ON MAGNETOSTRIC'JTON. 



19 



From the above table, we gather the fact that the magnetostric- 

 tion plays no important part in the use of nickel steel scales ; only 

 in measurements of extreme accuracy, it will be necessary to add a 

 very small factor of correction to the measured values, according as 

 the scale is placed in the magnetic meridian or perpendicular thereto. 

 As will be seen from the curves of elongation (Fig. 4), the difference in 

 a metre will generally be less than -r--/^ for measurements made in the 

 said di recti (Mis. 



One distinct featm-e of the curves of elongation is the effect of 

 annealing. In both 45% and 35% Ni, the wire elongates several 

 times more in the annealed than in the hard drawn state, so that in 

 cases when the change caused by magnetization is to be feared, we 

 shall be able to eliminate the errors due to magnetostriction in a con- 

 siderable degree by using the unannealed metal. 



§ 5. Effect of Longitudinal Stress on Magnetization 

 and the Reciprocal Relations. 



A remarkable feature of magnetostriction is the reciprocal rela- 

 tion between the strain caused by magnetization and the effect of 

 stress on magnetization. AYe have already examined the different 

 changes from this stand-point for iron and nickel, and found that the 

 relations between strain and stress are generally reciprocal in these 

 two metals. In the present experiment, we made special examination 

 into the effect (^f longitudinal pull on the magnetization of cobalt and 

 nickel steels in the same li<2:ht. 



The annexed dia^ji'ram shows the scheme of arranfi-ement for 

 examining the effect of pull on magnetization. A wooden lever LL, 

 furnished with a knife edge, rested horizontally against a vertical 

 pillar PP. The lever arm held a brass rod B, hang by a hook H ûom. 

 the ceiling; a cobalt cylinder C (length 27.3 cm., diameter 1.3G cm.) 



