alternations of stress which is required to rupture when the stress is applied 144 times per minute:

Ewing and Humfrey have subjected Swedish iron, with a breaking stress of 23'6 tons per square inch, to a series of pushes and pulls, 9 tons in magnitude, repeated 400 times per minute. On examination it was found that fine slip-bands appeared in a few crystals after a few-say 5000-reversals of stress, as shown in Fig. 58. With a greater number of reversals -say 40,000—the slip-bands increase in number, and those which first appeared broaden and develop into small cracks, as shown in Fig. 59. If the specimen be repolished, so as to clear off the slip-bands, the cracks alone become visible, as at A (Fig. 60). The crack, or flaw, gradually creeps right across the specimen when the number of alternations is still further increased, as shown in Fig. 61. Finally, the specimen breaks. Let me quote Ewing and Humfrey's own words :

"Whatever the selective action of the stress is due to, the experiments demonstrate that in repeated reversals of stress certain crystals are attacked, and yield by slipping, as in other cases of non-elastic strain. Then, as the reversals proceed, the surfaces upon which the slipping has occurred continue to be

surfaces of weakness. The parts of the crystal lying on the two sides of each such surface continue to slide back and forth over one another. The effect of this repeated sliding or grinding is seen at the polished surface of the specimen by the production of a burr or rough and jagged irregular edge, broadening the slipband, and suggesting the accumulation of débris. Within the crystal this repeated grinding tends to destroy the cohesion of the metal across the surface of the slip, and in certain cases this develops into a crack. Once the crack is formed, it quickly grows in a wellknown manner, by tearing at the edges, in consequence of the concentration of stress which results from lack of continuity. The experiments throw light on the known fact that fracture, by repeated reversals or alternations of stress, resembles fracture resulting from 'creeping flaw' in its abruptness, and in the absence of local drawing-out, or other deformation of shape." 1

The rupture of steel is not caused by the gradual growth of the crystalline structure of the metal under the influence of shocks and vibrations. The breakdown is due to fatigue. When fatigued, the metal breaks more readily. Again, when subjected to sudden shock, the metal has no time to "flow." The slipping of the crystal planes, or the plasticity of the metal, has no time to come into play. The metal, in consequence, appears to be abnormally brittle.

1 P. Kreuzpointer, Journ. Franklin Inst., 153. 233, 1902; J. A. Ewing, Nature, 70. 187, 1904; P. Breuil, Suppl., Journ. Iron and Steel Inst., 1904.