clamped in any position, and to the other was fastened a small angle-bar, from the end of which a cord was led horizontally to a pulley fitted with roller bearings. This pulley was carried on an adjustable arm, and by hanging a weight on the end of the cord a turning moment was produced in the flange which could be accurately measured. A large mirror was also attached, with a

1000

CALCULATED STRESS

per degree or 0.00286 for 300°, as being that given by good observers between 32° and 212°.

Finally, to test the validity of the formula given by equation (8) for f, the stress at B, a stiff frame was made with two stout flanges pivoted on centres, as shown in Figs. 5. To one of these flanges was attached a quadrant by which it could be readily

COMPARISON BETWEEN ACTUAL STRESSES AND THOSE FOUND BY EXPERIMENT.

mark and distant slit, in a movable piece of cardboard, to indicate the rotation of the flange. The idea of using this slit, instead of a beam of light in the usual manner, was not due to the Author. Between these two pivoted flanges, a pipe, bent at right angles and provided with ordinary jointing flanges, was fitted, and securely bolted to them, care being taken in bolting up that there should be no initial stress produced in the pipe. To the centre of the flange A, with the clampable bracket, was attached a long flexible steam-pipe, and to the other, B, a flexible drain-pipe and a pressure-gauge. The flange A having been clamped, and the position of the flange B noted by means of the mirror and slit apparatus, the steam was turned on and the pipe expanded, turning the flange B, which was next brought back to its original position by hanging weights on the cord. The pipe was thus in the same condition as if bolted to a rigid framework and expanded; while the temperature was obtained by means of the pressure-gauge, and the stress was calculated from the length of the arm and the weight on the cord which produced the turning moment necessary to bring the flange B back to its original position.

Figs. 6 show the stresses calculated by equation (8) for the two pipes experimented upon, compared with the actual stresses observed, due to the turning moment produced in the flange B by the weights applied to the cord at the different temperatures. These observed stresses are fairly consistent, and agree as nearly as may be expected with the theoretical values, considering the uncertainty in the values of E and of a, as well as the disturbances due to the "strut effect" and to the compression already discussed. At the higher temperatures pipe 2 was stressed beyond the elastic limit. Hence the falling off in the curve.

It is worthy of note that where pipes are subjected to considerable vibration, in addition to stresses caused by expansion, they quickly give way at the flanges where there is local softening and weakness, even with the most careful manufacture-a point which is of growing importance in these days of high speeds.

The Paper is accompanied by six tracings from which the Figures in the text have been prepared.

10.00*..

BAR A

(Paper No. 3098.)

"Note on the Endurance of Steel Bars subjected to repetitions of Tensional Stress."

By ERNEST GEORGE COKER, B.A., B.Sc.

THIS Paper relates to a series of experiments upon two mild stee bars, which were repeatedly subjected to tensile stress carried beyond the breaking-down point, the amount of extension being kept approximately the same throughout the experiments. The chief points it was desired to determine were:-(1) The total amount of extension to be obtained under such conditions; (2) whether the tensile stress at breaking-down point was influenced by successive stretchings and annealings; and (3) the ratio of the work done on a bar

Scale, 1 inch = 1 foot.

broken under ordinary conditions to that done on a similar bar broken under the conditions given.

Three bars of mild steel were cut from the same piece of boilerplate, and were annealed and machined to size by a special machine in the testing-house. They were 10 inches long, 2 inches broad, and 0.44 inch thick. A chemical analysis showed that the amount of combined carbon in the steel from which the test-pieces were made was 0.16 per cent. Two of the bars, marked A and B

respectively, were subjected to tensional stress followed by annealing, and the third bar C was broken in the ordinary manner.

The testing-machine first

used was of the Wicksteed type, with a beam to test up to 50 tons, the jockey weight being 1 ton. The testing-machine afterwards used was capable of exerting a pull of 100 tons, and was similar to the smaller one except in details, the most important being that the beam was short and the jockey weight was 2 tons. The extensions of the bars were measured by a Kennedy extensometer of the ordinary type, to measure to 0.001 inch, and, for the rougher measurements, a pair of beam compasses with fine points was used, which, with the aid of a magnifying glass, gave

measurements accurate to 0.01 inch.

It was found by experiment upon other similar bars that when the breaking down point

was

reached, the sudden extension at this point was rather less thɛn 0.25 inch, and it was decided to extend each bar 0.25 inch before annealing and commencing anev. Occasionally an extension of 0.3

inch was reached before the load could be taken off. At the conclusion of each experiment the bar was measured and afterwards annealed by an experienced tool-smith. The annealing was per

formed by heating the entire test-piece to a dull red heat in an ordinary blacksmith's fire, and allowing it to cool in lime. The percentage of combined carbon being small, there was little danger of burning the plate during heating.

The dimensions of the bars at the conclusion of the tenth, fifteenth, twentieth, twenty-fifth, thirtieth, thirty-fifth and last experiments are shown in Figs. 1, together with the corresponding stress-strain diagrams, Figs. 2. The figures marked with Roman numerals refer to bar A, and those with ordinary figures refer to bar B.

Extensions of the Bars.-The total extensions of the bars A and B, treated in the manner described, are approximately four times that of the bar C broken in the ordinary manner, Fig. . The bar A was subjected to thirty-seven extensions before it broke, involving as many annealings; and its length when fractured

measured 19.58 inches between the measuring points The bar B was subjected to thirty-nine extensions before it broke, and its length when fractured was 20.76 inches. The ratio of the extensions of the bars A and B to that of the bar C are as follows:

These results do not differ more than might be expected, having regard to the possible sources of error, such as extending the bar beyond the fixed amount (0.25 inch), uneven setting in the testing-machine, etc.

Breaking-Down Point.-The second point which it was sought to determine was whether the stress per square inch ator near the breaking-down point varied with the number of amealings to