186 PROCEEDINGS OF THE AMERICAN ACADEMY. 



In order to make sure that the reaction was complete the silver 

 chloride was again fused, and exposed to the action of hydrochloric 

 acid for several hours longer. As a rule, no change in weight was 

 observed. In all cases constant weight was obtained upon heating in 

 the same way for a third time. 



After making certain that only a small quantity of arsenic, if any, 

 remained in the silver chloride, the contents of the quartz tube were 

 dissolved in ammonia, and the silver chloride was reprecipitated by 

 boiling the solution to expel the ammonia and adding a small quantity 

 of sulphuric acid. The solution, after evaporation, was added to a 

 Berzelius-Marsh apparatus containing arsenic-free zinc and sulphuric 

 acid, and a mirror of arsenic was deposited in a hard glass capillary 

 tube in the usual way. The hydrogen was dried by calcium chloride 

 before passing into the hard glass tube, and the generating flask was 

 cooled with water to prevent the evolution of hydrogen sulphide. 



The arsenic mirror formed was compared with a photograph of 

 standard arsenic mirrors,^ the original mirrors showing that compari- 

 son with the photograph was equally satisfactory. The correction was 

 applied by assuming that the arsenic was present in the silver chloride 

 as arsenic trichloride, although if present as silver arsenate the correc- 

 tion would be much smaller. In any case the correction for residual 

 arsenic is so small as to be almost without effect upon the final result. 



Ebaugh used essentially the same method of heating the arsenate 

 in hydrochloric acid, although the periods were shorter, so that it 

 is probable that the small quantities of arsenate used (scarcely 

 over one gram in any analysis) did not retain weighable amounts of 

 arsenic. 



The U-tube beyond the quartz tube was washed out thoroughly with 

 dilute ammonia, and the solution was made up to definite volume after 

 nearly all the ammonia had been expelled by boiling. The silver con- 

 tent of the solution was then compared in the nephelometer with that 

 of standard solutions of silver, care being taken that the tubes were 

 treated in as nearly as possible the same way. 



The second method of analysis consisted in heating the silver arse- 

 nate in a platinum boat and, after weighing, dissolving the arsenate 

 in nitric acid and precipitating the silver as chloride or bromide. The 

 platinum boat was heated in a hard glass tube forming part of a bot- 

 tling apparatus,9 by means of which the boat could be transferred 



8 Sanger, These Proceedings, 26, 24 (1891). 



9 Richards and Parker, These Proceedings, 32, 59 (1896). 



