911 



X{r+A) = const., 

 in which onl}' tlie j)araiiietei' A changes from one conceiitratiou to 

 the other. Table II shows this : 



TABLE II. 



Therefore : the cliange in density of the oxygen only alters the 

 specific magnetisation, loithout changing the QvKm-constant. 



The pure liquid oxygen seems to form an exception : the straight 

 line ^ly=zf\T) for this differs considerablj' in direction from that 

 for the mixtures. We would call attention to the fact, however, that 

 each mixture was examined at constant concentration (see § 2) ; while 

 this was not the case with pure oxygen, which shrinks considerably 

 on cooling (10 7„ about between — 183° and — 210°) ; if we 

 calculate for any temperature (say — 195°) the specific magnetisation 

 that pure oxygen would have at that temperature and at the densities 

 corresponding to the other temperatures, using the CuiiiE-constant 

 which is common to the mixtures at constant concentration, the 

 values obtained thus (see Fig. 3 the points indicated by black disks') 

 fall in a natural way upon the general curve, which gives the specific 

 magnetisation as a function of the concentration at the temperature 

 under consideration ; the data obtained from these measurements form 

 therefore strong arguments in favour of the conclusion that in the 

 investigation of pure liquid oxygen at constant density a curve f or tlte 

 change of the magnetisation would appear, lohicJi only differed from that 

 for the mixtures by a new translation. This causes the anomaly to 

 disappear. The strict experimental proof of this conclusion can only 

 be obtained by a great number of experiments with a very concen- 

 trated liquid, or by means of a direct experiment, in which we begin 



