CHEMISTRY — RICHARDS. 155 



Parallel with this, and supporting its conclusions, there was conducted : 



(2) A revision of the atomic weight of potassium by the analysis of potassic 

 bromide, carried out with the assistance of Edward Mueller. This problem 

 involved a number of new points, because the preparation of potassic bromide 

 in a pure state involves many difficulties. The electrolysis of potassic oxalate 

 with a mercury cathode and the subsequent electrical decomposition of the 

 resulting amalgam was found to be one of the most convenient means of 

 obtaining pure potassic hydroxide used in making the bromide. The results 

 confirmed the work with the chloride in a satisfactor}' manner, giving K =^ 

 39.1143 by reference to silver, and 39.1135 by reference to argentic bromide. 

 The average 39.114 of the essentially identical results of the previously 

 described investigation and this one may be accepted as representing very 

 closely the true atomic weight of potassium, if silver is taken as 107.930. 



As a contribution towards the determination of the true value for silver, 

 the following investigation also was successfully brought to completion with 

 the assistance of Dr. G. S. Forbes : 



(3) The synthesis of argentic nitrate. Here again great care was used in 

 every stage of the work. The employment of vessels of fused quartz and 

 other precautions too numerous to mention carried the work to an unusual 

 degree of accuracy. The essentially identical results show^ed that if silver 

 is taken as 107.930, nitrogen can hardly be low^er than 14.037 ; or if nitrogen 

 is taken as 14.008, silver can hardly be higher than 107.88. Another some- 

 what similar research was : 



(4) A new determination of the atomic w^eight of sulphur, carried out with 

 the assistance of Grinnell Jones. Recent investigations, carried out under 

 the auspices of the Carnegie Institution of Washington, have given such 

 definite knowledge concerning argentic chloride that the conversion of the 

 sulphate into the chloride promised to yield trustworthy^ data for computing 

 the atomic weight of sulphur. It was found that the former salt is com- 

 pletely converted into the latter by gentle ignition in a current of hj'dro- 

 chloric acid gas ; and very careful experiments conducted in quartz vessels 

 indicated that the ratio of the equivalent weights is 100.000 : 91 .933. Hence, 

 if silver is taken as 107.93 and argentic chloride as 143.403, sulphur must 

 be 32.113, a much higher figure than that usually accepted (32.06) ; and 

 even if silver is as low as 107.89, sulphur must be almost 32.08. This out- 

 come was not a surprise, for it had been clear that impurities in Stas's silver 

 might have affected his results concerning sulphur as well as his work with 

 chlorine. A single research is not enough to furnish conclusive results 

 in a case of this kind, and the study of the atomic weight of sulphur will 

 be continued. 



In addition to these finished researches, other work of the same type was 

 begun, which had as its object the final decision of the uncertaint}^ concern- 



