62 



K. TSÜRÜTA. 



been led to conclude that the problem of comparing the thermo- 

 electric E. M. F. in different specimens of iron or in the same speci- 

 men under different conditions, is not to be attacked, even though all 

 other questions have been cleared up, till I shall be in possession of a 

 fuller knowledge of that relation, which, as I have convinced myself 

 by experiments, has indeed a very intimate connection with my present 

 subject of investigation. 



§ 35. Steel was the subject of the next experiments, one of 

 which is o-iven in the following: table. The wire was 0*5 m.m. in 

 diameter and was tested after being well annealed. 



Examining the curve in Fig. 4, PI. IV, in which the above 

 numbers are graphically represented, we see that the effect of the 

 thermo-electric hysteresis is here quite apparent, causing the " off " 

 branch to deviate from the "on." After a few steps of loading, the 

 initial gradient of the " on " branch becomes smaller, though very 

 little, and under heavier loads the curve runs almost straight. As to 

 the " off" branch the gradient becomes appreciably less and less as the 

 unloading goes on, and at a somewhat quicker rate near the close of 

 unloadiug. The "on " branch seems to show something like a point 



