919 per cent. loss of strength only, allowing nothing for friction between the surfaces (20). But friction in such a joint would certainly add 9 per cent. or even more if it could be utilised, and thus we find that the full strength of the solid plate becomes available, there being no loss whatever.

The amount of lap in such a joint carried out in the ordinary manner would be very great, but this may be avoided by the arrangement shown by Fig. 15. The bottom plate of the girder, or rather the plate subjected to tensile strain, instead of being made in one thickness, is divided into two plates of half the thickness, and they are arranged upon one another so as to break-joint. Thus when a thickness of 1 inch is required we should use two-inch plates: then if the top plate extends from A to B, the lower plate would extend from C to D, the junction at C being in the centre of the solid part of the plate A, B, &c. We thus secure all the advantages of spreading the rivets, without any loss by lap.

The only drawback to this method is, that two thin plates will be more subject to damage from rust than one thick one of equal area, not only because they would expose double surface to the elements, but also that the interstice between them would harbour the rain-water. This method is therefore most useful in large structures where thick plates are used, and even then, care should be taken by painting, &c., to obviate deterioration by rusting.

STEEL RIVETED-JOINTS FOR GIRDER-WORK.

(37.) The introduction of the Bessemer process in the manufacture of steel, and consequent reduction in cost, has led to its extensive use for all purposes as a substitute for wrought iron. The full value of its great tensile strength has not been quite realised with riveted joints, from the fact that steel rivets have comparatively a low shearing strength, which differs very little from that of wrought iron (42). This has led to the necessity for larger rivets than would otherwise have been required, resulting in some loss of strength. The best experiments we have are those of Mr. H. Sharp, from which we shall obtain the

strength of annealed and unannealed steel plates, solid, punched, and drilled; also the strength of the metal in riveted joints, and of the rivets in those joints.

(38.) "Solid or Unpunched Plates."-Table 9 gives the strength of solid steel plates, and shows the remarkable effect of annealing or heating to a dull red heat and cooling slowly in sand or ashes, the result being an increase in strength of 51 per cent. crossways of the grain, and 60 per cent. lengthways, the mean of the two 55 per cent.

=

Remarkable as this result is, it is confirmed by the experiments of Mr. Barnaby at H.M. Dockyard, Chatham, on steel plates inch thick, punched with holes about inch diameter, &c., as in Fig. 14: the average of eight annealed plates was 32.839, and of eight unannealed plates 21.097 tons per square inch, showing a difference of 38 839 21.097 1.5566, or 55.66 per cent., being almost exactly the same as with solid plates.

=

TABLE 9.-Of Experiments on the TENSILE STRENGTH of SOLID STEEL PLATES.

(39.) This is the more remarkable because the effect of annealing with wrought-iron plates was just the reverse, as shown by the direct experiments of Kirkaldy (14) on six kinds of Yorkshire iron; the mean result of eighteen experiments on

plates,, and inch thick being a loss by annealing of 5.6 per cent. lengthways, and 5.2 per cent. crossways. With Lowmoor iron, the loss was 4.8 and 1.8 per cent., and with Bowling, 8.0 and 9.1 per cent. respectively.

It would appear from this that steel plates, even in the solid or unpunched form should always be annealed With annealed plates, those strained lengthways of the grain are 10 per cent. stronger than those strained crossways, and with those not annealed, 4.2 per cent.

(40.) "Effect of Punching and Drilling."-Mr. Sharp made experiments on steel plates by punching and drilling rivet-holes of the diameter and pitch commonly used for riveted joints, the results of which are given by Table 10, which shows that by punching cold in the usual way, the metal left between the holes is damaged from 24.1 to 38 per cent., the mean being 33 per cent., which is very great as compared with wrought iron. Mr. Kirkaldy's experiments on Yorkshire iron in Table 4 gives the loss due to punching from 6.7 to 21.2 per cent., the mean of the whole being 15 per cent. only; Mr. Fairbairn's experiments on Lowmoor iron in single-riveted joints gave 1·0–·76 = ·24, or 24 per cent. loss, by col. 4 of Table 5.

After the punched plates were annealed, the tensile strength was restored to 35 86 tons per square inch, or nearly to that of the solid plate, which in this case was 36.22 tons, and this again is the strength of the metal in a drilled plate which by Table 10 = 36·3 tons per square inch. From this we find when the holes are drilled the metal left between holes is uninjured, its strength per square inch being equal to that in a solid plate.

It is not very clear how these experiments were made, but it would appear that the plates in Table 10 were all annealed to begin with; then after the holes were punched the strength was reduced from 36 22 to 24 333 tons per square inch, which by annealing a second time was restored nearly to its normal value, or to 35.86 tons. The drilled and annealed plates gave 36.3 tons, or practically the same strength as the annealed solid plate, which was 36 22 tons.

(41.) "Riveted Joints in Steel Plates."-Steel plates inch

32

STEEL RIVETED-JOINTS: EFFECT OF PUNCHING RIVET-HOLES.

TABLE 10. Of the EFFECT of PUNCHING and DRILLING RIVET-HOLES in STEEL PLATES.

thick, were riveted together with rivets, 13 pitch; the joints being in the six different forms given in Table 5. Unfortunately the results were vitiated by the weakness of the rivets. Of course when a joint fails by the rivets shearing it is no test of the strength of the plate: the only fair way is to take those cases where the rivets failed as giving the strength of the rivets, and vice-versa.

Taking from Table 11 the plates which failed, we have three with punched holes giving 40.98, 43.63, and 39.11 tons, the mean 41.24 tons per square inch. Then two plates with drilled holes failed with 39 25 and 42.93 tons respectively, the mean = 41.09 tons per square inch of metal between holes, which shows that with riveted joints, as with unriveted plates (40), the strength of annealed steel plates is the same whether the holes are punched or drilled. The mean of the five experiments on punched or drilled plates is 41-2, say 41 tons or 91,840 lbs. per square inch, which may be taken as the breaking weight of metal between rivet-holes in ordinary double-riveted steel joints. It will be observed that in some of the experiments from which that datum was derived, the joint had a back and front plate, the rivets being therefore subjected to a double-shear, but that fact seems to have made no difference to the results.

(42.) "Strength of Steel Rivets."-The resistance of mild steel rivets to a shearing strain may be obtained from Table 11:

TABLE 11.-Of Experiments on the STRENGTH of RIVETED JOINTS, in STEEL PLATES.

in seven cases the rivets were sheared with strains varying from 25.95 to 18.75 tons, the mean of the whole being 23.77 tons or 53,245 lbs. per square inch, which is remarkably low. It is generally admitted that the shearing and tensile strains are equal to one another, and (123) shows that this is correct so far as wrought iron is concerned. But the mean tensile strength of bar steel is 47 84 tons per square inch, this being the mean of sixty-six experiments in Table 1, so that it would appear that the fibres of steel are very seriously damaged in the act of riveting, the shearing strength being reduced to half the normal tensile strength.

This is the more remarkable because it is really lower than the shearing strength of wrought iron: the direct experiments of Mr. E. Clark in (123) give 24 14 tons per square inch as the mean of four experiments with single-shear, which is 1

D