CHANG-E OP RIGIDITY BY MAGNETIZATION. 5 



is evidently not due to the twisting of tlie rod, l)nt to its bending 

 by magnetization. Tlie case corresponds to the experiment of 

 Guillemin described in the preceding paper. Tlie nature of the 

 deflection and its amount coincided with the change of elasticity by 

 magnetization, which we have already studied. 



The steel pivot was then slightly brought in contact with the 

 agate cup ; if this contact was made in a suitfible degree, the deflec- 

 tion due to magnetization, when acted on by no twisting couple, 

 could be made negligibly small. In case the deflection could not be 

 sufficiently reduced, it was always corrected for. The contact being 

 so adjusted, a couple was applied b)^ suspending a weight. The 

 tension of the copper wire was next adjusted and the w^orking of the 

 apparatus tested by adding successive!}^ weights of 1, 10, 50 grams to 

 the pan. If the deflections of the mirror were proportional, the 

 adjustment was considered to be correct. 



To begin with, the ferromagnetic rod was demagnetized by 

 reversals, and then a current passed, taking tlie deflection as soon as 

 possible. These processes were repeated with successively increasing 

 currents. In order to prevent minute oscillations of the mirror, the 

 thin copper ware and the mirror should be protected from the air 

 currents. 



The resistance of the magnetizing coil was only 0.6 i?, so that the 

 heating of the core due to current was negligibly small up to the 

 strongest current used in the present experiment, and the creeping of 

 the imao^e of the scale was not at all observed. But we were careful 

 to read the deflection as quickly as possible. 



Since the couples corresponding to 1, 10, 50 grams produced 

 torsions proportional to their respective weights, the friction at the 

 pivot did not seem to disturb our results. That the observed deflec- 

 tion was really due to a rotation, but not t() the depression or eleva- 



