These experiments with elastic are necessarily somewhat more expensive than the wedge-block type previously described, because of the special preparation required by the specimens and the additional time occupied in ascertaining the numerous extensions necessary when testing them.

(Half length.)

Another class of experiment is adopted when material is sent of sufficient length. The extensions are measured upon a length of 100 instead of 10 inches; these experiments are specially important for ascertaining the minute extensions of material at stresses below the Elastic, as the long length enables such to be measured with great precision (see tabulated Reports L, M, and diagrams, Plates II. III.).

When full-size articles, such as tension-links, tie-bars, &c., are tested, the distances for extension range up to 300 inches. These readings, or extensions, are all given in reports upon specimens of long length. (See examples N, O, P.) This class of experiment does not allow of the readings being ascertained. right up to the moment of fracture, on account of the risk of damage to the fine measuring instruments, which, though removed before fracture occurs, are retained in place as long as practicable. It will be seen, on referring to N, that the readings are reported up to 60,000 on the entire length, and up to 50,000 on the body length of 200 inches. Besides stating in full all the information described under the preceding classes, the ultimate total extension upon the entire length is given, also the ultimate total extension upon the 200 or 100 inch distance, and further, the ultimate extensions in each of the 5 inch spaces which subdivided the same.

The reason for giving the ultimate extensions in the respective subdivisions is, that such afford information as to the uniformity or otherwise of the material throughout the long length, and also as to the position of and local effect produced by the fracture.

The Elastic stress was defined from the commencement of operations to be the amount of stress at which the rates of extension ceased to be in proportion to the increments of stress, and no reason has been found for change or departure from that definition. This is an accurate and practical method of determining the Elastic, namely, by observing the extensions at regular intervals or increments of stress, and INVARIABLY in every report in which it has been stated or given it has been honestly ascertained. There has never been any shirking of the work under the plea of being tedious, and guessing the amount as appears, most unfortunately, to be far too frequently done elsewhere.

After reports had been issued during many years containing the Elastic stress, it was found that they were being imitated in statements of tests by other persons; although Kirkaldy felt sure that the results set forth in such were not obtained in the careful and exact method pursued by him, he was unable to find out upon what those statements were based.

The matter was, however, cleared up in a Paper by Professor Kennedy (see "Proceedings Institution of Mechanical Engineers," April, 1881, p. 210) in which he rendered the service of stating the methods in use by other persons. In that Paper, after the description of his means of ascertaining "elasticity," the following paragraph occurs, which, though not quoted in full, the parts chosen give a fair idea of the subject: "Neither of the two points mentioned-namely the points where permanent set begins and where uniform "extension ends-can be determined without such special and tedious measurements of small extensions as "will enable such curves as those of Fig. 1, Plate 30, to be drawn out. Neither therefore can be noticed in "ordinary testing; and, consequently, neither of them is the point commonly fixed as the limit of elasticity. “.... What is called commercially the limit of elasticity will be found to be a point very considerably "higher than the limit which corresponds to any of the usual scientific definitions." The author further describes in that Paper that it is the point at which the steelyard drops during the test, that is considered the limit of elasticity in ordinary testing, and the change which occurs in the material he

terms "breaking down." Again (p. 212), “. . . . If, lastly, it be taken (as it practically is always for "commercial purposes) as the point where the material 'breaks down'-the point given in Column III. of the “Tables—it is not reached until at 68 per cent. of the breaking load." The placing on record of this method of procedure was a good service, but it is very much to be regretted that a few lines after the last quotation it should be stated, "In any further reference to limit of elasticity in this Paper, it is the point where the "material breaks down that will be referred to, unless a special statement is made to the contrary." Thus identifying himself with, and lending sanction to, a crude and unreliable method of testing. A method that gives high results will be best liked by manufacturers and contractors, but the preferences of such should neither be a reason for, or an inducement to, any man who desires to be independent to comply with such a method under the plea that it is customary. When a man shows in his own writings that he will conform to custom, and use a process, although it gives far too high results, how is it to be known that his work on other occasions is not done upon the same lines?

When designing the machine, David Kirkaldy considered most carefully his moral procedure, as well as the method of ascertaining the extensions; he aimed at obtaining extensions with great accuracy, and to as fine a degree as it would be possible to carry out faithfully right through all his work. He abhorred the idea of making a few experiments with very special instruments which would give the impression that he recorded the extensions to a very minute degree, if such instruments could not be used regularly; therefore he decided upon a system which would give very accurate results, and be, moreover, practicable, and he has been entirely successful: throughout the twenty-five years all classes of work have been treated with precision and with uniformity.

To work by the dropping of the steelyard not only renders the result fictitiously high, but is, moreover, uncertain and variable; if the excess by the process were a constant amount for all classes of material, or even materials of the same class, it would still be objectionable enough; but the excess is not constant, many kinds of material do not show this characteristic at all, others at a very late stage. In fact, it is only certain soft materials that show the characteristic in a marked degree, therefore Kirkaldy, though well aware of the "dropping of the steelyard," would have nothing to do with that as a means of determining the Elastic.

If Kirkaldy made experiments one day with wonderful minuteness, and yet on another was careless, how could clients know in which way any given work had been done? Whereas it has always been his ambition to carry out his work from year to year in such a way, and so systematically, that all should be exactly comparable. By adhering strictly to the practice of recording all the extensions and determining the elastic stress therefrom, the reliability of all work done has been maintained.

So much trouble having been taken and energy expended in maintaining the equality of his work, while exposed to great pressure to slacken his care, it is consequently discouraging to find other men, who, from the position they hold, ought not to think of what may be customary, but of what is right, falling into the very errors that he was so anxious to avoid. The sanction lent to such practices is especially annoying, because every additional instance renders it more and more difficult for Kirkaldy to maintain his system in its entirety.

THRUSTING TESTS.

Specimens of wrought iron and steel, alloys of copper, and other ductile materials are tested in two ways: 1st, in the form of specimens long enough and free to buckle, one standard length being 10 inches;

2nd, in the form of specimens short enough to resist buckling action, usually 2 inches long. With ductile materials there is no ultimate Thrusting stress if buckling does not come into play, because the material yields or spreads laterally the more the stress is increased. With all experiments of this kind the rate of depression, or shortening, is carefully ascertained, so as to determine the Elastic stress precisely upon the same system as described for the Elastic under Pulling Tests, for style of Report see G, I, K.

When treating cast iron both forms of specimens are tested as frequently as possible, i.e., the 10-inch and 2-inch lengths. The ultimate Thrusting stress is recorded because cast iron gives way by fracture, the amount of depression, or shortening, at time of fracture being also obtained upon the 2-inch length. The rates of depression are recorded when the work is in connection with a complete series of tests. But for ordinary specimens connected with Contracts the ultimate stress and depression only are given, with sizes (see Report SS).

(Half length.)

When sufficient material is sent, say of steel or iron plates or bars, specimens of long length are tested to compare with corresponding lengths under Pulling stress (e.g., Report L). These specimens may be 10, 25, 50, or 100 inches long, and are prevented from buckling by suitable apparatus. Such experiments are of special importance to ascertain the relative behaviour of the same material under opposite stresses; to obtain information as to whether the rate of decrement under Thrusting is the same as the rate of increment under Pulling, ought to be considered of vital consequence for large structures, such as bridges.

The reason for employing the term "Thrusting" may be explained; it is used because it really describes best the action that the specimen is subjected to, and is, moreover, appropriate in every case. It is surprising how continually other persons, when speaking and writing about tests, use the term "Crushing," which is often not appropriate. Again, the term "Compression" is frequently used, which shows lack of knowledge of what takes place. Material does not always become denser, or actually compressed, under test; in a great number of cases the specific-gravity of Thrusting specimens has been determined before and after testing, and the results obtained have amply justified the term adopted as being the most appropriate in all cases. Many metals become less dense when subjected to Thrusting stress, through extending laterally in a greater degree than is due to the shortening of length.

BENDING OR TRANSVERSE TESTS.

This kind of test has been designated by many as "Breaking," which appears strange, and even absurd, because ductile materials do not break under this test, but double up. Besides being inappropriate to many classes of material, that term is also confusing, because all specimens under tensile tests are broken; therefore the term, if used at all, would be more appropriate for that kind of test than for the Bending.

The greater number of experiments under Bending comprise the determination of the Elastic stress, as well as the maximum or Ultimate. The Elastic is determined from the recorded deflections observed at regular additions to the load, which is applied at the centre of the specimen or article being tested. The amount of addition or increment of the load for each observation is adapted to suit the material, size of specimen, and the span or distance between the supports.

With certain classes of work only the ultimate stress and ultimate deflection are determined, because

that information meets the requirements of the particular case. For instance, with cast-iron bars that are samples of what is being supplied under ordinary contracts, the minute readings, or rates of deflection, are not necessary, and therefore are not recorded. (For example of cast iron, see Report SS.) The machine proper is adapted for any span from 5 to 120 inches, and by the application of the two wing-beams the span can be increased up to 300 inches, as exemplified in Report S.

Standard spans have been adopted for regular work, which are maintained; exceptions have to be made occasionally to meet particular requirements, but such exceptions are avoided as much as possible. A few of the standards may be mentioned

RAILS are tested upon a span of 5 feet-length of pieces required, 6 feet each; ten samples, or pieces, being usually the minimum number.

ROLLED JOISTS: light sections 5 feet, and heavy sections 10 feet. Specially large joists, and when a series of experiments are being carried out, tests are made upon 5, 10, and 20 feet spans. Length required: at least 10 or 12 inches more than the span; extra length does not matter, as there is space for handling. The flanges of girders-that is, the top and bottom faces-should not exceed 12 inches in width.

TIMBER, in the form of joists or beams, 10 feet span.

CAST IRON-The usual form of bar is 1×2 inch, and 42 inches long, tested upon a span of 36 inches. CAST, rolled or hammered bars forming part of a series of experiments are planed to definite dimensions and tested upon standard spans of 20 and 30 inches.

STONES-Marbles and granites for steps, lintels, &c., are tested of the section to be used, or made to 6×6 inches-convenient span, for comparison, is 36 inches.

N.B.---Girders and joists requiring the application of only a proof-load can be satisfactorily dealt with upon a span of 10 feet, which is a convenient one, by calculating the amount of proof which will be equivalent to the proof-load required upon a given span.

When attention is paid to this plan much expense may be saved if the number to be proved is small, because the machine can be set for 10 feet without any loss of time, whereas considerable time is necessary to put the heavy wing-beams in position, and also to remove the same afterwards; therefore, there is no occasion to incur the necessary outlay for a larger span so long as the work may be equally well proved at a less span--of 10 feet or under. When a number of girders or rolled-joists are sent for proving at the same time, the larger span, if desired, is employed, without appreciably adding to the cost per test.

Apparatus for the application of distributed-loads has been provided, but some of the details have not been finished, because the opportunities for using the same have not as yet been afforded.

TWISTING OR TORSIONAL TESTS.

This is effected upon the

The Elastic stress is determined for every specimen tested excepting wires. same lines as in the preceding kinds of tests; the amount of torsion or twist is accurately recorded, in decimal parts of a turn, at regular intervals or increments of stress, from the commencement till fracture occurs. Although the rates of torsion are recorded in full in the experiment books, they are not given in extenso in the reports, for the reasons already stated, but only at certain intervals. (Refer for illustration to Reports H, J, K.) It will be observed from the woodcut that the specimens are double-ended, so that the uniformity or reverse of the material is very clearly demonstrated by the test. In some instances both halves of the specimen break together, or almost simultaneously, whereas in others one end will allow of being twisted as much as a half-turn after its neighbour has given way. The form of specimen at time of introduction was entirely new and is still unique.

The machine is so arranged that no bending action can come into play; specimens are permitted to lengthen or shorten under test-provision was made for ascertaining the end-movement, but it has not been necessary to make use of such. Friction is obviated, and therefore minute variations in amount of stress are recorded with precision.

Specimens are prepared on the premises to standard sizes and forms, the sizes being definite areas, and the lengths for torsion a definite number of diameters. The largest diameter tested is 3 inches, but the size depends upon the nature of the material.

N.B. When the machine was invented, provision was made for a very large twisting-yoke, in addition to the preceding. The drawings for which were completed, but the yoke, fortunately, was not made. During twenty-five years there has not been a single demand for the use of such. The machine, however, is perfectly ready for the reception of the large yoke, which can be fitted at any future time. The diameters of specimens might then reach 10 inches, and the length 250 inches.

SHEARING TESTS.

These tests are useful for many engineering details, such as pins for bridge-links, gib and cottar connections, connecting bolts, &c. Appliances are provided for regular standard sizes, which are each for definite sectional area. Specimens are usually tested in double-shear, as shown in woodcut. The amount of stress required to shear the specimen, and also the amount of detrusion at time of fracture, is ascertained and stated in the official report. (For example refer to Reports G, I.)

PUNCHING TESTS.

Some interesting experiments have been made in punching holes, but there is scope and need for far more extended investigations to find the best of various forms of punches, the difference due to changing the amount of clearance in dies or bolsters, and the effects produced on material of various grades and thicknesses through punching by the respective modes. The woodcut merely shows the nature of the test.

BULGING TESTS.

In this kind of test the facts ascertained comprise the amount of stress required to force the specimen, which is a disc 12 inches in diameter, through an aperture of 10 inches diameter, and also the amount of bulge produced in the process, or up to the time of cracking in experiments where the specimen has failed to pass through the aperture intact.