52 



RESEARCHES UPON ATOMIC WEIGHTS. 



Marignac ^ also determined the ratio of silver to silver bromide, with some- 

 what lower results, — 79.922, 79.904, and 79.915; average, 79.913. 



Scott,^ in his analyses of ammonium bromide, obtained six values for the 

 same ratio, varying between 79.899 and 79.911, with an average of 79.906. 

 One of his results is here rejected, since the silver used in this experiment was 

 known to be impure. 



Indirect Determinations. 



Dumas ' by heating silver bromide in chlorine found the values 80.02, 79.87, 

 and 79.94. 



In computing the atomic weight of bromine from these data, great weight is 

 always given to Stas's determinations, the value 79.918 being usually assumed 

 as the most probable one for the constant in question. Certainly, as pointed 

 out by Richards,* the true value must lie between 79.91 and 79.92. Clarke 

 calculates the value 79.912 as the weighted average of the diflferent investiga- 

 tions previous to Scott's.* 



* CEuvres Completes, i, 81. * Jour. Chem. Soc. Trans., 79, 147 (1901). 



• Ann. Chem. Pharm., 113, 20 (i860). * Proc. Amer. Phil. Soc, 43, 119 (1904). 

 ' A Recalculation of Atomic Weights, Smith. Misc. Coll., 1897. 



