1904.1 RICHARDS — THE ATOMIC WEIGHT OF NITROGEN. 119 



This percentage of silver in argentic bromide corresponds to an 

 atomic weight of 79.954 for bromine, if silver is taken as 107.93 ^ 

 and in my opinion this value is if anything rather too low than too 

 high, because most of the probable errors tend to diminish it. 



This value agrees almost exactly with that of Stas's work, and 

 hence there seems little room for doubt that if silver is 107.93 bro- 

 mine is between 79.95 and 79.96. Therefore Scott's bromine 

 (79.943) must have been impure, probably containing chlorine, 

 which is not easy to eliminate. It is also possible that the argentic 

 bromide was precipitated from too concentrated a solution and 

 hence contained nitrate. This error would have affected the 

 result in the same way. 



If there were no error in determining the amount of silver needed 

 for the precipitation, it would be easy nevertheless to correct for 

 this impurity of the bromine in the following way. Scott found 

 that 107.93 parts of silver corresponded to 97.995 parts of ammonic 

 halide and 79.943 parts of halogen. By subtracting the last figure 

 from the one preceding it the molecular weight of ammonium is 

 found, independent of the nature of the halogen, to be 18.052, and 

 hence that of nitrogen 14.022. This value is considerably higher 

 than that calculated by Scott on the assumption that his ammonic 

 bromide was perfectly pure. Unfortunately, however, the presence 

 of chlorine in the salt, by introducing the partly soluble argentic 

 chloride, complicates both the determination of the amount of sil- 

 ver required for the precipitation and the collection of the salt of 



