IRON AND STEEL UNDER CYCLICAL VARIATIONS OF STRESS. 41 



stress during the 10th cycle. The readings of the right-hand mirror had changed 

 679 divisions. Adding the two changes together, it appeared that the specimen had 

 increased in length by 1339 divisions. The "permanent extension" at 5330 repe- 

 titions is therefore 1339. 



Obviously, all the readings might have been taken at the minimum stress, when 

 the "permanent extension" would have appeared as 1304, or 35 divisions less. This 

 difference is, of course, equal to the " cyclical permanent set " at the time. It does 

 not appear to be possible to further separate the " cyclical permanent set " and the 

 " permanent extension," but, fortunately, this is not important. 



It has been usual in the experiments to keep the maximum stress constant and use 

 - the readings obtained at that stress for the calculation of " permanent extension." 

 The only exceptions occurred in the cases when cyclical variations of stress between 

 equal and opposite limits of stress were being observed. In these cases both loads 

 were increased numerically by equal amounts, the mean stress being always zero. The 

 mean of the extreme extensometer readings was then found to be practically constant. 



It will be readily recognised that the method of approaching the limiting conditions 

 by increasing the range from a small value is liable to be very costly and would only 

 be justified if the supply of material were limited. It is usual, in making fatigue 

 tests to destruction, to find any range which will break the specimen quickly, and 

 then, decreasing the range for a new experiment, to gradually approach the limiting- 

 range. This may often be obtained from three or four cases of fracture by slight 

 extrapolation. In a similar manner the observations in the present paper refer 

 principally to cycles of stress greater than the limiting value for safety. 



The Results of the Experiments on Axle Steel. The observations are given 

 graphically in figs. 1 to 5. Each figure corresponds to a different position of the 

 elastic limits of the material, and the scales for corresponding curves have been 

 maintained constant throughout the series, so that direct comparison can be made 

 from one figure to another. 



Equal Tensile and Compresnive Stresses (fig. 1). For equal stresses the obser- 

 vations were of the simplest kind. No permanent extension occurred and the 

 hysteresis loop was quite closed. The changes in the specimen showed themselves 

 entirely by the production of " cyclical permanent set," i.e. by an increase in the 

 extension during each cycle. 



After the specimen had been fixed in position, and before it had been loaded in 

 either direction, a reading was taken of the unstrained length. A similar reading 

 was recorded after the tension load had been applied and removed, and a third 

 reading after putting on and removing the compression load. The three readings 

 were alike and indicated complete elasticity, within the accuracy of measurement. 



The stresses 14'1 tons per sq. inch were then repeated automatically, and for 

 some time the straight line O in fig. 1, c, continued to represent the cycle of extensions. 

 The cross indicates the origin of both extension and stress. 



VOL. cox. A. G 



