( 819 ) 
c* - y (ivr-iv.) . (5) 
Thus just as if there were no correction for inhomogeneity the 
second side of the equation remains always proportional to the 
square of the field. Even without knowing that correction, if is 
itself a constant we should be able to deduce from the observations 
whether K is a function of the field or not. We see, however, that 
the constancy of f? requires that of A, i.e. that the field must remain 
homothetic no matter how great it should be. Now this is not the 
2<f 
case as can be seen from the quotients —- in tables V, VII, and 
H * 
VIII. Table V shows first an increase, then the quotient reaches a 
maximum and diminishes considerably; tables VII and VIII shew a 
change in exactly the opposite direction; this is just what one would 
expect if 0 were variable and K constant, for the tables refer to 
two .practically identical bodies, of which the one is dia- and the 
other para-magnetic. Now in either case the fundamental couple 
(uniform field) is in the same direction while the couple due to 
inhomogeneity changes sign with the susceptibility; should, therefore, 
the correction in the one case first increase and then decrease, it 
must in the other case first decrease and then increase. We shall 
return to this point in $ 4. 
Since this determination aims only at relative measurements, we 
have once and for all taken as the value of the susceptibility of 
oxygen at —183° C. the value that was given by the improved 
apparatus for measuring the magnetic rise. With the help of this 
value we have calculated the values of $ for each field from equation 
5): (see tables V and VI). These values fall pretty well on a curve of 
means. Finally the susceptibility at the lower temperatures is calcu¬ 
lated by means of the value of £ as a function of the field given 
by this curve. We shall take the opportunity of the corresponding 
series of observations to make some remarks upon the influence of 
the inhomogeneity for each of the three pole-gaps that were used. 
2. The inconstancy of the magnetization as a function of the 
azimuth. The general expression for the couple in a uniform field 
(JVj —N t )I*v sin rpcostp 
only reaches its maximum value just at <p = 45°, and consequently 
sin rp cos r = l U 8ince 1 remains constant during the torsion. Here 
again we see "a fundamental difference between the application of 
this method to the investigation of saturation magnetization and to 
that of a body of constant susceptibility. It is clear that in the first 
case the condition 1 = constant is, as it were, fulfilled by definition. 
