THE THERMO-ELECTRIC EFFECTS OF LONGITUDINAL STRESS IN IRON. 43 



shaken out, the representative point in the diagram rose considerably. 

 Instead of unloading as before, loading was began, continued up to 

 7 kilos., and then reduced to zero again. The course of the curve is 

 here again exactly the same as that we had in the minor cyclic curves 

 without agitation. 



§ 18. An examination of all the experiments enables us to 

 understand how the thermo-electric changes set in and go on when 

 the wire is brought to a certain stable condition by tapping. Indeed 

 they give us a confirmation of a remark of Prof. Ewing's/ which it 

 will be well to quote here : 



" The thermo-electric quality of a piece of iron which has been 

 shaken into a condition of molecular stability changes, first, towards 

 (say) positive, and later, towards negative, when stress is applied and 



increased And the changes have the reverse signs from those 



justed stated if the direction of the applied 

 stress is reversed. Thus, let a be a point 

 expressing the relation of thermo-electric 

 E. M. F. to load, which is reached by 

 tapping the metal or by exposing it to the 



Load process of demagnetising by reversals. If 



from this point we go on loading we have 

 the curve ab. If from the same point we had began to unload we 

 should have had the curve ac" 



§ 19. One of the results stated by Prof. Evving, which I could 

 not verify in the course of my experiments, is the formation of a 

 small loop in the cyclic curve when we turn from loading to unload- 

 ing at the uppermost limit of loading. In my experiments the " off" 

 branch usually goes nearly straight below the " on " and gets appre- 



(1) 1. c. § 46. 



