Ö99 



t'orivclion was needed i'ov the magnetism of the hadi or of the wall. 

 The ox\gen nsed contained nol more than (),()()1 foi-eign admixtnres. 

 The immediate resnlt of the observations is the snsceptihilitj A' per 

 unit of volume; to calcMdate the value of "/, the specific siisceplihility 

 (or specific magnetisation coefticienlj, we must know tiie densit}- o. 

 We have taken 1.44 for this, on the ground of our determinations 

 in 1910, and from a new one recently made, hut neitlier determi- 

 nation can lay claim to great accuracy, so that the values of /(and 

 also of K for the opaque modification if it should prove that in this 

 innumerable little splits are present) will have to be recalculated 

 when o is better i^nown. 



The four results refer to one freezing. 



TABLE I. 



A new freezing gave two \alues for K near 165.5. 10^^ (before 

 and after partial melting) at about the same temperature ( — 230° C). 

 This value is less to be relied upon than that in the Table. We had 

 very little time at our disposal for our joint work, and this was 

 even diminished by the apparatus being found somewhat deficient 

 in some points. On this account we hud not time to test a tempera- 

 ture lying between the melting point and the transition |)oint from 

 the transparent to the opaque modification. Hut we had already 

 postponed the resumption of our researches after the de[)arture of 

 one of us from Leiden, for two years, and we shall not have an 

 0p|)0rtunity in the near future to continue our joint research. We 

 therefore felt bound to publish what we had so far established. 



This is, besides the numei-ical values of our table within the limits 

 of accuracy given for it, that the transition from the transparent 



