IRON AND STEEL UNDER CYCLICAL VARIATIONS OF STRESS. 47 



elasticity very rapidly, until finally it would reach the maximum tensile stress of the 

 material when the superior limit reached the same point ; the range being then zero. 



General Remarks on the Diagrams.- When a specimen was being fatigued by the 

 application of an unsafe range of stress, the hysteresis loop produced retained the 

 same general shape for all ratios of the maximum and minimum stresses. 



The two parts of the loop which were traced out, as the stress varied from either- 

 extreme of the cycle towards the mean stress, were straight lines with a slope equal 

 to that which would have been given by the specimen if restored to an elastic 

 condition. Further than this, no appreciable change occurred in the value of YOUNG'S 

 modulus. 



The increase in width of the hysteresis loop, for a given increase in the range 

 applied, was greatest for the case of equal and opposite stresses, and gradually 

 decreased in such a way as to suggest that at the maximum stress there was no 

 tendency to form such a loop. The ordinary tensile test only differs from this 

 extreme case in the fact that a short time only is allowed for fracture, whilst an 

 indefinitely long one would be required as an extension of the case of varying stress. 

 This difference is unimportant, as careful experiments have shown that the rate of 

 test has little effect on the maximum stress obtained. 



The rate of permanent extension due to excess range, after the adjustment of the 

 elastic limits, is not so clearly shown. It is zero for the case of equal stresses and 

 becomes very great in the case of failur-e under high maximum stresses. 



The Determination of the Elastic Ranges for the Materials from the Observations. 

 Before the numerical values of these are given one further point of interest must be 

 noticed. Some specimens of Swedish iron and axle steel showed a large " permanent 

 extension" which ultimately ceased. In all cases "cyclical permanent set" was 

 produced by the initial " permanent extension," the amount of which decreased as the 

 test proceeded and in some cases disappeared. It would therefore seem that an 

 extension produced by repetitions of stress introduced a temporary want of elasticity 

 during each cycle, and that a specimen which is, in reality, quite safe may appear 

 inelastic. 



In further support of this statement may be mentioned a specimen which has been 

 described elsewhere.* A sample of Swedish iron was subjected to reversals of stress 

 of 8'3 tons per sq. inch and +8'9 tons per sq. inch at 1200 r.p.m. Due to some 

 cause, possibly the inequality of the stresses, a hysteresis loop was produced after 

 100,000 reversals, and this remained even after a million reversals, the specimen being 

 then unbroken. After many months of rest the elasticity was completely restored 

 over the whole range. One million reversals at the slightly higher range of 18 '2 tons 

 per sq. inch would have produced fracture. 



It may possibly be that the decreased range of stress found by REYNOLDS and 

 SMITH has some relation to the question of recovery, but further experiments are 



* 'Report of the National Physical Laboratory,' 1907, p. 58. 



