a 1-in. bar will support just before breaking, a term called the "ultimate strength." It has been found that the shape of the test bar and its method of loading has some effect upon the results, so it is now usual to turn a rod 5½ in. long down to 0.505 in. in diameter for a central length of 2-3/8 in., ending the turn with 1/2-in. fillets. The area of the bar equals 0.2 sq. in., so the load it bears at rupture multiplied by 5 will represent the "ultimate strength" in pounds per square inch.

Such a test bar is stretched apart in a machine like that shown in Fig. 9. The upper end of the bar is held in wedged jaws by the top cross-head, and the lower end grasped by the movable head. The latter is moved up and down by three long screws, driven at the same speed, which pass through threads cut in the corners of the cross-head. When the test piece is fixed in position the motor which drives the machine is given a few turns, which by proper gearing pulls the cross-head down with a certain pull. This pull is transmitted to the upper cross-head by the test bar, and can be weighed on the scale arm, acting through a system of links and levers.

Thus the load may be increased as rapidly as desirable, always kept balanced by the weighing mechanism, and the load at fracture may be read directly from the scale beam.

This same test piece may give other information. If light punch marks are made, 2 in. apart, before the test is begun, the broken ends may be clamped together, and the distance between punch marks measured. If it now measures 3 in. the stretch has been 1 in. in 2, or 50 per cent. This figure is known as the elongation at fracture, or briefly, the "elongation," and is generally taken to be a measure of ductility.

When steel shows any elongation, it also contracts in area at the same time. Often this contraction is sharply localized at the fracture; the piece is said to "neck." A figure for contraction in area is also of much interest as an indication of toughness; the diameter at fracture is measured, a corresponding area taken out from a table of circles, subtracted from the original area (0.200 sq. in.) and the difference divided by 0.2 to get the percentage contraction.