620 



calibration of the magnetic tield. This calil)ration may be estimated 

 to be accurate to a tew thousands. The deviations must be caused 

 by the circumstance that at large and at small values of //„ tiie 

 |)roportionality mentioned may not be expected to hold '). 



TABLE I. 



1) In fig. 2 the points for the higher field strengths show the same kind of 

 deviation from the Langevin curve at 4°, '25 K. as at 1°,9 K. In my opinion this 

 fact is caused by the absence of proportionality mentioned in the text. 



