270 SCIENCE PROGRESS 



The explanation of these differences lies in the alinement 

 of brittle impurities in the steel as a result of the rolling. 

 Sulphur and phosphorus, present as manganese sulphide and 

 as a solid solution of iron phosphide respectively, are present 

 in long-drawn-out threads elongated in the direction of the 

 forging. 



The presence of slag in the steel, which will also be drawn 

 out in the direction of the rolling, deleterious as it is from nearly 

 every point of view, may yet exert a considerable influence in 

 raising the resistance of the material to the impact test, provided 

 that the test-piece be cut in a longitudinal direction. As the 

 crack formed reaches a slag inclusion, which is perpendicular 

 to the direction in which it is growing, a tendency exists for a 

 change of direction, the crack travelling along the line of the 

 slag. Oriented in this way, slag fibres give an increased resis- 

 tance to impact. If, however, as in a transverse test-piece, the 

 slag lies parallel to the crack, it merely greatly facilitates its 

 passage through the material, very low impact values being 

 obtained. This point is very clearly dealt with by Mr. H. 

 Brearley (Journ. Manchester Association of Engineers, 191 7, 

 pp. 492 et seq.). 



The greatest triumph of the impact test has been the revela- 

 tion of a new and quite unexpected type of brittleness to which 

 certain alloy steels, especially those containing nickel and 

 chromium, were liable. Tempered at temperatures around 

 6oo-65o°C, and slowly cooled, the resistance of the steel to 

 impact may be almost non-existent, while when rapidly cooled 

 from the tempering temperature, the resistance is excellent. 

 This phenomenon, usually known as " temper-brittleness," is 

 frequently not revealed at all by the tensile test, as the following 

 figures, due to Mr. Brearley (loc. cit.), clearly show: 



The two sets of tests were carried out on the same material 

 similarly treated, except that sample A was slowly cooled in the 

 furnace after tempering, B being quenched in water. Although 

 the tensile test is to all intents and purposes unaffected, the 



