552 



with the first in tlie torsion balance, expecting to find a devia- 

 tion of, at the utmost, a few tenths of a mm. when observing in 

 the cellar, and practically zero in the vestibule. An observation of 

 August 24 in the cellar however gave, using the torsion rod, charged 

 with the two uranium oxide tubes, a difference of IF ^ JS" 2,1 mm.; 

 in the vestibule an observation of August 25 gave W "^ £"2,2 mm. 



The first tube was incidentally broken and part of the contents 

 lost. Tlie further experiments were made with the second tube 

 (balanced against quartz) and gave on August 28, 29, 31 and 

 September 2 tlie results W'^E 1, 1, 2.5, 3 mm. 



The deviations found widely exceed the errors of observation. 

 They prove (observations of August 24 and 25) that the two samples 

 of uranium oxide are not identical. Probably both or at least one 

 of the uranium oxides are contaminated by iron. ') By this hypothesis 

 we may understand that the magnitude of the deviation changes 

 with time (observations of August 28 and following days), and even 

 exhibits contrary signs (first observations in the cellar compared 

 with the later observations in the vestibule;. 



I had not yet an opportunity to test quantitatively the suggestion 

 as to the influence of a contamination by traces of iron. This rpust 

 be reserved to another occasion. Meanwhile, I have now to record 

 observations with uranyl nitrate, which go far to prove that radio- 

 active substances also follow the law of proportionality between 

 mass and weight with great accuracy. The uranyl nitrate was in- 

 cluded in a thin cylindrical tube and balanced against quartz. The 

 results of observations of September 10 and 11, made in the vestibule, 

 are plotted in figs. 7 and 8. 



The curves are not quite parallel to the axis of abscissae, a case 



A2- 



A3 



Uranylnitrate [Sept. 10, vestibule] 



"2 3 4 i 6 7 



Fig 7. 

 Uranylnitrate [Sept. 11, vestibule] 



'° hours 



k: hours 



Fig. 8. 



1) Owen (Ann. d. Phys. 37, 686, 1912) finds that the three preparations of 

 uranium, which were used in his magnetochemical experiments, contained much 

 iron. See also there (p. 672) some remarks on the omnipresence of iron. 



