CORRECTION FOR MAGNETISATION INDUCED BY THE EARTH. 609 



serving that the difference of the deflections is very small, we 

 deduce 



H sin a -sin a' a -a' 



Z sin a + sin a' 2 tan I tan a ' 



By a series of reversals the want of symmetry of the magnet 

 may be eliminated 



With a series of magnets Lament obtained for this term of cor- 

 rection fH. very different values, from 0*00053 to 0*00198. We 

 shall subsequently see that the coefficients / and </> have not the 

 same meaning. 



1181. Joule* has pointed out a method which enables us to 

 eliminate the effect of induced magnetisation in the observation 

 of oscillations. Two nearly identical bars M l and M 2 are taken. 

 They are arranged parallel to each other, and at such a distance 

 that the action of one on the centre of the other is sensibly equal 

 to the terrestrial field. Each bar being then in a field which is 

 almost null, only retains its rigid magnetisation. In this case the 

 oscillations of the system only depend on the sum of the mo- 

 ments Mj and M 2 , or of the directive couple 



(M 1 + M 2 )H = A. 

 The oscillations of the two magnets separately give the couples 



We deduce from these last two experiments the ratio 



M 2 ~A 2 m > 

 consequently 



M W^A_^_ 



i + m 



M H 



i+m 



1182. COMPARATIVE DEFLECTORS. For travelling instruments 

 it is unnecessary to know the distance of the deflecting bar from 



* JOULE. Proceedings of the Manchester Lit. and Phil. Society, Vol. vi., 

 p. 129. 1867. Scientific Papers, Vol. I., p. 561. 

 VOL. II. R R 



