other mat lines appeared to have good leads and no heavy strains thus far, it was decided to sink the outer portion and try to hold the inner portion by the slip lines. After the outer part was sunk some distance and all appeared to be going well, the drift started to move, buckled the mat in the middle, and tore away everything but the outer part of the head, which remained attached to the head lines, and the broken mat, drifting downstream, was, of course, lost. Only one 12-inch line was broken by this accident. The experience gained showed that spars could not be depended upon to hold the mooring barges from the bank, so large piles were driven and well braced to the bank, to form an abutment for the mooring barges. A second sill mat was immediately built and the strength increased. The one lost had lines of heavy poles, forming an extra head at 8 feet below the main head, and the new mat had a main head composed of five layers of large hardwood poles, and extra heads at 8 feet and at 16 feet below the main head, and the 2-inch sinking straps were made fast to all heads, and sinking lines were attached to these straps, the outer being 212-inch diameter, then five 2-inch and two 111⁄2-inch were attached, the latter being near the inside, and all of these were given long leads and strained as evenly as practicable.

The river rose to 13.4 feet on September 14 and fell to 12 feet on the 17th, the day of sinking. During the rise considerable drift had piled against the mooring barges; some had passed under them and lodged under the head of the mat; this was not an unusual occurrence, and the anchorage lines having been placed before the drift accumulated no special apprehension was felt about a successful sinking. The mat was heavily ballasted and the operation of sink-. ing conducted as usual. The head went down nicely until near the bottom, the slip lines leading nearly vertical and the water running over the entire length of the mat on the outside, when suddenly a movement of the drift under the mat bulged up its middle portion and broke it just below the head for some distance on each side of the middle. It held thus for several minutes, when an attempt was made to sink the outside portion, but this caused further drift movement and produced strains enough to break the outer 22-inch strap lines, and the mat then tore entirely across just below the 16-foot head. During the accident the two inside 11-inch mat lines broke, but all other fastenings remained intact. The loss was due to the drift, and the manner in which the mat broke was precisely similar to the breaking of the Memphis Harbor mat in

the mat. as the thrust a22.00 the 18AL VISI TE CAP &

account of the angaras of the fame v

nated. Most of the site,

with success, One agrident, be you and Tied part o sulted in the loss of the largest one tilt fining the wawn. Tha was the bottom erib of Dike No. 4. worn was 30 feet 7ez of which the inner portion, for a letzte of a fhe way if fat må and the balance, 210 feet long, was 30 feet wide the mar el no taining 87,120 cubic feet. The river at the date of éting was at only the 5-foot stage and rising, but maer drift bad artmadd at the outer end of the mooring harges and wone bad gotten inår the outer end of the crib. This drift had passed around the end of the drift barges above, which extended only 240 feet out from the bank, and while these stopped most of the drift they did not projeet far enough to stop it all. The crib was sunk in the usual manner, with a large amount of ballast, and had apparently landed safely on the bottom. After it had been down about two hours, alternate mooring lines were slacked, the pins removed, and the lines taken in, and at this time no extra strain was noted on the remaining lines, but about an hour later extra strain was observed to come on them; this was considered to be due to a slight shifting or Nettling of the crib, and as a safeguard more ballast was cast on it. A few hours later the remaining lines parted and the outer portion of the crib for a length of 160 feet turned over, dumping the stone from the pockets and rising to the surface. The inner and wider portion, with lines still attached and which was very heavily ballasted, held in place, with the turned-over part bending up from it to the surface and training down with the direction of the current. Lines were then attached to this and lead to the mooring barges, and after two days' work the crib was brought to very near its

original position, where it was properly anchored and sunk. The cause of the anchor was the drift under the crib, which held the upstream side off the bottom, thus tipping it and exposing such a large surface to the current that the pressure overturned it.

In October, during the progress of the work, a large cave occurred along the site of Dike No. 4, the sill mat of which was at that time under construction, with all the plant for building it swung out in the river. The caved territory was about 800 feet long, extending from about 200 feet above Dike No. 4 to about 100 feet below the location of Dike No. 5. The maximum amount of settling of the surface was about 30 feet, and it extended inshore to the middle of the levee or railroad embankment, the track on which settled from 6 to 8 feet for a length of 180 feet. All along the site of Mat No. 4 there was no disturbance near the water line, the settling not extending that far, but leaving a ridge of undisturbed ground there with badly shattered earth back of it. Nearer to and along Dike No. 5 the cave extended to the water line, but very little earth went into the river here, and the shape of the low-water contour at this place was changed very little by the caving. To determine what became of the earth displaced by the settling, soundings were made all along the caved territory, and these, compared with soundings made a few days before, showed no appreciable change. The inference, then, was that below the surface there was a stratum of quicksand, the fine material of which had been carried into the river by underground streams, leaving the more compact top stratum undermined, which, when the support became too weak, fell in.

Although the work done this season was incomplete, it prevented further caving for about eight years; but in recent years repairs and extensions have been necessary, but as these repairs and extensions were made at times when the fascine mattress was used a description of this work will be given later.

SPUR DIKES AT OTHER LOCALITIES

The only other locality where spur dikes alone have been used to protect a caving bank is at Caruthersville, Mo. (110 R), where in 1898 work was begun on a set of four dikes for this purpose. All the sill mats were placed, but only one crib could be built before the rising river stopped the work. When work was resumed the following season the plan was changed, and the dikes were abandoned for continuous revetment work. This will be described later.

Spur dikes have been built at a number of localities along pre

viously revetted banks as aids to this revetment, especially in "pockets" in these revetted banks, with the design of breaking up the violent eddy currents, which generally exist in such pockets and which are detrimental to the stability of the revetment. In some places these pockets were entirely covered with a floor mattress and the dike built on this; in other places, as where a large pocket has resulted from a recent failure of the revetment, the dikes were founded on the usual sill mats only. All such auxiliary dikes eonstructed at these various places will be described later.

REMARKS ON SPUR DIKES FOR REVETMENT

The spur dikes constructed at Columbus, Memphis, and Helena were primarily for the protection of caving banks, and for this purpose they were fairly successful. Reviewing the works at these localities, it is found that only at Columbus did the original construction prevent further eaving, and this work has not been subjected to a very strong attack of the current. At Memphis, where the dikes were located across pockets in which were strong eddies, the dikes were too far apart to break up these eddies sufficiently to stop caving between them, and after the first high water additional dikes had to be placed between the main dikes to accomplish this purpose, and afterwards considerable revetment was also required along the upper bank to prevent seour. At Helena there was no scour between the dikes for about eight years, when an increased current caused caving along two intervals, and the protection of these required two additional dikes and mattresses. Later, the bank eaved above the head of the dikes, or between them and the mattresses above. The cost of the additions to the original work at both Memphis and Helena has materially increased the original cost of these works, which has become quite high per foot of bank protected, and undoubtedly the banks at both places could have been more effectively protected with continuous work and at less cost. No dikes would have been needed at Helena, while at Memphis some might have been needed in the large eddies, but these should have been made of much smaller dimensions than the spurs built. Even at Columbus, where no additional work was done, it is believed that just as effective a revetment could have been made with continuous work and at less cost. When the spur dikes were first built it was expected that the area between them would fill extensively, but all subsequent examinations showed but minor tilling, confined to the upstream sides of the dikes,

The conclusion deduced from experience is, that where the caving bank is fairly regular it is both better and cheaper to use continuous revetment and that the use of spur dikes is justified only in connection with continuous work and in certain indentations of the bank, where it is desired to break up dangerous eddies, and such indentations, or pockets, should be first completely floored with

mattresses.

FASCINE MATTRESSES

As has been previously stated, this form of subaqueous mattress was first tried in 1892 at Daniels Point, Plum Point Reach. The next year, 1893, it was adopted as the standard, and practically all subaqueous mattresses constructed since then have been of this type. Since 1893 the changes made in either the manner or method of its construction have been of minor importance, and the description here will be of the fascine mattress in its latest and most approved form.

It is composed (Plates LV and LVI and Photos. Nos. 9 to 14) of brush fascines laid side by side and usually placed so as to be normal to the current (except in connecting and floor mattresses, where, for convenience, the fascines are placed parallel to the bank). Each fascine is about 12 inches in diameter and as long as the mat is wide. In the mattress they lie side by side and are held together by wire strands about a quarter of an inch in diameter, called sewing strands. These sewing strands pass over and around each fascine, then to the next, over and around it, and so on, being hauled taut at each turn by means of tackle blocks. These sewing strands are placed at about 8-foot intervals and serve to bind the fascines tight and at the same time to hold each in place in close contact with the adjoining fascines. To give the mattress longitudinal strength, larger continuous wire strands run under the mat along each sewing strand and are sewed in by it at each fascine and clamped to the sewing strand at frequent intervals. Each fascine is composed of willow brush from 1 to 4 inches in diameter at the butt, and as the brush is laid so as to break joints, and the fascine is afterwards tightly bound, the fascine itself has considerable strength in the direction of its length, and this is usually sufficient to give the mattress all the transverse strength that is required.

In the manner of constructing this mattress on floating ways, of holding its head by means of mooring barges, of ballasting and sinking the mattress, and of keeping off drift, no great changes