ATOMIC WEIGHTS 369 



the silver thus taken into solution was titrated with standard sodium 

 chloride. In three others the tungstate was treated directly with com- 

 mon salt, and the residual silver chloride collected and weighed. Here 

 again, on account of some complexity in Zettnow's figures, I am com- 

 pelled to reduce his data to a common standard. To 100 parts of AgCl 

 the following quantities of AggWO^ correspond : 



By First Method. 



161.665 

 161.603 



Mean, 161.634, ± .021 



By Second Method. 



161.687 

 161.651 

 161.613 



Mean, 161.650, ± .014 

 General mean from both series, 161.645, ± .012 



Hence W = 183.64. 



For tungsten hexchloride we have first, two analyses by Eoscoe, pub- 

 lished in the same paper with his results upon the trioxide. In one ex- 

 periment the chlorine was determined as AgCl ; in the other the chloride 

 was reduced by hydrogen, and the residual tungsten estimated. By 

 bringing both results into one form of expression we have for the per- 

 centage of chlorine in WClg : ^ 



53.610 

 53.632 



Mean, 53.621, ± .0074 



Hence W = 184.02. 



The investigation of tungsten hexchloride by Smith and Exner was 

 much more elaborate. They prepared the substance from scrupulously 

 pure materials, and further purified it by repeated sublimations. They 

 decomposed the chloride by means of water, and weighed the residual 

 tungsten trioxide. Their figures, with vacuum weights, are as follows, 

 with a percentage column added by myself: 



• The actual figures are as follows: 

 19.5700 grm. WCle gave 42.4127 grm. AgCl. 

 10.4326 grm. WCle gave 4.S374 grm. tungsten. 



