POISSON'S METHOD. 603 



magnetisation, may be represented by M(i+/H); if/ is the term 

 of correction for the distance R (1155), we have 



and the value of p will be determined by two experiments made 

 at different distances. We deduce from this 



ME 7 R 3 



~H~T~ 

 and consequently 



" B' R 3 



The component H is then expressed as a function of the data 

 of the experiment, and will be independent of the magnetisation 

 induced in the bar, if it may be assumed that the coefficients /and <f> 

 are sensibly equal. 



Corrections for temperature relative to the distance R and the 

 moment of inertia of the bar, are not of great importance ; for 

 they are of the order of expansion of the metals that is to say, 

 less than 0*0002 for a variation of 10. 



In the case of comparative measurements, it is useless to know 

 the moment of inertia of the bar, and the distance R, provided 

 the quantities are the same in all the experiments. 



1175. METHOD OF GAUSS. Poisson had thus established the 

 principles of the method ; but the use of oscillations has the in- 

 convenience of only giving the ratio of the fields F and H by the 

 difference of the squares of the numbers observed, and does not 

 carry with it great accuracy. The determination of the term of 

 correction /, in particular, will be obtained by an expression con- 

 taining differences of the numbers of oscillations, which are difficult 

 to estimate exactly. 



Poisson's method was very rarely used ; it seems even to have 

 been little known when Gauss* pointed out an experimental arrange- 

 ment easier of application, which since then has been generally 



* GAUSS. Intensitas vis Magnet., Comm. S. R. Getting., Vol. viu. 



