equal stress. As each pole, when tied down, formed a sort of fat arch, much of the brush toward the middle of the layer was imperfectly held down and was liable to displacement in sinking, thereby not only causing much of the stone ballast to drop through, but also leaving openings through which the current could scour the bottom.

The same types of mat were used as foot mats for dikes and as subaqueous mats in bank revetment. For the former the width varied from 40 to 100 feet, while the width of mattresses for revetment was made from 100 to 175 feet wide, the largest proportion being of the first mentioned width. They were constructed at all but the highest stages of river, and in one case a mat was afloat when the bank which was to be protected by this mat was entirely submerged.

No mooring barges or other device were used for holding up the head of the mat during construction or during sinking, and the absence of such a device was the cause of much damage to, and sometimes of loss of, the mat. In some cases, after the first launching, and owing to the stiffness of the poles and the weight of the mooring cables, the head of the mattress would lie under water, and where the current was strong enough the outer corner would be bent downstream, and in some cases it would be broken off. Mats were then started with a width of 35 to 50 feet near shore and were then gradually widened out as the work progressed. This, however, entailed a loss of work, as the length of mat was estimated for the full width only. In the case of one of the 175-foot mats a timber boom for holding up the heads was tried, but with such poor success that it was used but once.

To prevent the mat from being carried into the bank with a rising river, guide piles were driven about 25 feet apart along the inner edge of the mat, the pile driver keeping just below the mat barges. Yokes to encircle these piles were built into the mat. At first wire strands were used in place of yokes, but these would bite into the wood during sinking and hold up the mat, hence yokes made of poles were substituted. Even these would sometimes hug the piles and in such cases they had to be forced down from above. Frequently, especially in strong current, the work would crowd hard up against the piling, necessitating an extra amount of ballast to sink it to the bottom.

The absence of mooring barges or other suitable device to hold up the head of the mat was particularly felt during the operation

of sinking, always the most critical part of construction, and with the crude methods then in use it was difficult to foretell whether the mat would go to the bottom or down the river. Losses of work averaged 40 per centum, incurred mostly in the operation of sinking.

Two methods of sinking were in vogue (Plate XXVI), the downstream method and the upstream. In the former the stone barge was brought to the outside of the mat and a little below the head, and stone ballast was thrown off so as to submerge the mat sufficiently to permit of the loaded stone barge being floated over it, care being taken to keep the mat-head afloat for some time, for should this be depressed too soon the current would force it under and most likely destroy part or the whole of the mat by causing the mooring lines to pull out. The stone barge was never brought squarely across the mat, but in such a position that the outer edge of the mat was always sunk well in advance of the inner one, this angle tending to deflect the current instead of bringing the whole force of it against the submerged part of the mat. The stone barges had to be maneuvered by lines from shore, as no towboats were used. By changing the angle of the barge with the current the former could be made to approach the shore or recede from it. But this method was slow and difficult.

The second method of sinking--that is, from the downstream end of the mat upward-was decidedly the safest, especially in strong currents, as it kept the head of the mat well afloat until nearly all of it had been sunk, but it required more stone, as the tendency of the current was in this case to keep the mat afloat. Here the shore edge of the mat was sunk well in advance of the outer edge, thus deflecting the current instead of presenting an almost vertical surface square across it. As the stone barge had to be gradually hauled upstream, in addition to working it back and forth over the mat, this method was extremely slow. The best part of two days were usually occupied in sinking a mat.

The absence of any proper device at the head of the mat allowed drift to accumulate in front of and underneath it. This latter had to be sunk with the mat and sometimes, when sinking was progressing fairly well, saw logs or other heavy drift would break through the mat while still afloat, seriously damaging it. Again drift would sometimes run so strong that it began to pile on top of the head of the mat, depressing it by its weight sufficiently for the current to turn it under and tear out the mooring lines.

The mats were held in position by 2 and 12-inch manila lines fastened to trees, stumps, or "dead men" on the bank. From four to six of these lines held the head, and as the mat increased in length other lines were added.

At first the revetment work consisted of a mattress below low water only, but experience soon showed that at high water caving would take place above this low-water mat, and the revetment of the upper bank was thus shown to be necessary. Generally, the upper bank along a caving shore is steep and bluff, and before it could be revetted it had to be graded. Grading was done with the hydraulic jet, with a pump pressure of from 115 to 125 pounds, this giving at the nozzles from 80 to 90 pounds. Higher pressures would have been desirable, but the hose used at the time would not stand more. Usually three nozzles were used on each grader, one 14 inches and two 114 inches. The large nozzles moved the dirt, but the smaller ones gave the best results, as they wasted less water and would give a smoother slope.

In grading a bank it is essential that the proper slope should be made with the jet, if at all possible, as any trimming afterwards should not be encouraged, as it leaves the trimmed places loose and untamped. The obstacles to rapid progress of grading are buried logs and standing stumps. These have to be washed clear of all soil and then pulled out or cut down.

The grading of the upper bank was usually done before the mat was constructed, the reason for this being that it was expected that the material washed down would improve the usually steep and very irregular under-water slope. But there being in this case no mattress to stop the scour below low water, the caving would frequently go on unchecked and all the graded work would be lost before the mattress could be built. In consequence of this, a change of method was made so as to grade the bank after the sinking of the subaqueous mat.

During these first seasons the slopes to which the banks were graded were about 1 vertical to 2 or 22 horizontal. The upper bank protection was constructed in place. It consisted of brush laid normal to the river upon a grillage of poles spaced 6 feet, the courses of brush being laid shingle fashion, with the tops always toward the water. The first or lowest course was pushed outstream as far as practicable, the object aimed at being to make a lap over the previously sunk mattress. A top grillage corresponding to the bottom one was placed over the brush and the whole wired together

every 6 feet, after which the brush was covered with enough stone to hold it down when submerged.

During the season of 1882 the following work was done :

Ashport Bend. (Plate XXVII.) Revetment work was begun in 1882 at the head of the bend. This work was started at the foot of Forked Deer Chute and carried down to Ashport Landing, a distance of about 2,700 feet. It was originally intended to revet the whole of this bend down to the head of the Gold Dust Dikes, but when the work had progressed as just stated the need of additional mat parties to construct foot mats for the dikes caused a cessation of the work. The Ashport mat was of the woven type, 138 feet wide and 2,700 feet long. It was successfully built and sunk in three sections of about equal lengths, the success being mainly due to slack current and shoal water at the head.

Prior to the construction of this revetment the bank had been caving badly, and after its construction no further caving took place. Although, at the time, this was attributed to this revetment, it is doubtful if that had much, if anything, to do with it. Rapid caving with change of shape of banks was going on in the next bend above, and as a result of this the point of attack in Ashport Bend was moved below the revetment, while a deposit of silt was started at the head of the work. This filling has continued since, and there is now a high bar a quarter of a mile wide in front of the site of this first revetment, under which the work lies deeply buried. At Upper Osceola Bar (Plate XXVIII) revetment work was begun with the intention of completely revetting this bar between the upper and middle dike. This work began on the chute side of the head of the bar with a mattress 50 feet wide; on the outside of the bar they were made 100 feet wide and were completed to the end of the middle dike. About 1,400 feet of the upper bank was laid with brush, but only a small part was ballasted. Work on the lower bar below the middle dike was started, but before the first mattress was sunk the rise came and stopped the work.

Bullerton Bar.-(Plate XXIX.) The revetment work at this bar was started on the chute side of the bar about 500 feet below the head, and was then carried up and around the head and down the front side. About 4,500 feet of subaqueous mat 100 feet wide was built and sunk. All this mattress was built on the wire net, as above described. The upper bank was graded and the upper mattress work was started; about 2,000 feet of the mat, from 50 to 75

feet wide, were built in place, but when the high water came a large part of it was without stone ballast.

SEASON OF 1883

The high water of 1883, as has been stated, did no damage to the Ashport revetment, but this can not be said of the other work.

At Upper Osceola Bar (Plate XXX) much damage was done. Part of the upper bank was still unprotected, and the part where brush had been laid was still unballasted. This unballasted brush floated off and the whole stretch of upper bank was badly scoured, this extending in places down behind the subaqueous mat. This caving was especially bad near the upper end of the middle dike, where the caving was so serious as to leave the end of the dike exposed and allow the current to get behind it.

The mattress under construction along Lower Osceola Bar was caught afloat and washed away before it could be sunk. Some little caving took place near the head of this bar, leaving the end of the middle dike exposed. Along the middle of this bar the caving was small, while near the foot of the bar there formed around the head of the Osceola-Bullerton Dike a sand bar; however, the upper bank at this point was cut away slightly.

At the end of Bullerton Bar (Plate XXXII) the upper bank work had been caught almost entirely unballasted; a large part of the brush work was, therefore, washed away and the bank thus unprotected scoured down and behind the subaqueous work; the worst place was just at the end of the dike, where a large cave occurred, leaving this end of the dike exposed.

As soon as possible after the high water of 1883 revetment work was resumed and was continued throughout the season with the same type of work as used during the preceding year. Only one important change was made; it being found impossible to make the mattress conform to the minor irregularity of shore line, small connecting mattresses to cover these places were made and firmly wired to the main mattress.

At Upper Osceola Bar (Plate XXXI) no work was done, except the connection of the still existing revetment at the head of the bar with that built on the end of the new Cross Dike No. 1 constructed this season.

Lower Osceola Bar.-Work was resumed here in the summer of 1883 by constructing a mat 1,020 feet long on the inside shore and lapping it around the head. This mat was 100 feet wide and of the