ence, the bank has grown out and the bar has consolidated to abst the general line given by the ends of the dikes. This system, there fore, as far as its immediate object is concerned, has been a gre success. At the present time a few decayed remains of the tops of the piles can still be seen sticking up through the sand bar.

The Osceola-Bullerton system has also been a success. The trances have all built up to the general shore lines of the bar, and except for one slough at each entrance, to about the same height & the older bars. By 1890 the entrance between Osceola and Bulla. ton bars had built up to such a height that it was decided to ret along its front. At that time this new formation was practically & high as the older bars except at four places. One of these was a slough about 100 feet wide, open at the 6-foot stage; the other three, aggregating about 500 feet in width, were open only at t 15-foot stage or above. This opening was originally over a mi wide and was the main channel of the river at low water. Sin 1890 three of these sloughs have been permanently closed, and ons one, the deepest, now remains, but that is open only at higher stags. At the middle and upper Osceola entrances much the same cond tions now exist. At each is a narrow slough open to the passager water at high stages. Behind these bars the chutes have narrowe and filled up and are so overgrown with willow and cottonwoo that even when water passes through them the current is mu slower than in the main river, and as a result a deposit is taking place and the chutes are closed still further. At a few spots loa scour has taken place, but the amount is inappreciable. At ab mid-stage steamers still can get into the lower end of Bullertar Chute and can go part of the way up it, but at low water this chr and, to a greater extent Osceola Chute, are perfectly dry, and ro lead across them.

I

SEASON OF 1893

Gold Dust Dam.-(Plate LXIV.) At Gold Dust Chute t Elmot and Island 30 dikes, built in 1890, had been gapped in th following high waters and had, therefore, failed to close the chu Many local but minor changes took place from time to time, t year one part, the next year another part, of the whole dike woul be uncovered in a decayed condition, but up to 1893 no gener changes of importance had taken place since 1888. The slight f in the triangle at the upper corner of the dike system was still existence, though not quite so high. The broad flat in the mid

f the chute had not changed much, except that the deep channel hrough the gap in the lower cross dike had filled up above the lowvater plane. At the lower end of the chute the only change was hat deep and narrow channels had been cut through the gap in he Elmot and Island 30 dikes, and that the current through the atter dike had cut away a part of the Tennessee bank and with it he levee constructed in 1886. These deep cuts through the gap lisappeared a short distance above and below the dike. The chute is a whole was still closed at low water and its cross section had liminished.

As a still further closing of this chute was considered desirable, nother attempt was made in 1893, and as the previous experiences with pile dikes had shown them to be unsuited to the locality, a rush and stone dam was this year constructed across the chute. This dam was located (Plate LXIV) about 9,000 feet below the head of the Gold Dust training dike, or between Cross Dikes Nos. and 5. This was at the time the narrowest part of the chute and required a dam 2,900 feet long.

As the approach to the dam site was closed to boats at any but the higher stages of river, 14,000 cubic yards of stone were boated n during high water and unloaded on the banks at either side of the dam, about two-thirds on the Tennessee and one-third on the bar bank. Before the completion of the work, however, these quantities proved to be insufficient and had to be supplemented by nearly 5,000 yards additional, unloaded on the outside bank of the bar and transported across it.

The elevation of the bottom of the chute along the line of the dam varied from 3 to 13 feet above low water, and the dam was built up to the 16-foot stage. The bed of the chute at the site of the dam was composed mainly of sand, but there were at several places pools or depressions filled with a sort of ooze of soft mud or quicksand. The first work of construction was to fill up these oozy places with brush. This was put in without a grillage to the thickness of from 6 to 30 inches, the brush being placed transversely to the dam, or longitudinally to the general direction of the chute. me tr The next step was to build in place the sill mattress of the dam. This sill mattress was composed of two layers of brush with a pole 3 grillage on the top and bottom, tied together with wire; the lower layer of brush being longitudinal to the dam and the upper transmverse. The sill mat, when completed, was 3 feet thick, 100 feet in wide, and extended the whole length of the dam. The upstream

The

edge was stepped off, and at the downstream edge the upper layer of brush was extended over the lower one and sloped gradually downward.

On top of the sill mattress the other mattresses composing the dam were built, each composed of two layers of brush at right angles, with a top and bottom grillage, but the top grillage of carh mattress was utilized as the bottom grillage of the next. These gri mattresses were all the same width, 50 feet, but varied in length and thickness, this thickness being varied so that the top surface of each crib mattress might be made level irrespective of the irrega larities in the bed of the chute. The lowest crib mattress lapped the sill mattress on the upstream side by 8 feet, loose brush being laid longitudinal under this mattress to fill all holes. The next one lapped the first in the same manner by the same distance, and where it extended beyond the ends of the brush, the upstream edg was made fairly straight, and the lap on the sill mattress was here 16 feet. In the same manner the succeeding mattresses were built; a total of four crib mattresses being required at the deepest plac and only the sill mattress at the highest point on the line of the dam.

After the completion of the brush work, the dam was covered with a close layer of stone, varying from 6 to 8 inches in thickness, depending on locality. Some of the lower crib mattresses on th Tennessee side had each its own layer of ballast.

As an additional protectional against scour by overfall, foot mats were built below the dam in the localities of the swiftest current. The banks against which the ends of the dam abutted were protected in the usual manner by revetments 300 feet long on both banks.

Further details of the dam construction are shown on Plate XIV. Its cost was about $93,000.

During the period of construction the river was very low and the chute in the vicinity of the dam practically dry, but a few days after the completion of the dam, in January, the river began to rise slowly. The dam was impervious enough to cause a considerabl hydraulic head, and by the time the river had reached the 9-foot stage the head was as much as 3 feet. At this time the dam began to settle, but this settling was not uniform, being greatest where the soft, oozing bottom had been found before construction, which showed that the cause was the flowing out of this ooze under the pressure of this hydraulic head. The greatest settlement was about

In spite of the damage to the dam by the high water, large de pressions were made in the chute both above and below the dam. About 800 feet above the dam the cross section of the chute now showed no place below the 7-foot stage. (Plate LXV.) The project for the repairs called for the restoration of the elevation to 14 feet above zero where the dam was now below that level and the mattressing of the washout below the dam. As no plant could be floated into the chute, the mat for the washout was made in the following manner: Fascines of proper length were built on the dry bar, rolled into the pond, and then hauled into place and sewed together with one-fourth-inch strands every 8 feet of their length. Enough fascines were thus placed until a raft, sufficiently buoyant to carry the run plank and the necessary men, had been built, after which the fascines were built on the edge of the mat and then rolled into the pond and incorporated with the others. Care was taken to so shape the fascines that they would fit the contours of the pond, and the sewing strand was hauled perfectly taut, so as to have a dense and compact mat. Fascines were built 18 inches thick and were placed normal to the current. The mat was lashed to the old dam with wire strand, so as to prevent its sliding away in sinking, but the other sides were left free to move as the mat was sunk, so that it could accommodate itself to the bottom, the margins being afterwards covered with extra fascines sewed on to the mat. Stone for sinking was wheeled on as long as it could be done without wetting the men too much, after which the sinking was completed by means of a small scow built on the site.

To repair the break on the original line of the dam would have brought part of its base into the deep water of the washout; hence, in closing, a deviation upstream of 25 feet was made, thus making the new work curved instead of straight. For the base of the closure work a fascine mat was built similar to the one for the pond and extending 30 feet beyond the base of the dam. Upon this mat a course of brush 30 inches thick was laid and lashed down to the mat with wire strands 8 feet apart, the strands being normal to the brush, which latter was laid normal to the current. This whole structure, 4 feet thick, was then loaded with stone until its upper surface was about awash; another course of brush was then laid and ballasted, and the process continued until the whole rested on the bottom. Work was then continued by adding similar layers of brush, and loading them until the desired elevation was reached. Side slopes of 1 on 3 were carried up as the work was built up. The