sunk connecting mats to connect the dry bank with the river mat were constructed and ballasted, all brush and stone removed from the disintegrated revetment at the foot of Section A, and the work restored here. The grading and final completion of the bank work in this pocket were postponed for the present, as everything was safe against any further caving. River mat No. 2 was begun on the 20th and sunk on September 2. This mat was 1,150 feet long and, exclusive of shore connections, 200 feet wide, or, with latter included, from 230 feet to 255 feet wide. The method of construction followed for this and all subsequent river mats constructed during the season, except at Ashport Bend, was that laid down in "Details of construction" and its essential features are that the inshore edge of the mat is kept on the zero line by means of guide piles; that from this edge to the dry slope and as far up as desired the shore work is made a part of the river mat, and that by this metho of construction the connecting mats are done away with, thus effecting a saving of the material and labor lost by the necessary laps.

It soon became evident that one grader could not keep ahead of the revetment party, hence No. 4 was sent here on the 20th of August and began grading the next day. A small party of self-subsisting laborers had in the meantime been employed in clearing the bank in advance of the grading party. The intention had been to start a second revetment party here, but as the two graders could just about keep ahead of one party this had to be postponed. The cause of the apparently slow progress of the graders was primarily due to the large number of heavy deeply-embedded stumps, encountered. Each grader met with from 6 to 10 stumps per day, some as large as 4 feet 5 inches in diameter. The removal of these stumps from the slope was accomplished after grading, by blasting with dynamite, the blasting being usually done after quitting time. River mat No. 3 was begun on the 3d by constructing the head, getting out lines, cables, etc. On the morning of the 5th the plant was swung into position for weaving, and on the 16th this mat, 1,100 feet long, was sunk.

The preliminary work for mat No. 4, such as taking in cables from last mat, building head for the new one, putting out mooring barge and mattress cables, was at once begun by a small party, but as the weaving party had crowded the grading party closely during the construction of mat No. 3, and as the stonework was far behind, the weaving party did not swing into position until the 21st. Mat No. 4, 904 feet long, was sunk on the 30th of September. The preliminary work for mat No. 5 was done by a small party, the bulk of the force being engaged in bringing up the stonework. On October 5 the weaving party swung into position for mat No. 5 the last of the large river mats for this place, which they sank on the 14th. This mat was 915 feet long. It was constructed in the strongest current found along the bend, and sunk in water whose depth along the outer edge of the mat ranged from 46 feet to 79 feet below zero, the gauge at the time showing a stage of -1.10.

Second District Grader No. 40 having been put in order for work, and not yet being required for duty at Hopefield, was put to work on the 4th of September grading the bank in the pocket and worked there until the morning of the 12th, when it was sent to Hopefield. On September 28, Grader No. 4 having finished here was sent to Osceola Bar. On October 10, all grading having been completed here Grader No. 2 was sent up to Ashport Bend. Further and detailed information about grading will be found under the proper caption.

At the beginning of operations here some trouble and delay was encountered by the frequent oscillations of the river. A rise in the latter part of August compelled a stoppage of grading for five days. From the end of August until the final completion of mat work, in the middle of October, the weather and stage of river was everything that could be desired for rapid, good, and economical work. The difficulties encountered in grading will be more particularly enumerated under the head of "Grading."

The total length of river mat made here, exclusive of repair work, was 4,519 feet linear which is about 175 feet in excess of the length of bank operated upon, the excess being due to the overlapping of the successive mats. Further information regarding mattress construction will be found under the head of "Construction of Mattresses."

After the completion of the mattress work the party were employed in bringing the paving well up above the water line, repairing two small faults, one near the head of section B (1888), and one near the head of the 1888 work in section A. Both of these faults were located in strong eddies caused by projecting points just above, and to these the damage was due. The damage in neither case extending below the zero plane, heavy connecting mats to cover the damaged zone were all that were required, the deep holes being first filled with brush and dirt to restore the slope. On the 22d and 23d of October the force, having completed all mattress work and having brought the shore paving well above the then prevailing stage of river, was moved to Ashport bend with their plant.

The small party heretofore clearing bank and doing the other work preliminary to

the active revetment work were transferred to Fletchers Bend, where they began putting additional stone upon the old work, and in addition also began loading stone from the storage piles upon barges for use at Osceola Bar, all stone afloat on barges having by this time been consumed. They continued at these two classes of work until the 25th of November, when the Osceola Bar party having completed their work up to the 13-foot contour was transferred to here and both forces consolidated. The work of shore paving suspended on October 23d was resumed with stone taken from the storage piles. By December 1, all stone heretofore stored on the bank at this place had been expended. On the evening of that a tow of stone was received, but the urgent needs of the Ashport party only permitted the use of a small share of this stone here. After expending this the repair of a new fault, developed near the foot of the 1889 work section A, was begun. This fault in itself was of small magnitude only, but being situated almost at the head of the new work as well as on an exposed point it was deemed advisable to repair it in the very best manner. Hence, a river mat 100 feet wide by 225 feet long was sunk in front, a heavy brush crib built to fill up the hole or pocket, and a connecting mat overlapping the river mat placed over the restored slope, the whole being heavily ballasted. No more stone being available at present the force being employed here was largely reduced. Two more barge loads of stone were received here before the end of the month, with which shore paving was continued. On the 30th of October the party were transferred to Osceola Bar, and no further work was done here until January 15, 1892, when paving was resumed and continued at intervals as stone became available until February 12, when work was finally suspended for the season.

The originally contemplated height of the shore paving was the or 20 foot stage, but before much of it had been carried up to that plane the probable deficit in the supply of stone had become so apparent that this height was reduced 2 feet vertically or to the 18 foot stage. The difficulties of delivering stone on the works having been very much increased in January by the closure of the Mississippi River above Cairo by ice, and the approach to quarries on the Ohio being prevented by floating ice, this height was necessarily still further reduced, so that at the close of operations here the paving at some places did not extend above the 15-foot stage. It is believed, however, that the work as it stands now is perfectly safe until a more liberal supply of stone will enable us to complete it as originally projected. At the lower end of the section, or above what before the junction of the new with the 1888 work was section B, the current is very rapid and the water deep. Here the paving for 100 feet has been carried up to the crest of the slope as the current impinges directly against it. It may also be noted that while the slope was being graded here a strong vein of water discharging at the 2-foot contour was encountered. This was during the dry season, and hence it may be assumed that there is a permanent discharge here. As the soil around here was very soft, brush weighted down with stone and clay was embedded in the soft places and the shore mat laid over this. A phenomenon worthy of note occurred here during the rise in November. The inshore end of the head of Mat No. 4 was seen to rise to the surface and immediately a strip about 15 feet long and of nearly the width of the mat (200 feet) was turned over on the mat just like one leaf of a hinge folds back against the other half. This mat was sunk on the 30th of September, going down in fine shape, and I can only account for the above on the assumption that the lower end of Mat No. 3 and consequently also the head of No. 4 rested on some large stumps, of which there are many along the bottom here, and that therefore the first 20 or 25 feet of the mat had assumed a slope steep enough to permit some of the stone ballast to roll further on the mat, thus restoring some of the buoyancy of the head (which by reason of its construction is always the most buoyant part of the mat) until the increased current was able to raise and turn it over. Only one transverse cable was broken, but extra ones were afterward run from the mat to shore. A lot of bailed stone was thrown overboard in such a manner that it would drop on the damaged part, and by preventing any oscillation stop all further damage. It is probable that the piece turned over was not quite as long as the overlap of one mat over the other, for no bare bottom could be discovered with the lead.

Nothing has occurred here since.

The present condition of the new work here is as follows, viz:

Total length of river mat made

Total length of bank operated upon

Paving up to 13-foot stage (in pocket).

Paving up to 20-foot stage and over

Paving up to 18-foot stage.

Paving up to 17-foot stage

Paving up to 16-foot stage

Paving up to 15.5-foot stage.

Revetment-Ashport Bend, Tennessee.-The starting point for the revetment of this bend was located 3,500 feet below where the prolongation of the Ashport levee would strike the river bank, this being at the time above the locus of active caving.

The initial step was the cutting down of everything above 3 inches in diameter along the timbered bank for a distance of 100 feet back, this work, which was done by contract, costing $5 per acre. It was done in May and more for the purpose of preventing the formation of snags and wrack heaps; 23 acres were thus cleared. Nothing was removed at the time.

On October 1 a small party began the work of clearing the bank preparatory to the commencement of grading, and on the 5th Hydraulic Grader No. 4 began work at the upper end, being followed by No. 2 on the 10th. From the outcrop of the various strata composing the bank it was judged that hydraulic grading would be troublesome and the achievement of a good slope expensive. The bank is composed of layers of blue clay and sand stratified in some places, while other large pockets of sand flanked on either side by projecting points of tough clay were found. At one place the very expensive method of sluicing had to be resorted to. Details of this work will be found under grading.

On October 22 the Fletchers Bend party arrived here and at once began work preliminary to the construction of River Mat No. 1. There being no timber where the anchorage for the mat would come, "dead men" were used for the mooring and shackle lines. On the 24th the plant was swung into position for Mat No. 1, which, on account of the absence of current and the presence of other favorable conditions, was made 2,150 feet long, sections of it being sunk on two occasions before the final sinking. The river, which had, up to then, been at a stage below zero, began rising on November 14, and reports from headwaters indicating a heavy rise with much drift, the part of the mat still afloat was sunk on the 21st.

Mat No. 2, begun on the 23d, length 1,150 feet, was finished on the 10th of December, but, owing to the lack of stone, it could not be sunk until the 16th, by which date a small lot of stone had been received. In the construction of this mat, or rather of the shore work, a change was made. Heretofore the shore work had been joined to the mat so as to practically make a continuous piece of work from the out-stream edge up the dry slope. The line of juncture of the shore and river mat was considered a weak part of the work, as this line necessarily falls upon a part of the bank which, being always submerged, has a steeper slope than the graded portion, hence undue strains are cast upon the junction, tending to rupture or separate the one mat from the other. This was noticed in several cases of previous years' work. Therefore the following method was adopted: Anchor piles, 50 feet apart, were driven along the zero line. Against these the inner edge of the river mat rested, being secured thereto with yokes made out of -inch wire strand running across the widths of the mat and well fastened to it at every 16 feet. After the mat was sunk these piles were cut off close to the water. Connecting mats were then used to span the zone between the river mat and the dry slopes. The first 450 feet of this mat, being under construction before orders to change the method were received, were constructed in the usual manner.

At the beginning of the work here it was hoped that with two revetment parties about 8,000 feet of work might be accomplished during the season. When the second party had completed the Osceola Bar work on December 13, and were ready to be transferred to here, it was found impracticable to do so, as there was then a lot of work awaiting stone, only a small supply of the latter on hand, and the deliveries from the quarries not at all equal to the demand. The same cause, i. e., limited supply of stone and a large amount of unfinished work, compelled the cessation of mattress construction at this place when Mat No. 2 had been completed. As soon as this mat had been sunk, the force, which had been largely reduced, were employed in construction of the connecting mats and in paving the slope. The progress of this work suffered several interruptions from lack of stone. The force had been reduced by transfers to other works and by discharge, but a certain number of men had to be retained pending the arrival of stone. These men had to be subsisted during the period of no work, and this added considerably to the cost of the work. The original project was to carry all paving up to the 20-foot, or two-third stage, but before much of it had been carried that high it was found necessary to curtail the expenditure of stone by reducing the height of paving to the 18-foot stage. After the graders had completed the slope for Mat No. 2 and before the failure in the stone supply became apparent, they were moved below "Mud Point," which is a sharp point projecting from the bank, whose substrata are composed of a tough blue clay hard to erode. The intention was to run the revetment down close to this point, leave the point from its base outward unprotected, and then resume the revetment below. This, if carried out, would have cut away the point, leaving the bank in good shape for a subsequent connection of the two revetments. It was intended to start the second revetment party here, but, as already indicated above, this plan had to be abandoned. By the time that it became manifest that the supply of stone on hand and prospective

would not warrant the beginning of any work here, the graders had finished nearly 600 feet of bank, which may be considered so much lost work, as the slope is gradually caving in. The work in this bend, after suffering various delays from lack of stone and from natural causes, was completed for the season on January 20, 1892. The condition of the work is as follows:

The lower 550 feet may be considered as an experimental section, for the natural slope left by the receding water, and which required very little dressing, was taken advantage of and paved up to the foot of the bluff bank, rising from 12 to 15 feet perpendicularly above.

The intention was to distribute a quantity of stone along the top of this perpendicular bank, to be deposited on the slope as it caved in, thus forming a sort of riprap, but the prevailing poverty of stone prevented this being carried out.

It is to be regretted that a more vigorous prosecution of the work at this place was prevented by the scarcity of stone, for up to the latter part of December the stage of river and condition of weather were all that could be desired. Labor, too, was plentiful, and plant sufficient to work two revetment parties to their utmost capacity was on hand.

Caving is now quite active below Mud Point as well as immediately below the revetment, and an early resumption of work and completion of it during the coming season is strongly urged.

The work, as far far as finished, does not extend far enough downstream to have much effect on the sand bar opposite, whose most advanced point lies below Mud Point, but it may prevent its further growth.

Clearing of another strip along the timbered bank will no doubt become necessary shortly if the present rate of caving continues.

Daniels Point.-The work contemplated here called for an extension 500 feet upsteam of the revetment and the repair of a fault in the 1889 work about 400 feet below the head of it. Owing to the limited supply of stone, the former project had to be abandoned after the bank had been cleared and graded. The repairs were finished and consisted in a standard river mat 200 feet wide by 350 feet, long, two connecting inats 56 by 140 feet and 60 by 161 feet, respectively, to connect the river mat with the shore work or paving, a shore mat 36 by 140 feet, and a strip of paving 12 by 135 feet. Neither the shore work nor the paving were carried higher up than the two-thirds stage, the balance of the bank being left as found. Work was begun on December 3, 1891, and completed on January 9, 1892. Four days, from the 5th to 9th, were lost by reason of no stone being on hand to complete the work.

Osceola Bar. The project for this place was the extension upstream of the 1890 work 1,000 feet, putting a brush sill across Slough No. 2, and putting additional ballast on the 1890 work whenever the declining river exposed sparsely ballasted work. Work was begun on September 19, 1891. A sill mat 80 by 60 by 2 feet was constructed and sunk, which completed the closure of the slough; 2,076 by 355 feet of revetment was reballasted. While this work was going on a part of the force were clearing the bank, which was covered with willows, but no timber, preparatory to grading, and putting in dead men for the anchorages. Hydraulic Grader No. 4 began grading on the 28th. From the very beginning the bank sloughed and caved off badly, requiring a large amount of filling up and hand-dressing in order to obtain a fair slope. Various modifications in the use of the jet were tried, with a view of overcoming this difficulty, but as the bank was composed almost entirely of fine silt, quicksand, and in some cases beds of ooze, without success. Finally, after 290 linear feet had been graded, under great difficulties and expense, grading was discontinued, and as the cost and slow progress of sluicing made this method of grading prohibitory, it was decided to sink the river mat first and then resume hydraulic grading.

The mat party swung into position on the 6th of October and sank the mat on the 26th. It was 200 feet wide, by 1,037 feet long on the outer edge and 987 feet measured on the inner edge, the difference in length being caused by the convexity of the shore. This mat ran well down below the head of the 1890 work, which, as usual, had suffered some slight disentegration from the erosion above.

For 290 feet, where the bank was already graded, the shore work was connected directly to the river mat.

Grading was resumed as soon as the river mat had been sunk, and connecting mats and shore work were placed in position whenever a sufficiently long piece of bank was graded. Grading did not work well at all, even after the mat was sunk, but by means of a large amount of manual labor a good slope was obtained. Owing

to the oozy nature of the material near the foot of the slope, shore work was laid as a foundation for the paving, the shore work being carried up to firm bottom; 50 feet of the above water part of the 1890 work which was badly undermined and broken up was cut away, the slope restored, and a good junction of the old and new work effected.

All brush work was completed here on November 14, but the completion of the paving up to the 13-foot stage was delayed for want of stone until the 25th, when the party were transferred to Fletchers Bend. No further work was done here until December 30, when, anticipating the arrival of stone, a small party were sent down with instructions to carry up the paving to the 20-foot stage as soon as the stone was received. Information received the following day, however, destroyed these hopes and the party were therefore disbanded. No further work was done here until January 20, 1892, by which time one barge containing 396 cubic yards of stone could be spared from the work and used in raising the paving a little higher. This work, for which a small party was sent here on the above date, was completed on the 23d, and no further work has been done since. One condemned mooring and one mattress barge were sunk in Slough No. 2 behind the crib, and two condemned mooring barges were sunk at the head of Bullerton Tow-Head.

Elmot Chute Dike.—A large field of drift, the accumulation of the past two years, having massed in front of this dike, and the retention of this drift being very desirable as a means to the permanent closure of this chute, and faith in the ability of the dike to hold it long enough to accomplish this result having been considerably weakened by two breaks occurring shortly after its completion, it was decided to sink the drift and thus form a barrier independent of the pile dike. The plan adoptd for this work was to sink a zone of drift about 50 feet wide for the entire length of dike.

Heavy mattresses 50 feet wide by 100 long, distant from the front or upstream row of dike from 10 to 50 feet, were built over the drift and sunk. The length was subsequently increased to 200 feet and more. The mat could be made continuous if desired, in which case frequent detours, to avoid places where the drift is piled up much above the general elevation of the field, will be necessary. For the same reason no regular alignment of the mats is followed, but the most feasible site is followed, care being taken not to get too far away from the dike and to have a close juncture at the ends of the successive mats. As soon as a mat is finished it is heavily ballasted and stone is dumped upon it as long as it continues to sink. The thickness of the drift, as indicated by the submergence of the mats, is from 10 to 18 feet. As soon as enough drift moves over these sunken mats to warrant doing this, a second set will be constructed over the first. The nearest approach to the work is from the dike; hence the barges containing the material are brought up to it, planks laid over or under it, and the material carried or wheeled to the work. When the mats begin to sink a movable raft is used to wheel over and follow up the sinking mat. The mats, having to bear heavy loads, are inade of two layers of brush, one at right angles to the other. Lying in a heavy bottom grillage of 8-foot squares and finished with a similar top grillage forming pockets for the stone, they are well tied and wired together, both transversely and longitudinally. This work began on the 22d of February and is now in progress.

Similar treatment is contemplated for Island No. 30 Dike.

Unloading stone.-The first installment of stone for the season of 1892 having been received on March 3, the work of unloading it was begun in Ashport Bend and is now in progress. The machine used last season for unloading being found too light for the work, unloading is for the present done by wheeling out.

DETAILS OF CONSTRUCTION.

Construction of mattresses.—All river mattresses, except a short one laid to cover a fault in Section A, Fletchers Bend, were 200 feet wide, exclusive of shore connections. Their length varied from 350 to 2,150 feet. They were constructed as described in Appendix D 2, Report of the Chief of Engineers, 1891. At Osceola Bar, as already stated, owing to the difficulties encountered in grading in advance of mattress construction, and for the last 650 feet of the Ashport revetment, connecting mats to bridge the space between the river mat and shore work were used. In all other cases, except repairs to existing works, the shore work was joined to the mat, both classes of work being carried on simultaneously. Whenever the stage of river was above the zero line piles were driven every 100 feet so as to hold the inshore edge of the mat on or close to this line. Steel-wire cables were used for anchorage of the mooring barge and the mattress, and when well taken care of will last with unimpaired strength for years.

Though at first looked upon with disfavor by the foreman and like all innovations severely criticised, these sentiments soon changed when it was seen that when once stretched taut the cables require no attention while in use. The height above the