846 



are preseiir, one slioulii expect the resiill of iiiakiiig settings by 

 adjusting beats to zero and by adjusting beats to a fixed number to 

 be different. The resnits however indicate tiiat this efïect is alisent. 

 Thus J 3.2 and 13.1 are I lie readings obtained on the scale aS' by 

 the two methods. 



_(/. CaVihrdtion. Since all the ciianges in the frequency are very small 

 the change in the frequency is very nearly proportional to the 

 change in capacity or to the change in inductance that causes it. 

 Therefoie changes in A' necessary to compensate for two different 

 changes in inductance are proportional to these changes. If one 

 change in iiuiiH-lance is known or if its meaning a.s a susceptibility 

 is known the other is derived by multiplication into the ratio of 

 the two settings of K. This is the principle of the calibration em- 

 ployed. The calibration divides itself into the following parts: 



a. To produce a change in inductance which has a direct inter- 

 pretation as a susceptibility. 



For this purpose two glass tubes were attached to the paramag- 

 netic sample at opposite sides of a diameter — while the sample 

 was in the warm condition. Copper wires accurately drawn and 

 measured could be inserted into these Tiie length of the wires was 

 nearly equal to the length of the column of paramagnetic substance 

 employed. Roughly the wires may be said to exclude the high fre- 

 quency magnetic field from their interior. To a first approximalion 



1 



they are therefore equivalent to a material of .susceptibility . 



If the positive effect of a paramagnetic sample is equal to the effect 

 of a wire of a certain size, its susceptibility must be then equal to 



— times the ratio of the volume of the wire to the volume of the 

 4jï 



substance. Since the field is not quite excliuied from the interior of 



the wire, its diamaguetic action is not quite as large as we have 



just supposed but a correction for this may be applied. Taking the 



field to be a homogeneous one along the axis of the wire the 



correction factor is — f^i ii {q) -\- i where 



2 ber q bei' q — ber' q bei q 



ber^ q -\- bei'' q 



where q = 1/ «, ber and bei are the Kelvin functions, and 



ft, <T, « are respectively the permeability, resistivity, and radius of 



tu 



the wire used at the frequency — . 



