strength and do not elongate nearly as much as short ones, but this is true only for materials which contract at the point of fracture. Mild steel contracts very much, so that results as to elongation would be misleading unless the length of the sample is mentioned. The general practices as to length are

Board of Trade, 10 ins.

Admiralty, Lloyd's, and Continent, 8 ins., or 200 mm.
Artillery experts, Whitworth,

and railway tyres

about 2 ins.

This short length of 2 ins. is due to the want of dimensions in the gun barrels. The removal of the scale reduces the elongation and increases the tenacity.

The Testing Machines all work on the principle that the sample is attached by one end to a weighted lever, and by the other end either to a screw or to a press which can be adjusted to make up for the elongation. With the older machines the screws can be worked only after the load has been removed, and the testing of ductile materials cannot be carried out in one operation, but this does not seem to affect the results.

In some machines the samples are placed horizontally, in others vertically, and the levers are either simple or compound, and the loading is done either by a jockey weight travelling on a lever, by weights added to the extremity of the lever, or by a fluid pressure or a pendulum weight.

The oldest machines are worked by added weights. On the Continent (chiefly Austria) water is run into a large bucket at the end of the lever. Generally a weighing machine is permanently stationed under the bucket, so that it can easily be weighed, but a gauge glass is also attached.

In England single levers and jockey weights are now customary. The stretch is taken up by steam power acting on a ram, and the speed is about 3 inches stretch per minute.

At some German works the lever is a loaded pendulum, which rises while a hydraulic ram stretches the sample.

In France and Belgium the lever is a crank, which presses on an iron disc resting on a sheet of leather, which covers a basin filled with mercury, and which communicates with a graduated vertical glass tube. The total load can be read off on a scale. This arrangement gives very accurate results, and can with advantage be used for ascertaining the drop (or elastic limit); but it is unsatisfactory in cases of dispute, because the column of mercury sinks back to zero at once when the test piece is broken. The defect could be remedied, or by attaching an ordinary steam indicator automatic diagrams could be obtained.

Descriptions and illustrations of testing machines will be found in the following publications: J. H. Wicksteed, M. E.,' 1882, p. 384, and 1886, p. 27; Prof. A. B. W. Kennedy, 'C. E.,' 1887, vol. lxxxviii. p. 1; U. R. Towne, M. E.,' 1888, pp. 206, 448. Engineering,' vol. xxxi. p. 57; vol. xxxiv. p. 254; vol. xxxv. p. 346; vol. xxxvi. p. 146; vol. xli. p. 180; vol. xliii. pp. 414, 572; vol. xliv. pp. 649, 652; vol. xlv. p. 458, &c.; vol. xlvi. p. 21.

Many testing machines are so arranged that they can be used for measuring compression-bending and torsion stresses. Appliances for

bending samples can be attached to almost any machine, but that is not the case as regards compression and torsion. In the former case very careful adjustments are necessary, in order that the thrust may be perfectly central and the bearing perfectly normal. In both respects most machines are very imperfect. Almost any lathe can be used for the torsion test. It is a valuable one, but not carried out often enough, and then the lessons it teaches are not properly understood (see p. 125). Strain Indicators.-When the elastic limit has to be ascertained strain indicators are indispensable. A description of these will be found in C. E.,' 1887, vol. lxxxviii. p. 1. The principle on which they are designed is the accurate measuring of the elongation or contraction of a test piece when loaded, but these changes of length are so minute that they have to be magnified at least 200 fold, better still 1,000 fold. This can be done with the help of microscopes; but they are very expensive instruments, and although accurate, are very inconvenient to work with. Micrometer screws have also been used, but unless they make and break contact electrically, they do not seem to be very accurate; besides, they are liable to have their threads stripped or damaged. Most of the strain indicators in practical use consist of some lever arrangement, and this always involves friction, which masks the effect of after-strains and may even show permanent sets before they occur. Rollers can of course be used instead of levers, but they are also not free from friction. The author has been very successful with rolling pins consisting of hard steel wires of diameters ranging down to 1 in., to which are attached light balanced, straw pointers of 3 in. length; a magnification of 300 is obtained, which for many purposes is ample. The author has also adapted the phenomena of interference of monochromatic light waves to the measurement of strains. With such instruments a movement of one millionth inch can be detected; this is the elongation which occurs in one inch of steel when a stress of 30 lbs. per square in. is applied. The instrument is the only one that seems to have given satisfaction when used for the direct measurement of Poisson's Ratio. (C. E. Stromeyer, 'Proceedings,' 1894, vol. lv. p. 377.)

1

When using strain indicators one cannot be too careful to have the pull quite central in the test piece. Thus if in a test piece of in. thickness the pull falls in. to one side of the central line, or if the test piece is bent in., then a strain indicator placed on one side will give readings which are 6 per cent. too high or too low. For this reason wedges should not be used for holding test pieces when these delicate instruments are to be used; in fact, whatever the mode of attachment, the correct arrangement is to surround the test piece with three identical strain indicators. Their readings will tell one not only whether the pull is central or not, but also what the stress is at any particular section. This is of great importance when investigating limits of elasticity, for then it is not the average stress, but the maximum stress which is required to be known.

Speed of Testing does not seem to affect the final result, at least not unless the speed is excessively great, as in percussion tests, or unless the test is prolonged for hours or days. See J. Bauschinger, 'Mitt. München,' 1896, vol. xx.

Percussive Tensile Tests are carried out by securing the top end of

a tensile test piece to a strong but narrow beam, and attaching a crossbar to its lower end. A heavy weight, whose lower end is forked, is then dropped on it, striking the cross-bar. Another plan is to attach the weight to the lower end of the test piece and a cross-bar to its top end; the whole is then raised and dropped; the cross-bar is arrested by stops, while the weight tears the sample asunder. This is the more usual plan, and seems to be the better of the two.

Colonel Maitland (C. E.,' 1887, vol. lxxxix. p. 114) was able to obtain a still more sudden rupture by shaping the test piece somewhat like a dumb-bell, surrounding the central, thinner part with gun cotton, and inserting both into a strong tube open at both ends. On firing the explosive, the two ends were driven out with great violence in opposite directions. It was found that during rupture the samples had elongated very much. This, however, might be due to combined action of the longitudinal pull and the surface pressure on the bar, acting like the forces which come into play when drawing wire (see p. 167).

Chipping, Machining, and Scratching have repeatedly been proposed for use, as they seem capable of giving valuable information. Thus the turning tool of a lathe will show up very distinctly the various slabs and layers in an iron bar, exposing, as it were, very slight differences in the hardness or toughness of the material. With a hammer and chisel there is no difficulty in distinguishing between iron and steel, and some boiler-smiths profess to be able to tell, with the help of these tools, where a particular piece of steel has been manufactured. This test has not yet been made practical; the same may be said of etching and the microscope, and magnetic tests.

Micro Structure of Steel. If carefully polished surfaces are treated with dilute acids, the micro structure can be studied and valuable deductions drawn from it. Steel is now looked upon as being a conglomerate of various compounds of iron and carbon to which names have recently been given.

Ferrite is pure iron. Cementite is Fe3C.

Pearlite is an intimate striated mixture of ferrite and cementite. Martensite, Austensite, Troostite, and Sorbite are four other compounds found in steel.

Ferrite is not attacked by strong nitric acid. Dilute solutions develop its crystalline structure.

Cementite remains bright after etching with iodine solution or infusion of liquorice.

Pearlite and Martensite can be coloured by iodine solution. 'M. E.,' 1899, p. 53.

As yet no name has been applied to the sulphur and phosphorus compounds.

Researches on micro structure may one day grow to be looked upon as valuable supplements to chemical analysis, but neither can supersede mechanical tests as guides in commercial transactions.

Prof. Egleston, Am. M. E.,' 1879, vol. v. p. 140. S. M. Saxby, 'N. A.,' 1884, vol. ix. p. 61; 1885, vol. x. p. 119. Prof. D. E. Hughes, 'M. E.,' 1884, p. 36. C. M. Ryder, 'Metall. Rev.,' 1877-8, vol. i. p. 317. Prof. K. Keller, O. I. A. V.,' 1879, vol. xxxi. p. 163. Dr. H. Wedding, Stahl und Eisen.' J. A. Ewing, 1891. A. Sauveur, Trans. Am. I. M. E.,' 1893, vol. xxii. p. 546; T. Andrews, 'Proceedings,'

1895, vol. lviii. p. 59; Roberts-Austen, Transactions,' 1896, vol. clxxxvii. p. 417; I. and S. I.,' 1891, p. 100; 1896, vol. i. p. 486; 1897, vol. i. p. 42; 1897, vol. ii. p. 115; 1898, vol. i. p. 145; 1899, vol. ii. p. 102. Etching at a red heat: E. H. Smith, I. and S. I.,' 1898, vol. ii. p. 137.

With the help of some of the above-mentioned tests it has been attempted to solve the various problems of the behaviour of iron and steel. Some have been cleared up, others remain mysteries.

The Influence of Temperature, particularly of heat, on the strength of materials is a subject of the first importance. The relation existing between the boiler pressures and temperatures will be seen from the following table :

50 100 150 200 250 300 400 500

°F. 212 281 328 358 382 401 417 445 467 °C. 100 138 153 181 194 205 214 230 242

Original researches on this subject will be found in the following papers-Franklin Inst.,' 1836, ii. pp. 82-208; M. Baudrimont, An. Ch. Ph.,' 1850, iii. vol. xxx. p. 304; W. Naylor, M. E.,' 1866, p. 76; W. Fairbairn, 1856, 2nd ser. p. 96; Sir W. Fairbairn, Manch. L. Ph.,' 1871, vol. x. p. 86; Portsmouth Dockyard Experiments,' 1877; C. Huston, Frankl. Inst.,' 1878, vol. lxxv. p. 93; N.,' Glaser's An.,' 1880, vol. vii. p. 165; Dr. J. Kollmann, 'Ver. Gew.,' 2nd ser., 1880, vol. lix. p. 92; J. F. Barnaby, 1881 and 1882; J. E. Howard (Watertown Arsenal), I. and S. I.,' 1889, ii. p. 460; Le Chatelier, Comp. Rend.,' 1889, vol. cix. p. 58; Board of Trade Report (No. 257) on 'Boiler Explosions'; Prof. Martens, 'Mitt. Berlin,' 1890, p. 159; A. Bleichenden, M. E.,' 1891, p. 320. See also p. 195 as regards elastic limits at high temperatures.

6

The Influence of Cold on the tenacity of metals is dealt with by the following experimenters :-Sir W. Fairbairn, Parl. Report of the Commissioners of Railway Structures,' p. 321; ibid. Brit. Assoc.,' 1857, p. 405; K. Styffe, 1869; N.,' 'Glaser's An.,' 1880, vol. vii. p. 165; Capt. Bernardo, Rev. d'Art.,' 1890, p. 485; W. Brockbank, Manch. L. Ph.,' 1871, vol. x. p. 77; W. W. Beaumont, C. E.,' 1876, vol. xlvii. p. 43; T. Andrews, C. E.,' 1887, vol. lxxxvii. p. 340; Spangenberg, Glaser's An.,' 1879, vol. v. p. 165; T. Andrews, C. E.,' 1891, vol. cv. pp. 161, 169; W. Rudeloff, Mitt. Berlin,' 1897.

6

Experiments on the influence of temperature on copper and other materials: Frankl. Inst.,' 1836, ii. p. 39; Dr. Kirk, Enging.,' 1887, vol. xliv. p. 661; W. Parker, 'N. A., 1889, vol. xxx. p. 47; Le Chatelier, Comp. Rend.,' 1889, vol. cix. p. 24; Le Chatelier, Paris, 1891. Dr. J. Kollmann's experiments on iron and steel, 150 in number, seem to be the most reliable. His conclusions are that the ultimate strength of both materials decreases with rising temperature, and that it shows a very serious drop at about 450° to 500° F. The elongation is at a maximum at about 900° F. The contraction of the fractured sectional area steadily increases till it reaches 90% at a red heat. The limit of elasticity decreases steadily.

Another result is that the ductility, as measured by bending tests,

increases with rising temperatures till 450° to 500° F. (blue heat) is reached, when the material is rotten. At higher temperatures it is pliable once more. It would seem as if medium quality steel is exceedingly brittle when cold (0° F.), while tougher qualities of the same tenacity remain ductile; at any rate, boiler-smiths and ship-platers have come to the conclusion that it is risky to handle steel or iron plates in cold weather, and where serious hammering or drifting is contemplated, heaters are almost invariably applied. If the steel contains much phosphorus it is exceedingly likely to break under the percussive bending test, if carried out in cold weather.

Sudden cooling, not necessarily from a red heat, is said to produce brittleness (see T. Andrews, 'C. E.,' 1891, vol. ciii. p. 231); and repeated heatings have also produced brittleness, but the exact conditions for effecting this change are not known. E. Wehrenfennig, 'O. I. A. V.,' 1879, vol. xxxi. p. 153; J. P. Barnaby, 1881 and 1882; A. Ledebur, 1884, p. 646; C. E. Stromeyer, C. E., 1886, vol. lxxxiv. p. 122; E. B. Martens, 'Ing. Civ.,' 1886, p. 607; E. Wehrentennig, Organ,' 1884, vol. xxi. p. 216; A. E. Sherk, Enging.,' vol. xliv. p. 458; C. E. S., ibid. vol. xliv. p. 491; B. H. Thwaite, ibid. vol. xliv. pp. 505, 536; Th. Edington and Son, ibid. vol. xliv. p. 505.

Occlusion of Gases. In some of the above-mentioned cases the brittleness may have been due to the absorption (occlusion) of injurious gases, and experiments prove that hydrogen is readily absorbed, and that it injures the material. Acids and other corrosive influences are also said to produce brittleness, but the real cause may be the hydrogen which is evolved during these processes.

Dr. Schahäntl (Bayrisches Kunst- und Gewerbeblatt,' June 1863) gives an analysis of the various layers of an exploded boiler plate, and shows that on the water side it contained an excess of oxygen, while on the fire side occluded sulphurous acid was found. That solids do absorb gases is proved by the well-known fact that soaped window glasses turn a mauve colour after a time, which is due to a chemical action of the oxygen of the air on the manganese salts in the glass.

M. Bustein (An. Mines,' 1883, viii. vol. iii. p. 28) gives the chemical analysis of three qualities of steel which had been exposed for 112 days in flue gases or in boiler water.

Previous Treatment of Sample

Prof. Hughes (Tel. Eng.,' 1880) deals with this subject.

Influence of Pickling.-A. Ledebur (Stahl und Eisen,' 1887, vol. vii. p. 682) gives analyses of eight qualities of steel (wire), and finds that pickling reduces the elongation about 15%, and the ductility 39%; that exposure to the atmosphere for two months reduces both qualities about 50%, but that annealing puts matters right again, though the ductility is not perfectly restored. The action of zinc in galvanic contact with the wires is worthy of notice; it prevented corrosion, but the wires grew very brittle, and on