THE THERMO-ELECTRIC EFFECTS OF LONGITUDINAL STRESS IN IRON. 47 



greatest, as it probably is, the thermo-electric curve must have its 

 initial trends always in the sense dictated by that greatest effect. 



§ 23. Besides the initial trends, I have in the above figures 

 indicated the courses of the thermo-electric curve when the loading or 

 unloading is further continued. It will be seen that after passing 

 through a maximum or minimum point the curve goes over to lie 

 again in the fourth or second quadrant. Thus, the manner of forma- 

 tion of the main cyclic curve is repeated, almost everywhere in the 

 field of the diagram, although not so fully developed in some cases 

 as in others. This must have arisen from a complex change in the 

 mechanical constitution of the iron, arising from the interaction of the 

 effects of the hysteresis and stress. It will be of course premature and 

 quite speculative to try to make out in a precise manner what is that 

 change and how it takes place. 



§ 24. After all the previous experiments had been finished, I 

 made some others on iron wires in the hard drawn state in which they 

 had been purchased/ The general result obtained is, that although 

 in this case the thermo-electric curves are somewhat peculiar in form, 

 they still have almost all those features that were pointed out as 

 characteristic in connection with annealed soft iron wires ; in fact, 

 some features which had been formerly hard to discover are found 

 singularly well developed in these experiments. 



§ 25. Corresponding to the experiments of § 8, I have the 

 following observations which were made on iron wires 0'7 m.m. in 

 diameter and exceedingly soft when annealed. By trial I ascertained 

 that wires from the same specimen in the hard drawn state could be 

 loaded up to 36 kilos, and thus became appreciably more hardened 

 still. 



(1) Prof. Ewing, 1. c. § 42-44. 



