Professor Thurston, Materials,' 1883, vol. ii. p. 576. Prolonged excessive straining causes rupture.

Collingwood, Am. C. E.,' 1880, vol. ix. p. 171. Tenacity of wire changes after a time.

L. Fletcher, C. E.,' 1884, vol. lxxx. p. 136.

A. J. Maginnis, 'Engr.,' 1885, vol. Ix. p. 447; C. E. Stromeyer, 'C. E.,' 1886, vol. lxxxiv. p. 187. Sketches of plates which cracked spontaneously while not in use.

J. Harrison, Engr.,' 1886, vol. lxii.

'Army N. J.,' 1887, vol. xxiv. p. 65. A long list of steel armour plates which cracked spontaneously before being fitted.

Collingwood, Am. C. E.,' 1880, vol. ix. p. 171, and W. Hewitt and Felton, Am. M. E.,' 1888, vol. ix. p. 47. Strength and ductility of wires and plates change 24 hours after rolling.

H. M. Howe, 1890, p. 195, old armour plates are brittle; p. 210, time removes injury caused by cold working.

For other cases, see p. 24.

Although not as yet published, there are a few cases of boiler shells cracking under the hydraulic test, and particularly of furnace saddle corners cracking while out of use or if struck with a hammer. One of the most curious cases is perhaps the following: A steel steamer which had been on fire was being repaired, and several of the buckled plates had been taken out to be straightened. It was found that some of these cracked spontaneously while lying on the ground, although they had not cracked while being removed from the ship.

Besides these various failures other instances will readily suggest themselves about the influence which time has on the quality of iron and steel. There is the case of the gun of H.M.S. 'Collingwood,' which burst with a light charge after two years' rest, succeeding on its trial with several proof charges. These and similar failures are so very mysterious that they have been attributed to hidden or incipient flaws; but they have never been seen, and if the point is conceded that certain qualities of steel have a natural tendency to change and this is stoutly maintained by experienced steel manufacturers-then the difficulty is somewhat reduced. That important changes are slowly occurring in steel is proved by the fact that ordinary test samples which have been successfully bent to a small radius crack spontaneously some time afterwards; that armour plates, after having been fired at, emit strange sounds for a long period, and that the elastic limit of steel test pieces which have been stretched a few per cent. slowly grows higher and higher when at rest.

Time can hardly be called a treatment, but the mystery attaching to its effects should be a sufficient excuse for most carefully investigating all such cases as may possibly have been produced by it.

Influence of Punching.--One treatment which produces injurious effects in steel is punching holes into the plates. The experiments on the subject are too numerous to be mentioned here, but many of them will be found in the chapter on Mechanics,' under Riveted Joints,' p. 223. The thickness of the plate, the diameters of the punch and die, as well as the hardness and chemical composition of the plate, affect the pliability of such samples. It has been found that rimering out in. of the holes removes all bad effects; but if

it is true, as stated by Mr. Beck-Gerhard (Gorni J.,' 1884, p. 347), that the curves of stress slowly extend at least 5 ins. away from the hole, then punched plates ought to be annealed or rimered at once. This experiment was as follows: A-in. plate was polished on one side and punched (in a cold atmosphere). Spiral curves then showed themselves, which were first washed with aqua regia. The piece was then planed into several strips and each tested, when the spiral curves reappeared and were perceptible to the touch.

This may explain the curious curved markings near punched holes and sheared edges. In every one of these lines the surface scale has fallen off, showing that here stresses have been at work producing local deformations of at least %, for it is only after steel has been stretched this amount that the mill scale falls off the plates. Illustrations of similar effects will be found in Kirkaldy's works. pp. 156 and 165.)

(See

Influence of Severe Stresses. An important matter is the behaviour of steel and iron under severe stresses, and a good deal may be learnt on this subject by watching a tensile test piece under the following conditions: After being fixed in the machine a strain indicator is attached, and the loading carefully proceeded with, preferably by the addition of small loads, and not by moving a jockey weight. The elongation produced can then be accurately reduced to zero by removing the weights.

After-strain. If a properly shaped solid bar of lead be struck with a hammer, it emits a musical note whose number of vibrations enables the modulus of elasticity to be calculated. It will be found to be much higher than that which can be calculated from strain indicator readings taken in a testing machine, and this shows that the elasticity of lead differs according as to whether the stress is applied quickly or slowly it is a function of time. The gradual elastic extension of the material with time, however small the stress, is called after-strain.' Apparently it is a property possessed by nearly all materials, and though of no great consequence, it has made itself felt in a most annoying fashion in aneroid barometers. taken to a high altitude, these instruments adapt themselves to the reduced pressure, and of course give wrong readings, and then, after having been brought down again to the sea level, they once more adapt themselves to the increased pressure and give correct readings. The slight but gradual springing back of newly bent boiler shell or furnace plates may perhaps be attributed to this cause. In accurate research work it is well to bear this in mind, even although steel generally shows but slight after-strain.

If

Dynamic Elasticity. So called by Lord Kelvin, as it is dynamic modulus of elasticity which has to be reckoned with where quick changes occur, as in vibrating objects. It is only natural that even the slightest change of form leads to a slight change of temperature and of volume.

We have therefore to distinguish three elasticities: Static elasticity, dynamic elasticity, and after-strain.

Viscosity. It will be found that even with small loads a very small permanent set takes place. This is generally attributed to viscosity of the material, and is best studied on wires subjected to

torsion. Pitch possesses this quality in a high degree, and quartz is said to be absolutely free from it; probably it is this property which makes a metal non-sonorous, from which it would follow that hard steel, silver, and glass are not of a viscous nature. (H. Tomlinson, 'Phys. S.,' 1887, vol. viii. p. 171; 1888, vol. ix. pp. 49, 67.)

Modulus of Elasticity. The loading of the test piece can now proceed, careful readings of the strain indicator being taken. It will be noticed that for steel and iron the elongation is almost proportional to the stress, being at the rate of 10 in. in 10 ins. for about every 13 ton per square in., from which it follows that the modulus of elasticity is about 13,000 tons, or 30,000,000 lbs. per square in., or 20,000 kils. per square millimeter. With cast iron and various other metals there seems to be a change in this modulus when the stress is increased beyond a certain point, and some people have called this the limit of elasticity, but a better name is limit of proportional elongation. The more accurate the strain indicators are, the more gradual does this change appear, and the obvious conclusion is that with these metals the modulus of elasticity is a variable quantity, growing smaller as the stress increases. It also decreases about 1 per cent. for every 100° F. rise of temperature (see H. Tomlinson, Phys. S.,' 1887, vol. viii. p. 171).

Elastic Limit. It will be noticed that the pointers of the strain indicators oscillate slightly for every newly added load, coming to absolute rest only after a very long period. But when a certain stress has been reached this action ceases and the pointers acquire a slow onward motion. The elastic limit has now been reached, and to verify whether this is so or not, several or all of the weights are removed, and the pointers will either return to the positions previously occupied or not. This check is necessary, as the giving way of the attachments of the test pieces sometimes produces strange effects, and may even cause one of the pointers to travel backwards. For this reason also the determination of the elastic limit should always be made with the help of three strain indicators. Mild steel shows an elastic limit of about 15 tons. If very mild and previously tempered it is sometimes as low as 10 tons, but in such cases it is hardly perceptible and no breakdown point can be noticed.

6

Breakdown Point.-Continuing the straining of properly annealed samples, a point will be reached when the slow motion of the pointers gives way to a very rapid one, which is not stopped even by reducing the stress by several tons. This is called the 'breakdown point,' because of the behaviour of the material, or the drop,' the lever falling through a considerable angle. It has also been called the limit of plasticity, because above this point the material behaves as if it were plastic. This point is often mistaken for the limit of elasticity, and the two points sometimes fall together.

Irregular Stretching. Even now it is of interest to watch the strain indicator, for it will be found that the plastic elongation proceeds very irregularly. Generally after adding a weight it commences slowly, increases, and then diminishes, until at last no further motion can be detected. Very often, particularly if only small additions are made to the load at one time, the pointers vary their

speed repeatedly, increasing and diminishing their velocity several times without any additional weights being added to the lever.

Irregular Elongations. An explanation will suggest itself if several (say, four) short strain indicators are attached along the length of the sample, for it will then be noticed that first one and then the other span is elongating, showing that waves of plasticity pass along the samples. This may also be noticed when the breakdown point is reached, for then the mill scale falls off, first at the extremities, and then more towards the centre. The scale falls off when the stretch

exceeds % of the length.

The same phenomena, but more marked, may be noticed when twisting wire in a torsion machine. Instead of proceeding uniformly, it will be noticed that the twist commences at one end (fig. 115), and that, like a wave, it travels to and fro till the sample breaks.

FIG. 115

Changes in the Limit of Elasticity. The next point to be noticed in a test piece is that when once the elastic limit has been passed, and the sample then unloaded, it will not again elongate permanently until the previous stress has been reached. This is only natural, for the second testing is but a continuation of the first, and it is not difficult to accept the statement that the elastic limit of a sample is raised by preliminary testing.

R. H. Thurston, 1883, p. 601, has suggested that this behaviour would enable one to detect whether a broken structure had been overstrained. A test piece would have to be carefully cut from the plate and its limit of elasticity determined. Unfortunately experiments show that, even with the most careful handling during preparation, the elastic limit of the test piece has again fallen to its original value, or perhaps the shock of the rupture has produced this effect.

Influence of Time. The test piece under consideration could now be stretched till it has elongated 5%, which will raise the stress to about 25 tons. It may then be left in the machine overnight with the full load on it, or it may be put aside for a day or two. On re-testing it will be found that the elastic limit has risen considerably above 25 tons, which was the last stress to which it was subjected. The following experimental results will illustrate this:

A long list of experiments on this subject by Bauschinger is contained in Civil I.,' 1881, vol. xxvii. p. 1; also Mitt. Munich,' 1886, vol. xiii. 1891, vol. xx. He investigated the behaviour of 14. samples of iron and steel, and also copper and gunmetal. Unlike the author's experiments, none of his showed an increase of elastic limit beyond the ultimate strength of the material, but even with him the influence of time in raising it is very marked. With copper and gunmetal the elastic limits only rise as high as the preliminary stress. In all these cases the second elastic limit and breakdown point fall together, and the drop is now very much greater than with an annealed sample.

The question naturally arises, What would happen if the sample had first been subjected to a compression test? This experiment has also been carried out, and it was found that the elastic limit for tension had been reduced from 19.3 and 16·3 to 11.8 tons, and the ultimate tenacity raised to 307 tons, elongation 7.2 %.

A preliminary compression stress at right angles to the axis of the sample (produced by drawing it out under a hammer) raised the limit to 20.5 tons. This also increased the tenacity to 32 tons, and reduced the elongation to 12%.

These experiments readily suggest that, as the elastic limit is a changeable value, it cannot be a reliable measure of the working strength of a material. When a preliminary test has raised the elastic limit considerably it may be dangerous to repeat it, because if the new elastic limit is accidentally exceeded a very considerable breakdown occurs, which may lead to rupture. Those parts of a structure which have been subjected to excessive compression stresses should not be exposed to severe tension stresses, as their elastic limits of tension have been lowered and their ductility reduced. The reverse is also probably true.

Fatigue. Člosely related to the last proposition are the deductions drawn from Wöhler's celebrated experiments on the effect of alternating stresses, better known by the name of fatigue.

They have been and are still being repeated, and it would appear— Firstly, if the experimental tension and compression stresses are sufficiently low, these may be repeated an infinite number of times without producing rupture.

Secondly, rupture will be produced by alternate stresses, if these are sufficiently high, but kept below the elastic limit.

Thirdly, the more intense the alternate stresses are, the sooner will rupture occur.

Fourthly, if the alternate stresses are equal in intensity, they produce rupture more quickly than if one of them is small or does not exceed zero.

Attempts have been made to find a relation between the power of a metal to resist fatigue and its other known qualities, such as elastic limit, strength, and elongations, but as yet these are not reliable.

6

A good deal of information on this subject will be found in the following publications:-Wöhler, Zeit. Bw.,' 1860, vol. x. col. 583; 1863, vol. xiii. col. 243; 1866, vol. xvi. col. 67; 1870, vol. xx. col. 90. Spangenberg, Zeit. Bw.,' 1874, vol. xxiv. col. 482; 1875, vol. xxv. col. 79. Ibid. 'Glaser's An.,' 1879, vol. v. col. 6.