623 



So the values given in table II have been found. 



d. Corrections for diamagnetism of Ihe liquid bath and of the 

 anion could be left out of consideration. 



e. xAs regards the accuracij and Ihe sources of error maj firstly 

 be pointed out that the kelhnntemperatures are rather uncertain, 

 especial I}- the lower ones. There was no room for a special stirrer 

 and so the liquid could be stirred only so much as was possible 

 by moving the floating system up and down. Therefore probably 

 the temperature was not always evenly distributed and not perfectly 

 well defined. This is especially important at temperatures below the 

 maxiniuni of density; then the cooling at the surface by evaporation 

 does not give rise to downward convectional currents. However the 

 lower temperatures are not only somewhat indefinite, but the values 

 accepted are not very accurate. They have been determined graphi- 

 cally by means of the total existing material for helium vapour 

 pressures'), but this leaves at the temperatures between 1° and 3° K. 

 uncertainties of the order of 0,1 of a degree. 



TABLE n. 



ƒ. Much care was bestowed on the adjusting of the sample to 

 the proper place in the magnetic field, or more accurately, of the 

 adjusting of the magnet to the sample, the cryostat not being movable. 



') H. Kamerlinoh Onnes and Sophus Webek, these Proceedings 18, p. 493; 

 Leiden Comra N^. 147b ; H. Kameklingh Onnes, Leiden Coram. N". 159 p. 35. 



