789 



is therefore more limited by the accuracy of the ammeters, than by 

 that of the apparatus itself. With two of tlie ammclers, in eacii of 

 which we constantly adjusted on one of the scale divisions, it was 

 possible to bring the relative sensibiHty to within a thousandth. The 

 third was almost equally accurate. In the relative measurements of 

 the susceptibility calculated, we can thus rely in general upon an 

 accuracy of 0,2 to 0,3 7o- '^ he absolute accuracy suffers somewhat 

 from the fact that we are not certain of the homogeneity of the 

 experimental object (see above); but it is principally limited by the 

 accuracy of the determination of the magnetic ticld, the square of 

 which occurs in the formula for the susceptibility, and not by that 

 of the apparatus. 



There is one more important factor to be considered, viz. the 

 constancy of the temperature of the bath. Disturbances in this will 

 have a different effect with different substances, as the change in 

 the susceptibility with the temperature is very unequal for different 

 substances, especially when we approach the absolute zero. 



In our experiments in general we did not observe any disturbances 

 from irregularity of temperature. Only in the case of gadolinium 

 sulphate, the susceptibility of which changes most with the temperature 

 (1 7o for 0,2 degrees at 20° K.) the measurements in different fields 

 (at different moments therefore) did not agree so well with each other 

 as might have been expected if we only considered the accui-acy 

 of the magnetic determinations. 



(February 26, 1914). 



