553 



realized in the cellar observations Figs. 3 and 4, but exhibit a 

 small slope due to the rise of temperature in the vestibule. This 

 slope is however the same for the E. and W. -curves. The lines are 

 even practically coincident for the greater part of the observations 

 of September 11, and the small residual differences exhibit contrary 

 signs in the observations of September 10 and 11. The observations 

 justify us in concluding that for uranyl nitrate a deviation of the 

 law of the constancy of the ratio of mass and weight is less than 

 1 i?i tiventy million. 



It seems extremely improbable that the behaviour of uranium 

 should be otherwise, as far as a so fundamental property as mass 

 is concerned, in an oxide than in a nitrate. The deviation found in 

 the case of an oxide, is therefore most probably due to a magnetic 

 contamination. If the deviation found in the case of the oxide were 

 really due to a change of mass, than the nitrate should exhibit 

 about half the effect of the oxide or about 1,5 mm., allowance being 

 made foi- the quantity of uranium in the two combinations. But the 

 effect is most certainly less than 0.2 mm. 



Perhaps 1 may be allowed to add that electric disturbances during 

 the observations were excluded sometimes by a short exposure of 

 the inner case, opened to this purpose, of the balance to radium 

 rays, sometimes by placing a i^w scrapings taken from the leaden 

 box containing the radium preparation. 



I have projected recently some improvements of the apparatus 

 and the method of observation, by which I hope to be able to 

 increase the accuracy, according to an estimation at the safe side, 

 ten times. 



I hope to return to this subject on another occasion. 



