However, notwithstanding the fact that theoretically the test specimen should be higher than broad, it is quite the universal custom to determine the crushing strength of stone by testing cubes; and this practice is likely to continue so that the results may be comparable with those hitherto obtained and published. Theo retically the strength of a cube of stone is about 9 per cent greater than that of a prism one and a half times as high as broad.
Size of the Oabe. Although the cube is the form of test specimen generally adopted, there is not equal unanimity as to the size of the cube; but it has been conclusively proved that the strength per square inch of bed area is independent of the size of the cube, and therefore the size of the test specimen is immaterial. A two inch cube is most frequently used in compression tests.
General Gillmore, in 1875, made two sets of experiments which he claimed proved that the relation between the crushing strength and the size of the cube can be expressed by the formula in which y is the total crushing pressure in pounds per square inch of bed area, a is the crushing pressure of a 1-inch cube of the same material, and x is the length in inches of an edge of the cube under trial. For two samples of Berea (Ohio) sandstone, a was 7000 and 9500 lb., respectively.* But the testing machine was too crude, the experiments were insufficient in number, and the cubes were too small and too nearly the same size to establish any such law. Results by other observers with better machines, particularly by General Gillmore himself f with the large and accurate testing machine at the Watertown (Mass.) Arsenal,$. uniformly show this supposed law to be without any foundation. Unfortunately the above relation between strength and bed area is frequently quoted, and has found a wide acceptance among engineers and architects, notwithstanding the fact that it is not true.
The position of the test specimen with reference to the bedding planes of the rock jn its native position has a very important relation to the strength of the stone. The direction in which a stone splits
most easily is called the rift, and the next easiest the grain, while the direction in which the resistance to splitting is the greatest is called the head. The test specimen will be the strongest, if the pressure is applied perpendicular to the rift. In most cases the rift is hori zontal, i.e., is parallel to the natural bed; but in some cases the rift makes an angle with the natural bed. Horizontally bedded stones are quarried by blasting or wedging along the natural lines of cleavage, or by channeling and wedging along the laminations; but if the rift is not horizontal, it is common practice to cut out blocks without reference to the natural seams, since the quarrying machines run most easily upon horizontal tracks, and since it is desirable to main tain a level quarry floor. In the last case, then, there is a difference between the rift and the "natural bed"; and in preparing test specimens care should be taken to have two sides of the cubes par allel to the rift, which should be marked so that the pressure may be applied on these faces. It is not sufficient to have the faces of the cubes parallel and perpendicular to the broadest face of the original block, for the latter may not have been cut out with reference to the rift. The rift can be determined by a careful examination of the block. The failure to apply the pressure perpendicular to the rift doubtless accounts for part of the large difference in strength of different specimens found by most experimenters.
If the specimen is dressed by hand, the concussion of the tool greatly affects its internal conditions, particularly with test specimens of small dimensions. With 2-inch cubes, the tool-dressed specimen usually shows only about 60 per cent of the strength of the sawed sample. The sawed sample most nearly represents the conditions of actual practice. Unfortunately, experimenters seldom state whether the specimens were tool-dressed or sawed. The disintegrat ing effect of the tool in dressing is greater with small than with large specimens. This may account for some of the difference in strength of different sizes of test specimen as seems to be shown by some experiments.