90,S 



alone. To compensate the vacuinn caused bv the contraction of the 

 mixtnre by each decrease of temperature, we have each time added 

 pure 7iitro(/tm (stirring of course eacii time) ; our object was the 

 inv^estigation of the influence of the mean distance of the paramag- 

 netic molecules upon the magnetisation, and the process described 

 above apparently comes to the determination of the changes in the 

 magnetisation with the temperature, for every mixture at constant, 

 distance of the oxygen molecules. 



After the measurements the ^■apo^r products were collected, and 

 analysed with pyrogallic acid. This analysis is a useful check, but 

 cannot lay claim to great accuracy, as the comparison with the 

 synthesis assumes the knowledge of the total (juantity of condensed 

 nitrogen (including the amount added during the measnrements), which 

 quantity is for various reasons somewhat uncertain. Moreover it is 

 necessary that the vapour products should be very completely col- 

 lected, as a considerable weight attaches to what is vaporised last, 

 as being almost pure oxygen. On the other hand we may remark 

 that the deduction of the concentration (see above) trom the 

 synthesis presumes oidy an accurate knowledge of the loeujlit of 

 oxyyen and not that of the nitrogen, and therefore can claim a 

 greater degree of accuracy. 



The susceptibilities are expressed in absolute units by comparison 

 with that of pure liquid oxygen, which are previously measured in 

 absolute value by the method of rise ^); this calibration comes simply 

 to this, that a series of measurements are made under the same 

 conditions as the former, but with pure oxygen instead of with the 

 mixtures. Moreover, a calculation of the absolute values based upon 

 the values of the field according to the measurements made by Dr. 

 OosTERHUis in Leiden after the measurements, lead to results which 

 agreed with the experiments previously obtained, well within the 

 limits of experimental accuracy. Further the magnetic corrections 

 were applied for the glass of the carrier and for the nitrogen of the 

 bath as well as of the mixture ^). 



§ 3. Conclusions and experimental results. The mean numerical 

 data of the measurements are given in table I, where •/ signifies the 

 magnetisation of 1 gram of oxygen in eacli of the mixtures or in 

 (he pure oxygen; for the latter the \'alues are calculated on the 



^) Kamerlingh Onnes and Perrif.r, Comm. No. 116. Cf. note 1, p. 912. 



-) These corrections must be made even in a purely relative measurement, as 

 they are by no means proportional to the susceptibility of the experimental object 

 taken as a whole. 



