MI: .1 \irii; ON TIN: i;i:i <>\I.I;N OF IKON I I:M UVI-:I;STI:.\IN -j| 



\\ hidi completed the large cycle of loading. The hysteresis exhibited by a cycle such 

 as this may lx- represented numerically by expressing the breadth of the cycle at any 

 stress, as a fraction of the total elongation of the specimen. If this be done, the 

 hysteresis, in tin- relation nf strain to stress, which recently overstrained iron has JIM 

 U-en shown to exhibit, may IM- compared \\itli that observed with ordinary material l.v 

 Professor EWIM; in experiments on very long wires.* In Professor KWIXQ'S experi- 

 ments, the wires were subjected many times to a certain range of stress, and the 

 extension at half the range was observed l*>th as the load was applied and as it was 

 removed. The latter extension w;is found, due to hysteresis, to be greater than the 

 former ; and the difference being expressed as a fraction of the extension produced l>y 

 the maximum load, values were obtained ranging from ^ l a in the case of high carbon 

 steel, to T Jff in the case of an iron wire in the hard-drawn state, or j^j with mild 

 steel wire annealed and then hardened by stretching. The hysteresis shown at half 

 the range of stress in the cycle described above (Curve 8, Diagram VI.) is about -^ of 

 the extension produced by the maximum load of 40 tons per square inch. 



In order to see how far the hysteresis in the relation of strain to stress exhibited by 

 recently overstrained iron is statical in character, or how far it depends on the rate of 

 l'.-iiling, a cycle was performed allowing ten minutes to elapse after the addition of 

 every 4 tons of stress. The only effect was to produce a series of little notches in 

 the curve obtained, similar to the notch shown at the stress of 34 tons in Curve 7, 

 1 Magram VI. The area of the cycle was thus not appreciably affected. The time 

 allowed after the addition of every 4 tons was ample so far as the amount of creeping 

 was concerned, as is clearly shown by the creeping at the stress of 20 tons in Curve 7, 

 Diagram VI. If a much longer time had been allowed to elapse, then recovery of 

 elasticity would have taken place as in Curves 3 and 7, Diagram VI., and the question 

 of the static character of the hysteresis would have become complicated. 



The back creeping which occurs after the removal of the load from a specimen 

 which has been several times overstrained (for example, the creeping shown to have 

 occurred during ;" minutes in Curve 8, Diagram VI.) is not simply due to the 

 immediately preceding loading, but to all previous loadings. It was often observed 

 that if sufficient time were allowed to elapse after the removal of a load, the zero 

 reading would become negative. That is, the bar would become shorter than it was 

 before the loading was commenced an effect which is no doubt to be ascribed t<> 

 previous overstrains, and is analogous to phenomena which have been observed in the 

 residual charge of dielectrics, and in the torsional strains of glass and other materials. 



Before leaving this section of the paper, attention should be called to the close 

 analogy between the hysteresis effects shown in Diagram VI.. and the known 

 characteristics of magnetic hysteresis in iron.t 



* " On Hysteresis in the Relation of Strain to Stress," ' B.A. Report,' l*s;i. |>. 502. 

 t EWINO. " r.\| 'imi'iital Researches in Magnetism," Phil. Trans.,' lM, or Intok on 

 Induction in Inni and nthi-i Metals." 



