319 



Mean, 50.388, ± .0043 



The three series seem to be fairly in agreement between themselves, 

 and with Reiser's work, but diverge seriously from the electrolytic data. 



Keller and Smith also attempted to determine the atomic weight of 

 palladium by heating the palladiammonium chloride in sulphuretted 

 h3'drogen, and so converting it into the sulphide, PdS. These data were 

 obtained : 



Mean, 65.651, zh .0051 



Hence Pd =^ 106.55. This result, however, is affected by the work of 

 Petrenko-Kritschenko,* who has shown the existence of the sulphide 

 PdS to be uncertain. 



Joly and Leidie,t in their determinations of this atomic weight, re- 

 turned to the potassium palladiochloride, K^PdCl^. In their first series 

 of experiments the salt was dried in vacuo at ordinary temperatures. It 

 was then electrolyzed in a solution acidulated with hydrochloric acid, 

 both the deposited palladium and the potassium chloride being weighed. 

 The palladium was dried, ignited in a stream of hydrogen, and cooled in 

 an atmosphere of carbon dioxide. The results were as follows, with the 

 column added by me giving the Pd equivalent to 100 parts of KCl : 

 K.,Pda^. Pd. 2KCI. Ratio. 



1.0255 -3919 .5520 70.996 



1.2178 .3937 .555' 70.924 



I. 2518 .4048 .5687 71.016 



Mean, 70.979, ± .0188 



This series was rejected by the authors, because the salt was found to 

 contain water — in one case 0.23 per cent. This error, however, should 



* Zeit. Anorg. Cheni., 4, 251. 1893. 

 tCompt. Rend., 116, 147. 1893. 



