This was the most serious difficulty encountered on any part of the work, and, coming at the very start, was exceedingly discouraging. During the shut-down the broken plates were reinforced temporarily with steel ribs and reinforced concrete (Fig. 1, Plate LXXIII) which, on completion of the work, were replaced by cast-steel segments, as described elsewhere. Practically, no further movement of iron took place, and the loss of grade caused by the settlement of the shield, which was by far the largest that ever occurred in this work, was not sufficient to require a change in the designed grade or alignment of the track. Work was resumed with the shutters in use at the face as an aid to excavation. The features of extreme seriousness did not recur, but for two months the escape of air continued to be extremely large, an average of 15,000 cu. ft. per min. being required on many days during this period.



In Tunnel B, after passing out from under the bulkhead line, in April, 1906, the loss of air became very great, and blow-outs were of almost daily occurrence until the end of June. At the time of the blows the pressure in the tunnel would drop from 2 to 8 lb., and it generally took some hours to raise the pressure to what it was before the blow. During that time regular operations were interrupted. In the latter part of June a permit was obtained allowing the clay blanket to be increased in thickness up to a depth of water of 27 ft. at mean low tide. The additional blanket was deposited during the latter part of June and early in July, and almost entirely stopped the blows.

By the end of the month the natural clay, previously described, formed the greater portion of the face, and, from that time forward, played an important part in reducing the quantity of air required. During April and the early part of May the work was under the ferry racks of the Long Island Railroad. The blanket had to be placed by dumping the clay from wheel-barrows through holes in the decking.

In Tunnel A a bottom heading had been driven 23 ft. in advance of the face at the time work was stopped at the end of 1905. During the ten months of inactivity the seams in the rock above opened. The rock surface was only from 2 to 4 ft. below the top of the cutting edge for a distance of about 60 ft. Over the rock there were large boulders embedded in sharp sand. It was an exceedingly difficult operation to remove the boulders and place the polings without starting a run. The open seams over the bottom heading also frequently caused trouble, as there were numerous slides of rock from the face which broke up the breasting and allowed the soft material from above to run into the shield. There were two runs of from 50 to 75 cu. yd. and many smaller ones.

[Footnote D: The lead of the shield is the angular divergence of its axis from the axis of the tunnel and, in this tunnel, was measured as the offset in 23 ft. It was called + when the shield was pointed upward from grade, and - when pointed downward.]

GUIDING THE SHIELDS.

Little difficulty was experienced at any time in driving the shield close to the desired line, but it was much harder to keep it on grade. In rock section, where the cradle could be set far enough in advance to become hard before the shield was shoved over it, there was no trouble whatever. Where the cradle could be placed only a very short time before it had to take the weight of the shield, the case was quite different. The shield had a tendency to settle at the cutting edge, and when once pointed downward it was extremely difficult to change its direction. It was generally accomplished by embedding railroad rails or heavy oak plank in the cradle on solid foundation. This often had to be repeated several times before it was successful. In soft ground it was much easier to change the direction of the shield, but, owing to the varying nature of the material, it was sometimes impossible to determine in advance how the shield should be pointed. It was found by experience at Manhattan that the iron lining remained in the best position in relation to grade when the underside of the bottom of the shield at the rear end was driven on grade of the bottom of the iron, but if the rate of progress was slow, it was better to drive the shield a little higher.

In the headings from Long Island, which, as a rule, were in soft ground, the cutting edges of the shields were kept from 4 to 8 in. higher, with respect to the grade line, than the rails. The shields would then usually move parallel to the grade line, though this was modified considerably by the way the mucking was done and by the stiffness of the ground at the bottom of the shield.

On the average, the shields were shoved by from ten to twelve of the bottom jacks, with a pressure of about 4,000 lb. per sq. in. The jacks had 9-in. plungers, which made the average total force required to shove the shield 2,800,000 lb. In the soft ground, where shutters were used, all of the twenty-seven jacks were frequently used, and on several occasions the pressure exceeded 6,000 lb. per sq. in. With a unit pressure of 6,000 lb. per sq. in., the total pressure on the shield with all twenty-seven jacks in operation was 5,154 tons.

INJURIES TO SHIELDS.

There were only two instances of damage to the essential structural features of the shields. The most serious was in Tunnel D where the cutting edge at the bottom of the shield was forced up a slightly sloping ledge of rock. A bow was formed in the steel casting which was markedly increased with the next few shoves. Work was suspended, and a heavy cast-steel patch, filling out the bow, was attached to the bent segments, as shown in Fig. 2, Plate LXXIII. No further trouble was experienced with the deformed portion. The other instance was in Tunnel B, from Long Island, where a somewhat similar but less serious accident occurred and was treated in a like manner.

Bulkheads.—At Manhattan, bulkheads had to be built near the shafts before the tunnels could be put under pressure. After 500 ft. of tunnel had been built on each line, the second bulkheads were constructed. The air pressure between the first and second bulkheads was then reduced to between 15 and 20 lb. When the shields had been advanced for 1,500 ft., the third set of bulkheads was built. Nearly all the broken plates which were removed were located between the first and third bulkheads at Manhattan. Before undertaking this operation, the doors of the locks in the No. 3 bulkheads were reversed to take pressure from the west. By this means it was possible to carry on the work of dismantling the shields under comparatively low pressure simultaneously with the removal of the broken plates.

At Long Island City the roofs of the caissons served the purpose of the No. 1 bulkheads. Two other sets of bulkheads were erected, the first about 500 ft. and the second about 1,500 ft. from the shafts.

SETTLEMENT AT SURFACE OF GROUND.

The driving of such portions of the river tunnels, with earth top, as were under the land section, caused a settlement at the surface varying usually from 3 to 6 in. The three-story brick building at No. 412 East 34th Street required extensive repairs. This building stood over the section of part earth and part rock excavation where the tunnels broke out from the Manhattan ledge and where there were a number of runs of sand into the shield. In fact, the voids made by those runs eventually worked up to the surface and caused the pavement of the alley between the buildings to drop 4 or 5 ft. over a considerable area. The tunnels also passed directly under the ferry bridges and racks of the Long Island Railroad at East 34th Street. Tunnels B and D were constantly blowing at the time, and, where progress was slow, caused so much settlement that one of the racks had to be rebuilt. Tunnel A, on the other hand, where progress was rapid, caused practically no settlement in the racks.

CLAY BLANKET.

As previously mentioned, clay was dumped over the tunnels in varying depths at different times. A material was required which would pack into a compact mass and would not readily erode under the influence of the tidal currents of the river and the escape of the great volumes of air which often kept the water in the vicinity of the shields in violent motion. Suitable clay could not be found in the immediate vicinity of the work. Materials from Shooter's Island and from Haverstraw were tried for the purpose. The Government authorities did not approve of the former, and the greater portion of that used came from the latter point. Although a number of different permits governing the work were granted, there were three important ones. The first permit allowed a blanket which roughly followed the profile of the tunnels, with an average thickness of 10 ft. on the Manhattan side and somewhat less on the Long Island City side. The second general permit allowed the blanket to be built up to a plane 27 ft. below low water. This proved effective in checking the tendency to blow, but allowed considerable loss of air. Finally, dumping was allowed over limited and marked areas up to a plane of 20 ft. below low water. Wherever advantage was taken of this last authority, the excessive loss of air was almost entirely stopped. After all the shields had been well advanced out into the river, the blanket behind them was dredged up, and the clay used over again in advance of the shield.

Soundings were taken daily over the shields, and, if marked erosion was found, clay was dumped into the hole. Whenever a serious blow occurred, a scowload of clay was dumped over it as soon as possible and without waiting to make soundings. For the latter purposes a considerable quantity of clay was placed in storage in the Pidgeon Street slip at Long Island City, and one or two bottom-dump scows were kept filled ready for emergencies. Mr. Robert Chalmers, who had charge of the soundings for the contractor, states that "the depressions in the blanket caused by erosion due to the escape of air were, as a rule, roughly circular in plan and of a curved section somewhat flat in the center." Satisfactory soundings were never obtained in the center of a violent blow, but the following instance illustrates in a measure what occurred. Over Tunnel B, at Station 102+80, there was normally 36 ft. of water, 7 ft. of clay blanket, and 20 ft. of natural cover. Air was escaping at the rate of about 10,000 cu. ft. per min., and small blows were occurring once or twice daily. On June 22d, soundings showed 54 ft. of water. A depth of 18 ft. of the river bottom had been eroded in about two days. On the next day there were taken out of the shield boulders which had almost certainly been deposited on the natural river bed. Clay from the blanket also came into the shields on a number of occasions during or after blows. The most notable occasion was in September, 1907, when the top of the shield in Tunnel D was emerging from the east side of Blackwell's Island Reef. The sand in the top was very coarse and loose, and allowed the air to escape very freely. The fall of a piece of loose rock from under the breast precipitated a run of sand which was followed by clay from the blanket, which, in this locality, was largely the softer redredged material. Mucking out the shield was in progress when the soft clay started flowing again and forced its way back into the tunnel for a distance of 20 ft., as shown in Fig. 3, Plate LXXIII. Ten days of careful and arduous work were required to regain control of the face and complete the shove, on account of the heavy pressure of the plastic clay.

The clay blanket was of the utmost importance to the work throughout, and it is difficult to see how the tunnels could have been driven through the soft material on the Manhattan side without it.

The new material used in the blanket amounted to 283,412 cu. yd., of which 117,846 cu. yd. were removed from over the completed tunnels and redeposited in the blanket in advance of the shields. A total of 88,059 cu. yd. of clay was dumped over blows. The total cost of placing and removing the blanket was $304,056.

IRON LINING.

The standard cast-iron tunnel lining was of the usual tube type, 23 ft. in outside diameter. The rings were 30 in. wide, and were composed of eleven segments and a key. The webs of the segments were 1-1/2 in. thick in the central portion, increasing to 2-3/8 in. at the roots of the flanges, which were 11 in. deep, 2-1/4 in. thick at the root, and 1-1/2 in. at the edge, and were machined on all contact faces. Recesses were cast in the edge of the flanges, forming a groove, when the lining was in place, 1-1/2 in. deep and about 3/8 in. wide, to receive the caulking. The bolt holes were cored in the flanges, and the bosses facing the holes were not machined. The customary grout hole was tapped in the center of each plate for a standard 1-1/4-in. pipe. In this work, experience indicated that the standard pipe thread was too fine, and that the taper was objectionable. Each segment weighed, approximately, 2,020 lb., and the key weighed 520 lb., the total weight being 9,102 lb. per lin. ft. of tunnel. Fig. 1 shows the details of the standard heavy lining.

In addition to the standard cast-iron lining, cast-steel rings of the same dimensions were provided for use in a short stretch of the tunnel, when passing from a rock to a soft ground foundation, where it was anticipated that unequal settlement and consequent distortion and increase in stress might occur, but, aside from the small regular drop of the lining as it passed out of the tail of the shield, no such settlement was observed.

Two classes of lighter iron, one with 1-in. web and 8-in. flanges and the other with 1-1/4-in. web and 9-in. flanges—the former weighing 5,166 lb. per lin. ft. of tunnel and the latter, 6,776 lb.—were provided for use in the land sections between East Avenue and the Long Island City shafts. Two weights of extra heavy segments for use at the bottom of the rings were also furnished. The so-called XX plates had webs and flanges 1/4 in. thicker than the standard segment and the YY plates were similarly 1/2 in. heavier. The conditions under which they were used will be referred to later. All the castings were of the same general type as shown by Fig. 1.

Rings tapering 3/4 in. and 1-1/2 in. in width were used for changes in alignment and grade, the former being used approximately at every fourth ring on the 1 deg. 30' curves. The 1-1/2-in. tapers were largely used for changes in grade where it was desired to free the iron from binding on the tail of the shield. Still wider tapers would have been advantageous for quick results in this respect.

No lug was cast on the segments for attachment to the erector, but in its place the gadget shown on Fig. 4, Plate LXX, was inserted in one of the pairs of bolt holes near the center of the plate, and was held in position by the running nut at one end.

In the beginning it was expected that the natural shape of the rings would not show more than 1 in. of shortening of the vertical diameter; this was slightly exceeded, however, the average distortion throughout the tunnels being 1-7/16 in. The erectors were attached to the shield and in such a position that they were in the plane of the center of the ring to be erected when the shove was made without lead and just far enough to permit placing the segments. If the shield were shoved too far, a rare occurrence, the erection was inconvenienced. In driving with high vertical leads, which occurred more frequently, the disadvantage of placing the erector on the shield was more apparent. Under such conditions the plane of the erector's motion was acutely inclined to the plane of the ring, and, after placing the lower portion of the ring, it was usually necessary to shove the shield a few inches farther in order to place the upper plates. The practical effect of this action is referred to later.



At first the erection of the iron in the river tunnels interfered somewhat with the mucking operations, but the length of time required to complete the latter was ample for the completion of the former; and the starting of a shove was seldom postponed by reason of the non-completion of a ring. After the removal of the bottom of the diaphragms, permitting the muck cars to be run into the shield and beyond, the two operations were carried on simultaneously without serious interference. The installation of the belt conveyor for handling the soft ground spoil in Tunnel A was of special benefit in this respect.

Preparatory to the final bolt tightening of each ring as erected, a 15-ton draw-jack, consisting of a small pulling-jack inserted in a light eye-bar chain, was placed on the horizontal diameter, and frequently the erectors were also used to boost the crown of the iron, the object being to erect the ring truly circular. Before shoving, a 1-1/4-in. turn-buckle was also placed on the horizontal diameter in order to prevent the spreading of the iron, previous to filling the void outside with grout. The approach of the supports for the upper floor of the trailing platform necessitated the removal of these turnbuckles from all but the three leading rings, but if the iron showed a tendency to continue distortion, they were re-inserted after the passage of the trailing platform and remained until the arch of the concrete lining was placed.

The cost of handling and erecting the iron varied greatly at different times, averaging, for the river tunnels, $3.32 per ton for the directly chargeable labor of handling and erecting, to which must be added $7.54 for "top charges." The cost of repairing broken plates is included in this figure.

Broken Plates.—During the construction of the river section of the tunnels, a number of segments were found to have been broken while shoving the shield. The breaks, which with few exceptions were confined to the three or four bottom plates, almost invariably occurred on the advanced face of the ring, and rarely extended beyond the bottom of the flange. A careful study of the breaks and of the shoving records disclosed several distinct types of fracture and three principal known causes of breakage by the shield.

In the first case, the accidental intrusion of foreign material between the jack head and the iron caused the jack to take its bearings on the flange above its normal position opposite the web of the ring, and resulted usually in the breaking out of a piece of the flange or in several radiating cracks with or without a depression of the flange. These breaks were very characteristic, and the cause was readily recognizable, even though the intruding substance was not actually observed.

In the second case, the working of a hard piece of metal, such as a small tool, into the annular space between the iron and the tail of the shield, where it was caught on the bead and dragged along as the shield advanced, was the known cause of a number of broken segments. Such breaks had no particular characteristic, but were usually close above the line of travel of the lost tool or metal. Their cause was determined by the finding of a heavy score on the underside of the segment or the discovery of the tool wedged in the tail of the shield or lying under the broken plate when it was removed. It is probable that a number of breaks ascribed to unknown causes should be placed in this class.

The third cause includes the largest number of breaks, and, while difficult to define closely, is the most interesting. Broadly speaking, the breaks resulted from the movements of the shield in relation to the position of the tunnel lining. While shoving through soft ground, it was frequently difficult to apply sufficient power to the lower jacks to complete the full shove of 30 in. on the desired alignment. The shield, therefore, was driven upward at the beginning of the shove, and, as the sand packed in front of the shield and more power was required, it was furnished by applying the upper jacks. The top of the shield was slowly pushed over, and, at the close of the shove, the desired position had been obtained; but the shield had been given a rocking motion with a decided lifting of the tail toward the close of the shove. A similar lifting of the tail occurred when, with high vertical leads, the top of the shield was pushed over in order to place the upper plates of the ring. Again, when the shield was driven above grade and it was desired to descend, the passage of the shield over the summit produced a like effect. In all these movements, with the space between the tail of the shield and the iron packed tight with pugging, the upward thrust of the shield tended to flatten the iron in the bottom and occasional broken plates were the result. The free use of the taper rings, placed so as to relieve the binding of the lining on the tail of the shield, forces the tunnel to follow the variations in the grade of the shield, but reduces greatly the injuries to the rings from this action.

In Tunnel D, where very high vertical leads were required through the soft sand, combined with a marked tendency of the shield to settle, the shield was badly cramped on the iron and dragged along it at the top. The bearing of the iron on its soft foundation tended to thrust up the bottom in this case also, as shown by the opening of the bottom cross-joints when the bolts were slackened to relieve the strain during a shove. The anticipated cracks in the crown plates, which have been more frequently observed in other tunnels, did not occur here, and were not found elsewhere except in one place in Tunnel B where they were traced to a similar action of the shield. The cracks resulting from the movements of the shield, as briefly described above, in this third case were not confined to any particular type, but occurred more frequently at the extreme end of the circumferential flange than at any other point.

The number of broken plates occurring in the river tunnels was 319, or 0.42% of the total number erected. Of these, 52 were found and removed, either before or immediately after a shove, by far the greater number being broken in handling before or during erection. The remaining 267 are considered below.

Repair of Broken Plates.—On the completion of a shove, the tail of the shield lacked about 5 in. of covering the full width of the last ring, and the removal of a plate broken during the shove, therefore, would have exposed the ground at the tail of the shield. With a firm material in the bottom, this introduced no particular difficulties, and, under such conditions, a broken plate was usually removed at once. In the sand, however, and especially on the Manhattan side where it was quick and flowing, the removal of a plate was attended with some danger, and such plates were usually left to be removed on the completion of the tunnel. Many of these had been reinforced by the use of XX, YY, and steel segments placed adjacent to the break in the following rings.

After the meeting of the shields, the postponed replacement of the broken segments was taken up. The pressure was raised sufficiently to dry thoroughly the sand outside the segments, which were drilled and broken out usually in quarters as shown on Fig. 1, Plate LXXIII. A steel segment was then inserted in the ring and drawn into place by turnbuckles. The application of the draw-jack, with a pull of about 30 tons to each end successively, brought the plate to a firm bearing on the radial joints at the ends.

Where the broken plate was isolated and was reinforced by steel or extra heavy segments in the adjacent ring, the crack, if slight, was simply caulked to insure water-tightness. If, however, the crack was opened or extended to the web of the plate, the cross-flanges were tied together by a 1-1/2-in. by 7-ft. bolt, inserted through the bolt holes nearest the broken flange. The long bolt acted in the nature of a bow string, and was provided at its ends with two nuts set on opposite sides of the cross-joints to replace the standard bolts removed for its insertion. Fig. 4, Plate LXXIII shows one of these bolts in place. In addition, all broken plates remaining in the tunnel were reinforced with 1-in. twisted-steel rods in the concrete lining, also shown in Fig. 4, Plate LXXIII.

Special Construction at River Shield Junctions.—Dismantling the shields was started as soon as they came to rest in their final position with the cutting edges together. The plans contemplated their entire removal, with the exception of the cylindrical skins and cast-steel cutting edges. Inside the former the standard tunnel lining was erected to within 4 ft. of the heels of the cutting edges. Spanning the latter, and forming the continuous metal tunnel lining, the special construction shown by Fig. 2 was built. This consisted of a 1-1/4 in. rolled-steel ring, 7 ft. long, erected inside the cutting edges, with an annular clearance of 1 in., and two special cast-iron rings shaped to connect the rolled-steel ring with the normal lining. One flange of the special cast-iron rings was of the standard type, the other was returned 9 in. in the form of a ring, the inside diameter of which was the same as the outside diameter of the rolled-steel ring to which it was bolted.

The space between the standard and special construction was of varying width at the various shields, and was filled with a closure ring cast to the lengths determined in the field. Fig. 2 shows the completed construction.

Hook-bolts, screwed through threaded holes and buried in 1 to 1 Portland cement grout ejected through similar holes, reinforced the rolled-steel ring against external water pressure. In two of the tunnels the concrete lining was carried completely through the junction, and covered the whole construction, while in the remaining two tunnels it was omitted at the rolled-steel ring, leaving the latter exposed and set back about 3 in. from the face of the concrete.



GROUTING.

Except as previously noted, the voids outside of the tunnel lining were filled with grout ejected through the grout holes in each segment. The possibility was always present that Portland cement, if used for grout in the shield-driven tunnels, would flow forward around the shield and set hard, "freezing" the shield to the rock or the iron lining, or at least forming excrescences upon it, which would render its control difficult. With this in mind, the contractors proposed to substitute an English Blue Lias lime as a grouting material. Grout of fresh English lime containing a moderate quantity of water set very rapidly in air to the consistency of chalk. Its hydraulic properties, however, were feeble, and in the presence of an excess of water it remained at the consistency of soft mud. It was not suitable, therefore, as a supporting material for the tunnel.

An American lime, made in imitation of the Lias lime, but having greater hydraulic properties, was tried, but proved unsatisfactory. Two brands of natural cement were also tried and rejected, but a modified quick-setting natural cement, manufactured especially for this work, was eventually made satisfactory, and by far the largest part of the river-tunnel grouting was done with this material mixed 1 to 1 by volume. East of the Long Island shafts the work which was built without shields was grouted principally with Portland cement and sand mixed 1 to 1 by volume.

In the river tunnels large quantities of the English lime were used neat as grout over the top of the tunnel in attempts to stop losses of air through the soft ground. It was not of great efficiency, however, in this respect until the voids outside of the lining had been filled above the crown. Its properties of swelling and quick setting in the dry sand at that point then became of value. The use of dry lime in the face, where the escaping air would carry it into the voids of the sand and choke them, was much more promptly efficacious in checking the loss.

With the exception of the English lime, all grout was mixed 1 to 1 with sand in a Cockburn continuous-stirring machine operated by a 3-cylinder air engine. The grout machine was placed on the lower floor of the trailing platform shown on Plate LXXII, while the materials were placed on the upper platform, and, together with the water, were fed into the machine through a hole in the upper floor. The sand was bagged in the yard, and the cars on which the materials were sent into the tunnels were lifted by an elevator to the level of the upper floor of the trailing platform before unloading.

Great difficulty was experienced in preventing the waste of the fluid grout ahead of the shield and into the tail through the space between it and the iron lining. In a full soft ground section, the first condition did not usually arise. In the full-rock sections the most efficient method of checking the waste was found to be the construction of dams or bulkheads outside the lining between it and the rock surface. For this purpose, at intervals of about 30 ft., the leading ring and the upper half of the preceding one were disconnected and pulled forward sufficiently to give access to the exterior. A rough dam of rubble, or bags of mortar or clay, was then constructed outside the iron, and the rings were shoved back and connected up. In sections containing both rock and soft ground, grout dams were built at the cutting edge at intervals, and were carried up as high as circumstances permitted.

The annular space at the tail of the shield was at all times supposed to be packed tight with clay and empty bags, but the pugging was difficult to maintain against the pressure of the grout. For a time, 1/2-in. segmental steel plates, slipped down between the jackets and the iron, were used to retain the pugging, but their displacement resulted in a number of broken flanges, and their use was abandoned. In their place, 2-in. segmental plates attached to the jack heads were substituted with more satisfactory results. Notwithstanding these devices, the waste of grout at the tail was very great.

The soft ground material on various portions of the work acted very differently. The clay and "bull's liver" did not cave in upon the iron lining for several hours after the shield had passed, sometimes not for a day or more, which permitted the space between it and the iron to be grouted. The fine gray or beach sand and the quicksand closed in almost at once. The quicksand has a tendency to fill in under the iron from the sides and in places to leave a cavity at about the horizontal diameter which was not filled from above, as the sand, being dried out by the air, stood up fairly well and did not cave against the iron, except where nearly horizontal at the top.

The total quantity of grout used on the work was equivalent in set volume to 249,647 bbl. of 1 to 1 Portland cement grout, of which 233,647 bbl. were ejected through the iron lining, an average of 14.93 bbl. per lin. ft. The cost of grout ejected outside of the river tunnels was 93 cents per bbl. for labor and $2.77 for "top charges." East of the Long Island shaft the corresponding costs were $0.68 and $1.63, the difference being partly due to the large percentages of work done in the normal air at the latter place.

CAULKING AND LEAKAGE.

Up to August, 1907, the joints between the segments of the cast-iron lining were caulked with iron filings and sal ammoniac, mixed in the proportion of 400 to 1 by weight. With the air pressure balancing the hydrostatic head near the tunnel axis, it was difficult to make the rust-joint caulking tight below the axis against the opposing water pressure; this form of caulking was also injured in many places by water dripping from service pipes attached to the tunnel lining. A few trials of lead wire caulked cold gave such satisfactory results that it was adopted as a substitute. Pneumatic hammers were used successfully on the lead caulking, but were only used to a small extent on the rust borings, which were mostly hand caulked. Immediately before placing the concrete lining, all leaks, whether in the rust borings or lead, were repaired with lead, and the remainder of the groove was filled with 1 to 1 Portland cement mortar, leaving the joints absolutely water-tight at that time. The subsequent development of small seepages through the concrete would seem to indicate that the repair work should have been carried on far enough in advance of the concreting to permit the detection of secondary leaks which might develop slowly. The average labor cost chargeable against the caulking was 12 cents per lin. ft., to which should be added 21.8 cents for "top charges."

Unfortunately, it was necessary to place the greater part of the concrete lining in the river tunnels during the summer months when the temperature at the point of work frequently exceeded 85 deg.; and the temperature of the concrete while setting was much higher. This abnormal heat, due to chemical action in the cement, soon passed away, and, with the approach of winter, the contraction of the concrete resulted in transverse cracks. By the middle of the winter these had developed quite uniformly at the ends of each 30-ft. section of concrete arch as placed, and frequently finer cracks showed at about the center of each 30-ft. section.

While the temperature of the concrete was falling, a like change was taking place in the cast-iron lining, with resulting contraction. The lining had been erected in compressed air, the temperature of which averaged about 70 deg. in winter and higher in summer. Compressed air having been taken off in the summer of 1908, the tunnels then acquired the lower temperature of the surrounding earth, slowly falling until mid-winter. The contraction of the concrete, firmly bedded around the flanges of the iron, and showing cracks at fairly uniform intervals, probably localized the small corresponding movements of the iron near the concrete cracks, and resulted in a loosening of the caulking at these points. With the advent of cold weather, damp spots appeared in numerous places on the concrete, and small seepages showed through quite regularly at the temperature cracks, in some cases developing sufficiently to be called leaks. Only a few, however, were measurable in amount.

Early in January small brass plugs were firmly set on opposite sides of a large number of cracks, and caliper readings and air temperature observations were taken regularly throughout the winter and spring. The widths of the cracks and the amount of leakage at them increased with each drop in temperature and decreased as the temperature rose again, but until spring the width of the cracks did not return to the same point with each return of temperature.

The leakage was similar in all four tunnels, but was largest in amount in Tunnel D, where, at the beginning of February, the ordinary flow was about 0.0097 cu. ft. per sec., equivalent to 0.00000347 cu. ft. per sec. per lin. ft. of tunnel. Of this amount 0.0065 cu. ft. per sec. could be accounted for at eight of the cracks showing measurable leakage, leaving 0.0032 cu. ft. per sec. or 0.00000081 cu. ft. per sec. per lin. ft. of tunnel to be accounted for as general seepage distributed over the whole length.

It was not feasible to stop every leak in the tunnel, most of which were indicated simply by damp spots on the concrete; a rather simple method was devised, however, for stopping the leaks at the eight or ten places in each tunnel where water dripped from the arch or flowed down the face of the concrete. The worst leak in any tunnel flowed about 0.0023 cu. ft. per sec. To stop these leaks, rows of 1-in. holes, at about 4-in. centers, were drilled with jap drills through the concrete to the flange of the iron. These rows were from 3 to 18 ft. long, extending 1 ft. or more beyond the limits of the leak. The bottoms of the holes were directly on the caulking groove and the pounding of the drill usually drove the caulking back, so that the leak became dry or nearly so after the holes were drilled. If left alone the leaks would gradually break out again in a few hours or a few days and flow more water than before. They were allowed to do this, however, in only a few cases as experiments. After the holes were drilled, the bottom 4 in. next the flange was filled with soft neat cement mortar. Immediately on top of this was placed two plugs of neat cement about 2-1/2 in. long, which were 5 or 6 hours old and rather hard. Each was tamped in with a round caulking tool of the size of the hole driven with a sledge hammer. On top of this were driven in the same way two more plugs of neat cement of the same size, which were hard set. These broke up under the blows of the hammer, and caulked the hole tight. When finished, the tamping tool would ring as though it was in solid rock. Great pressure was exerted on the plastic mortar in the bottom of the hole, which resulted in the re-caulking of the joint of the iron. No further measurable leakage developed in the repaired cracks, during a period of four months, and the total leakage has been reduced to about 0.002 cu. ft. per sec. in each tunnel, an average of 0.00000051 cu. ft. per sec. per lin. ft.

SUMP AND PUMP CHAMBERS.

To take care of the drainage of the tunnels, a sump with a pump chamber above it was provided for each pair of tunnels. The sumps were really short tunnels underneath the main ones and extending approximately between the center lines of the latter. They were 10 ft. 9-1/2 in. in outside diameter and 44 ft. long. The water drops directly from the drains in the center lines of the tunnels into the sumps. Above the sumps and between the tunnels, a pump chamber 19 ft. 5 in. long was built. Above the end of the latter, opposite the sump, a cross-passage was constructed between the bench walls of the two tunnels. This passage gives access from either tunnel through an opening in the floor to the pump chamber and through the latter to the sump.

From the preliminary borings it was thought that the sumps were located so that the entire construction would be in rock. This proved to be the case on Tunnels C and D, but not on Tunnels A and B. The position of the rock surface in the latter is shown by Fig. 3. After the excavation was completed in Tunnel B, January 1st, 1908, the plates were removed from the side of the tunnel at the cross-passage, and a drift was driven through the earth above the rock surface across to the lining of Tunnel A. The heading was timbered as shown by Fig. 3. There was practically no loss of air from the drift, but the clay blanket had been removed from over this locality and the situation caused some anxiety. In order to make the heading as secure as possible, the 24-in. I-beams, shown on Fig. 3, were attached to the lining of the two tunnels. The beams formed a support for the permanent concrete roof arch of the passage, which was placed at once. At the same time plates were removed from the bottom in Tunnel B over the site of the sump, and a heading was started on the line of the sump toward Tunnel A. As soon as the heading had been driven beyond the center line of the pump chamber, a bottom heading was driven from a break-up westward in the pump chamber and a connection was made with the cross-passage. The iron lining of the pump chamber was next placed, from the cross-passage eastward. The soft ground was excavated directly in advance of the lining, and the ground was supported by polings in much the same manner as described for shield work. On account of bad ground and seams of sand encountered in the rock below the level of the cross-beams, the entire west wall of the pump chamber was placed before enlarging the sump to full size. This was also judicious, in order to support as far as possible the iron lining of the tunnels. The sump was then excavated to full size. The iron lining of the sump and the east wall of the pump chamber were placed as soon as possible. The voids outside the iron lining of the sump and the pump chamber were filled as completely as possible with concrete, and then thoroughly grouted. Finally, the concrete lining was put in place inside of the iron.

As shown by Fig. 3, the excavation of these chambers left a considerable portion of the iron lining of the tunnels temporarily unsupported on the lower inner quarter. To guard against distortion, a system of diagonals and struts was placed as shown.

The floor of the pump chamber was water-proofed with felt and pitch in a manner similar to that described for the caissons at Long Island City. It was not possible to make the felt stick to the vertical walls with soft pitch, which was the only kind that could be used in compressed air, and, therefore, the surfaces were water-proofed by a wall of asphalt brick laid in pitch melting at 60 deg. Fahr. Forms were erected on the neat line, and the space to the rock was filled with concrete making a so-called sand-wall similar to that commonly used for water-proofing with felt and pitch. The bricks were then laid to a height of four or five courses. The joints were filled with pitch instead of mortar. Sheets of tin were then placed against the face of the wall and braced from the concrete forms. As much pitch as possible was then slushed between the brick and the sand-wall, after which the concrete in the main wall was filled up to the top of the water-proofing course. The tin was then withdrawn and the operation repeated. This method was slow and expensive, but gave good results. Ordinary pitch could not be used on account of the fumes, which are particularly objectionable in compressed air. The 60 deg. pitch was slightly heated in the open air before using.



The sump and pump chamber on Tunnels C and D differed from the one described only in minor details; but, being wholly constructed in rock, presented fewer difficulties and permitted a complete envelope of water-proofing to be placed in the top.

CONCRETE LINING.

The placing of concrete inside the iron tube was done by an organization entirely separate from the tunneling force. A mixing plant was placed in each of the five shafts. The stone and sand bins discharged directly into mixers below, which, in turn, discharged into steel side-dump concrete cars. All concrete was placed in normal air.

The first step, after the iron lining was scraped clean and washed down and all leaks were stopped, was the placing of biats, marked B on Plate LXXIV. These were made up of a 6 by 12-in. yellow pine timber, 17 ft. long, with two short lengths of the same size spliced to its ends by pieces of 12-in. channels, 3 ft. 9 in. long, clamped upon the sides. These biats were placed every 5 ft. along the tunnel in rings having side keys. Next, a floor, 13 ft. wide, was laid on the biats and two tracks, of 30-in. gauge and 6-1/2-ft. centers, were laid upon the floor. There were three stages in the concreting. Fig. 2, Plate LXXIV, shows the concrete in place at the end of the first, and Fig. 3, Plate LXXIV, at the end of the second stage. The complete arch above the bench walls was done in the last operation.

Two 3 by 10-in. soldiers (SS in Figs. 1 and 2, Plate LXXIV) were fastened to each biat and braced across by two horizontal and two diagonal braces. To each pair of soldiers a floor template, T, was then nailed. The form for the center drain was then suspended as shown in Fig. 1, Plate LXXIV. Three pieces of shuttering, FFF, 20 ft. long, were then nailed to the bottom of the soldiers. One is all that would have been needed for the first concrete placed, but it was easier to place them at this stage than later, when there was less room. Three rough shutters were also nailed to the curved portion for the floor template. Opposite each biat, a bracket, bb, was then nailed, which carries a set of rough boards which formed the risers for the duct steps. Everything was then ready for concreting except that, where refuge niches occurred, a form for the portion of the niche below the seat was nailed to the shuttering. This form is shown at R in Fig. 1, Plate LXXIV.



The concrete was dumped down on each side from side-dump cars standing on the track, and, falling between the risers for the duct steps, ran or was shoveled under the forms and down into the bottom. The horizontal surface on each side the center drain was smoothed off with a shovel. The workmen became very skillful at this, and got a fairly smooth surface. This concrete was usually placed in lengths of 45 or 60 ft. After setting for about 24 hours, the brackets, bb, were removed, together with the shuttering on the steps. The triangular pieces, t in Fig. 1, Plate LXXIV, were not removed until later. Instead, a board was laid upon this lower step on which the duct layers could work. This and the triangular piece were not removed until just before the bench concrete was placed. This was important, as otherwise the bond between the old and new concrete would be much impaired by dirt ground into the surface of the old concrete. The ducts were then laid, as shown in Fig. 2, Plate LXXIV.

The remaining shutters for the face of the bench walls were then placed. The remainder of the forms for the refuge niches, RR, in Fig. 1, Plate LXXIV, were nailed to the shutters, the steel beam over the niche was laid in place, the forms for the ladders, L in Fig. 2, Plate LXXIV, which occur every 25 ft., were tacked to the shutters, the shutters and forms were given a coat of creosote oil, and then all was ready for placing the bench concrete.

The specifications required a 2-in. mortar face to be placed on all exposed surfaces and the remainder to be smoothed with a trowel and straight-edge. After about 48 hours, the biats were blocked up on the bench, and all forms between the bench walls below the working floor were removed.

The centering for the arch concrete consisted of simple 5 by 3-1/2 by 5/16-in. steel-angle arch ribs, curved to the proper radius, spaced at 5-ft. intervals. Each rib was made up of two pieces spliced together at the top. Two men easily handled one of these pieces. After splicing, the rib was supported by four hanger-bolts fastened to the iron lining as shown in Fig. 3, Plate LXXIV.

In the early part of the work, two additional bolts were used about half way up on the side between the upper and lower hanger-bolts. It was soon found that by placing the strut between the tunnel lining and the crown of the rib, these hanger-bolts could be dispensed with. The lagging was of 3-in. dressed yellow pine, 12 in. wide, and in 15-ft. lengths. Each piece had three saw cuts on the back, from end to end, allowing it to be bent to the curve of the arch; it was kept curved by an iron strap screwed to the back. The arches were put in, either in 15, 30 or 45-ft. lengths, depending on what was ready for concrete and what could be done in one continuous working. The rule was that when an arch was begun, the work must not stop until it was finished. An arch length always ended in the middle of a ring. The lagging was placed to a height of about 6 ft. above the bench before any concreting was done. When the concrete had been brought up to that point, lagging was added, one piece at a time, just ahead of the concrete, up to the crown, where a space of about 18 in. was left. When the lagging had reached the upper hanger-bolts, they were removed, which left only the two bottom bolts fixed in the concrete. Most of these were unscrewed from the eye and saved, as tin sleeves were placed around them before concreting. Two cast-iron eyes were lost for every 5 ft. of tunnel. To place the key concrete, a stage was set up in the middle of the floor, and, beginning at one end, about 2 ft. of block lagging was placed. Over this, concrete was packed, filling the key as completely as possible. This was done partly by shoveling and using a short rammer, and partly by packing with the hands by the workmen, who wore rubber gloves for the purpose. Another 2 ft. of lagging was then placed, and the operation was repeated, and thus working backward, foot by foot, the key was completed. This is the usual way of keying a concrete arch, but in this case the difficulty was increased by the flanges of the iron lining. It was practically impossible to fill all parts of the pockets formed by these flanges. To meet this difficulty, provision was made for grouting any unfilled space. As the concrete was being put in, tin pipes were placed with their tops nearly touching the iron lining, and their bottoms resting on the lagging. Each pocket was intended to have two of these pipes, one to grout through and the other to act as a vent for the escape of air. Each center key ring had six pipes, and each side key had eight. The bottoms of the pipes were held by a single nail driven half way into the lagging. This served to keep the pipes in position and to locate them after the lagging was taken down.

The cost of labor in the tunnels directly chargeable to concrete was $1.80 per cu. yd. The top charges, exclusive of the cost of materials (cement, sand, and stone), amounted to $3.92.

ELECTRIC CONDUITS.

In one bench wall of each tunnel there were fifteen openings for power cables and in the other, between the river shafts, there were forty openings for telephone, telegraph, and signal cables. East of the Long Island shaft, the number of the latter was reduced to twenty-four. The telephone ducts were all of the four-way type. The specifications required that the power ducts should have an opening of not less than 3-1/2 in., nor more than 3-7/8 in., and that after laying they should pass a 4-ft. mandrel, 3-3/8 in. at the leading end and 2-5/8 in. at the other. The outside dimension was limited between 5 and 5-3/8 in. The openings of the four-way ducts were required to be not less than 3-3/8 in., nor more than 3-5/8 in., and after laying to pass a 5-ft. mandrel, 3-1/4 in. at the leading end and 2-1/2 in. at the other. The outside dimensions were limited between 9 and 9-1/2 in. All were to be laid in 1/4-in. beds of mortar. The specifications were not definite as to the shape of the opening, but those used were square with corners rounded to a radius of 3/8 in. The four-ways were 3 ft. long, and the singles, 18 in.

A study of the foregoing dimensions will show that the working limits were narrow. Such narrow limits would not pay for the ordinary conduit line in a street, where there is more room. In the tunnel greater liberality meant either reducing the number of conduits or encroaching on the strength of the concrete tunnel lining. The small difference of only 1/8 in. in the size of the mandrel, or a clearance of only 1/16 in. on each side, no doubt did increase the cost of laying somewhat, though not as much as might at first be supposed. All bottom courses were laid to a string, in practically perfect line and grade, and all joints were tested with mandrels which were in all openings, and pulled forward as each piece of conduit was laid. As the workmen became skillful, the progress was excellent.

All costs of labor in the tunnel chargeable to duct laying amounted to $0.039 per ft. of duct; top charges brought this up to $0.083.

The serious problem was to guard against grout and mortar running into the duct opening through the joints from the concrete, which was a rather wet mixture. Each joint was wrapped, when laid, with canvas, weighing 10 oz. per sq. yd., dipped in cement grout immediately before using. These wraps were 6 in. wide, and were cut long enough to go around the lap about the middle of the duct. As soon as all the ducts were laid, the entire bank was plastered over with fairly stiff mortar, which, when properly done, closed all openings. The plastering was not required by the specifications, but was found by the contractor to result in a saving in ultimate cost.

The concrete on the two sides of the bank of ducts was bonded together by 2 by 1/8-in. steel bonds between the ducts, laid across in horizontal joints. Both ends were split into two pieces, 1 in. long, one of which was turned up and the other down. These bonds projected 1-1/2 in. into the concrete on either side. Where the bond came opposite the risers of the duct step, against which the ducts were laid, recesses were provided for the projecting bond. This was done by nailing to the rough shutters for the steps a form which when removed left a dove-tailed vertical groove. This form was made in two pieces, one tapering inward and the other with more taper outward. As the bonds were placed, these grooves were filled with mortar.

The ducts usually received their final rodding with the specification mandrel a month or more after they were laid, after which all openings into splicing chambers were stopped by wooden plugs, 8 in. long tapering from 3-3/4 in. at one end to 2-3/4 in. at the other end, and shaped to fit the opening tightly. At first the plugs were paraffined, to keep them from swelling and breaking the ducts, but were not successful, as the paraffin lubricated them so that they would not stay in place. They were expensive, and there was some swelling in the best that were obtained. A better plug was made by using no paraffin, but by making six saw cuts, three horizontal and three vertical, in the larger end, cutting to within about 2 in. of the smaller end. The swelling of the wood was then taken up by the saw cuts and the spring of the wood.

The splicing chambers are at 400-ft. intervals. They are 6 ft. long, 4 ft. 9 in. high, with a width varying from 3 ft. 2 in. at the top to 1 ft. 2 in. at the bottom.

THE END