18

THE MEMPHIS BRIDGE.

adjustable portion of the bridge. It would have been preferable to makeit stiff, but the plan was adopted in conformity to the practice hithertofollowed by the Chief Engineer of the bridge, he not being satisfied withthe detail connections of any stiff system which he had worked out at thetime these plans were prepared.

The transverse bracing is rigid throughout and is made in the formof lattice frames, so as to stiffen, not merely the centre of each post, butintermediate points above. The principle which led the engineer toadopt this form of bracing was his belief that the two trusses of a bridgeshould be made as nearly as possible a single truss. So far as the chordsare concerned this cannot be done; but if the compression members ofthe web are united by stiff frames they become in a large degree a singlecompression member of the full width of the truss, this single compressionmember transferring its strain to two chords at the top and two chords atthe bottom. This principle of transverse bracing was followed through-out the whole superstructure. In the end posts of the intermediate andcentral spans and the posts of the cantilevers which carry the weight ofthe intermediate spans the stiffening was carried further by extendingportal plates down the sides of each separate post, thereby increasing thewidth of the posts and uniting the two sides with a stiff portal. In thecase of the large vertical post over the piers, these portal frames wereunited with the floor beams below, so that the two posts form from topto bottom practically a single member.

The floor system consists of two stringers placed eight feet betweencentres, these stringers being riveted to the webs of the floor beams at eachpanel and half panel point. At the full panel points the floor beams aresupported on short posts to which they are riveted, which posts restdirectly on the pins. At the half panel points the floor beams are sus-pended from the web intersections above.

Although the width between trusses is sufficient for a double trackbridge, the bridge, being, designed but for a single track, has but twolines of stringers. (The management did not feel justified, at the timethe bridge was built, in bearing the additional expense which a doubletrack structure would have entailed.) It should, however, be mentionedthat the Memphis Bridge, though designed for a single track only, is atleast two and one half times as strong as some important single trackbridges built less than fifteen years ago.

, The charter from the General Government required that the bridgeshould be adapted to the passage of vehicles and animals. This provisionis met by making a tight floor 20 feet wide; on each side of the floor are

steel fences supported at the panel points, these fences being in fact lat-tice girders which carry the ends of the ties and so make practically fourlines of stringers. The ties are planked longitudinally with 3-inch pineplank sized to 2f inches, and diagonally with 2-inch oak plank sized toIf inches, this making 4£ inches of solid planking. Between the rails theoak plank is laid longitudinally. It is doubtful whether this expensivehighway floor will be much used.

In many respects the design of the superstructure may be criticisedas not strictly economical. This is admitted; but such criticisms are illconsidered unless they include, not merely the metal in the superstruc-ture, but all the material in the piers and masonry. The substructure ofthe bridge, which under the present design cost more than the superstruc-ture, would have been rendered much more expensive by those changeswhich mere economy of superstructure design called for.

In the design attention was everywhere given to stiffness as well asto strain. This is a matter to which too little attention has been givenand which has often been overlooked in competitive design. It is per-fectly possible to design a structure in which no metal under any ordi-nary supposition will be overstrained, and yet without such overstrainvibrations can exist which would be utterly inadmissible. This mayoccur in trusses of extreme depth and also in structures with cantileverdetails in which loose fitting is permitted at expansion joints.

It was with a view to avoid vibrations as much as possible that stiffchords were adopted throughout, and that the principle was followed ofuniting the compression members of opposite trusses as far as possibleinto single members. For the same reason, the sliding joints at the endsof the cantilevers were so carefully fitted that lost motion in these placespractically does not exist.

While these ideas undoubtedly made the structure more expensivethan a competitive design might have been, the engineer believes that theresults fully justify his work; but in studying details these facts shouldbe remembered.

Special Details. —With these conditions stated, attention is calledto a few of the special details of the superstructure.

1. The stiff chords of the central span, including one full panel ofthe bottom chords of each adjoining cantilever, were made with four webs.This form of construction was adopted for two reasons. First, to reducethe thickness of the metal and so to reduce the lengths of tlie rivets inthe splices. With the present arrangement the maximum length ofrivets is 3f inches between heads, and the larger part of the rivets are in

double shear, the splices being balanced on the two sides of each web,this being the condition under which the rivets are least likely to getloose and least likely to cause trouble if they become loose. The secondobject was to reduce the bending strains on the pins; the connectionswith the web members are all made in the narrow spaces between theoutside and the interior webs, so that the pins, by which the whole hori-zontal strains are transferred to the chords, are supported at four points,and the unsupported length is reduced to a minimum. The objection tohigh theoretical bending strains in pins is not so much the danger to thestructure from the failure of the pins as the fact that by the distortion ofthe pins the strains may be distributed unequally among the several piecesattaching to them. With the arrangement adopted here, this danger isreduced to a minimum.

The four web chords are confined to 10 full (20 half) panels of thetop chord of the central span, to 11 full panels of the bottom chord ofthe same span, two full panels of each cantilever arm and two full panelsof the anchorage arm. Two web chords were used in the remaining bot-tom panels of the cantilever and anchorage arms and throughout thewhole length of the suspended spans. In the bottom chords of the sus-pended spans the rivets became objectionably long.

The sections of the stiff chords were generally made in full panellengths, 56 ft. 5£ in. long, the riveted joints being placed midwaybetween the panel points and over the intermediate points; the only ex-ceptions to this rule were in the upper chords of the intermediate spansand the lower chords of the cantilever arms and anchorage span. Thedetails of the riveted joints are given on Plate 41.

2. The shortness of the anchorage arm insures the connection beingalways in tension and never in compression. The details of the anchorageare given on Plate 37. (Under each truss there are 16 rods in the ma-sonry, these rods connecting with eight blocks which carry the single pinon which four anchor bars connect.) The anchor bars are simply flat barsof steel with pin holes at top and bottom. (After the adjustment of thestructure, folding wedges were placed between the steel plate and theblocks and tightened up so as to prevent any possible horizontal bendingstrain on the anchor rods.) On the centre of the pier was placed a tetra-pod of steel, the four corners of which were anchored to the masonry,and the apex of which fits into a guide at the centre of the end floorbeam, the connection being such that no weight can be thrown upon it,but a full horizontal stiffening is obtained.

The general arrangement of the anchorage is such as to place all