extension become more nearly equal; they are exactly equal for strains of 4842 pounds per square inch, while for greater strains the amount of extension from tensional strains is greater than the amount of compression from the same compressive strain. With wrought-iron and mild steel the compression under small strains is even in a much more marked ratio than under small tensional strains owing to the "Mowing" of the metal under compressive strains. There are no reliable experiments recorded, however, on which any formulae could be based for wrought-iron, it may be safely assumed, however, to be perfectly elastic up to 18000 pounds, (per square inch), compressive strain and 25000 pounds, (per square inch), tensional strain. With heavier strains the shortening or lengthening will be in very much quicker ratio than the increase in their respective strains. With steel it will depend upon its nature. Mild steel will approximate nearly to the perfect elasticity of wrought-iron, under safe strains, while hard steels will become more imperfect in elasticity the nearer they approximate the nature of cast-iron. Time plays a very important part in considering the final effect on metals of any strain.

It has been found that a moderate strain on a bar, if left on a long time, will gradually increase its effect, either extending it more and more as time passes on, or shortening it more and more. This is known as the "fatigue" of the metal. If the strain is within the elastic limit the increase in "set" will continue for a long time, but will finally become so infinitesimally small, as to be practically nothing. If, however, the strain is beyond the limit of elasticity the set will continue to grow and the piece will frequently break after many years under a strain which was considered well within the original ultimate breaking strength, but which, as time passed, fatigued, that is tired out, the material. Severe strains should therefore never be borne for a lengthened period, but should only be imposed for short periods, and they should be well within the elastic limit, to give the metal a chance to recover from the set produced by such strains.

Oft-repeated strains, even where put on entirely without shock, greatly weaken a material. It has been found that a load, which originally was borne with perfect safety, will, if often removed and replaced, finally break the piece.

Time Tests.

Intermittent

Strains.

Planat concludes by comparing many experiments on fatigue of iron: that if a strain of 58700 pounds, per square inch, can be repeated 170000 times before breaking a test piece, the same piece will break under a strain of only 48000 pounds, per square inch, if repeated 480000 times; or under a strain of 42700 pounds, per square inch, if repeated 1320000 times; or under a strain of 38400 pounds, per square inch, if repeated 4035000 times; while a strain of 32000 pounds can be repeated forever and would never break the piece.

These experiments were made by A. Wohler and confirmed by Spangenberg, Baker, Baushinger and others: they are usually carried out by means of revolving wheels or by pistons which alternately put on and remove the strain, the number of revolutions or pressures being carefully computed. Piston-rods of engines offer good examples of oft-repeated off-and-on loads, or the driving-rods of locomotives, in fact many parts of machinery.

Where a load is put on suddenly, with shock, that is falls on, or is forced on, suddenly without regard to whether it jars or not, the effect is practically double that of a stationary (static) or dead load. This is called "impact." The subject is too extensive to enter into here, the actual effect on the beam depending upon the length of span, height of fall, and permanent load; but it will be safe enough in all such cases to simply double the factor-of-safety, or in other words allow only one-half of the safe-stress, that would be allowed for an intermittent load put on without shock. Rolling loads are considered the same as loads put on with shock; the effect depending, of course, on the velocity; all such loads, moving, rolling, jarring, etc., are called dynamic as opposed to static or dead loads.

If the strain is reversed at each application, that is alternately tension and then compression, the effect is equal to the sum of both, that is, double that of the same amount of strain if not reversed. To break anything we instinctively bend it one way and then the other, by doing this we reverse the strains in the fibres from compression to tension and vice-versa at each bend, the result being that the piece soon breaks.

Box has combined these rules in a Table, which is here given as Table XXXII.

Impact or Rolling loads.

Reversed

Intermittent Strains.