Tests on large timbers are considered necessary to furnish designers data on the probable strength of the different sizes and grades of timber on the market; their coefficients of elasticity under bending (since the stiffness rather than the strength often determines the size of a beam); and the manner of failure, whether in bending fibre stress or horizontal shear. It is believed that this information can only be obtained by direct tests on the different grades of car sills, stringers, and other material in common use. When small pieces are selected for test they very often are clear and straight-grained, and thus of so much better grade than the large sticks that tests upon them may not yield unit values applicable to the larger sizes. Extensive experiments show, however, (1) that the modulus of elasticity is approximately the same for large timbers as for small clear specimens cut from them, and (2) that the fibre stress at elastic limit for large beams is, except in the weakest timbers, practically equal to the crushing strength of small clear pieces of the same material.[57] [Footnote 57: Bul. 108, U. S. Forest Service: Tests of structural timbers, pp. 53-54.] MOISTURE DETERMINATION In order for tests to be comparable, it is necessary to know the moisture content of the specimens at the zone of failure. This is determined from disks an inch thick cut from the timber immediately after testing. In cases, as in large beams, where it is desirable to know not only the average moisture content but also its distribution through the timber, the disks are cut up so as to obtain an outside, a middle, and an inner portion, of approximately equal areas. Thus in a section 10" x 12" the outer strip would be one inch wide, and the second one a little more than an inch and a quarter. Moisture determinations are made for each of the three portions separately. The procedure is as follows: (1) Immediately after sawing, loose splinters are removed and each section is weighed. (2) The material is put into a drying oven at 100° C. (212° F.) and dried until the variation in weight for a period of twenty-four hours is less than 0.5 per cent. (3) The disk is again carefully weighed. (4) The loss in weight expressed in per cent of the dry weight indicates the moisture content of the specimen from which the specimen was cut. MACHINE FOR STATIC TESTS The standard screw machines used for metal tests are also used for wood, but in the case of wood tests the readings must be taken "on the fly," and the machine operated at a uniform speed without interruption from beginning to end of the test. This is on account of the time factor in the strength of wood. (See SPEED OF TESTING MACHINE, below.)