff, and this was graded and paved to the 28-foot stage.

The extension of the revetment amounted to 926 feet and made the total length of bank revetted over 4,900 feet, all continuous except two gaps, one just above Dike No. 1, 500 feet long, and the other between Dikes Nos. 3 and 4, about 200 feet long. Along the former gap there had already been some slight bank erosion.

SEASON OF 1899

The high water of this spring considerably increased the bank erosion along the upper gap, extending for some distance up along the mat of 1889, the upper bank revetment of which had been omitted on account of the tenacious materal of which this bank was composed.

The season's work consisted in revetting the gap and paving the upper bank along it and the 1889 mat. One subaqueous mat, 600 feet long by 250 feet wide, was sunk across the gap, and three connecting mats along the 1889 mat to join it with the shore, from which it had become somewhat separated by the caving. The upper bank was then graded and paved down to the head of Dike No. 1, the paving being carried to the 23-foot stage. Dike No. 4 was also reinforced with an additional crib, and a small quantity of stone placed on the shore ends of the four upper dikes.

The work of this season left the work continuous from the head for 4,900 linear feet, except the gap of 200 feet between Dikes Nos. 3 and 4.

No work has been done here since 1899, and the work itself seems be still in good condition, but there has been a continuance of the cracking that has been going on in the bank. These cracks are ongitudinal, and along some of them there has been a vertical setting of as much as 5 or 6 feet. Outside of these cracks the work is apparently unharmed, and the only cause that can be assigned for this cracking is the removal by underground flow of some sand <ratum, after which the superincumbent earth settled into the roid space.

The cost of this work from 1896 to date, inclusive, has been $124,000 and the total cost since the beginning has been $199,000.

CONCLUSIONS AS TO BANK REVETMENT

The history of the bank revetment constructed in these districts is Do completed. Beginning over twenty years ago, the work has progressed with varying success; alterations have been made from time to time as experience showed them to be desirable, until there was at length evolved a successful type of work that seems suited to the conditions met with here. During this period a number of different types of work have been tried and much experience has been gained, and before closing the subject it will perhaps be well to state briefly the most important conclusions that have been dereloped.

The object of bank revetment is to protect the banks of the river from destruction by the currents. The active agency in this bank destruction is erosion, and bank revetment must therefore prevent this erosion. Perhaps the easiest and most obvious method of doing this is by laying over the exposed surface of the erodable bank a non-erodable covering, and this is the method adopted. This nonerodable covering must, of course, be fairly durable, must have sufficient strength to withstand any strain put upon it, and must be free from interstices through which scour might take place.

Owing to the fluctuations in the river, the different parts of the bank are subjected to different conditions. Below the low-water line the bank is always wet; above that line sometimes wet and sometimes dry, and this difference in conditions allows, if it does not require, that the portion of the bank above and below low water be treated differently.

Above low water the bank can be seen and the non-erodable covering can be laid with ease, and it has been found that here a properly laid stone pavement forms an efficient protection. The in

dividual stones of this paving must, of course, be so large that they can not be moved by the current, and it was found by experience that stones of this size are not of themselves sufficient, as through their interstices scour could take place when the bank was sandy and the current strong. This has been prevented by first covering the bank with a layer of spalls or crushed rock of such thickness and closeness as to prevent scour through them and to hold this down with a layer of larger stone. Such protection, when carefully laid, has been proven to be all that is required for the upper bank.

Below low water the conditions are different. As it is impossible to construct the non-erodable covering actually in place, the next best thing is done, constructing it on the surface of the water above where it is to lie, and when constructed sinking it. In this way the continuity of the covering can be assured, and it can be placed exactly where it is needed. As has been stated, this covering must not contain interstices through which scour can take place, for in such a case not only would the purpose of the covering not be fulfilled, but by such scour its own eventual destruction would be assured.

The covering actually used has been a mattress of brush, and the only kind that has been found to prevent scour in very rapid currents is the fascine mattress. In moderate currents another type of mattress might do, but as changes in current conditions are always likely to happen the fascine mattress only should be used.

As the immediate object of bank revetment is to prevent erosion, it must cover all parts of the bank where erosion can take place, or at least where its effects are at all serious. As at one stage or another erosion may take place at any point between the top of the bank and the foot of the subaqueous slope, it is necessary, in order to stop all erosion, that the revetment work should cover the entire bank from its top to the deepest water.

In the work that has been done the paving has only in few cases been carried to the top of the bank; in the early works, because it was not considered necessary, and since 1897, because of the insufficiency of funds. The Hopefield experience of 1897 showed the absolute necessity of such an extension in very exposed places, and as every place where revetment is needed may be at some time subject to severe attack it is safest, and usually cheapest in the long run, to extend the paving to the top in the beginning.

The subaqueous mat should, in general, and where possible, be