A REVISION OF THE ATOMIC WEIGHT OF ARSENIC. 87 



In the second place, both methods for determining the ratio of the arsenate 

 to the chloride give essentially identical results. This is shown by the agreement 

 of Series I and II, and that of Series IV and V. 



Finally, the per cent of silver in silver arsenate found in Series I and II agrees 

 within less than 0.002 per cent with that found in Series III. This agreement, 

 together with that of the individual analyses of each series, indicates both uni- 

 formity in the material employed and purity of the hydrochloric and hydro- 

 bromic acids, as well as accuracy in the analytical work. The agreement of 

 Series IV and V with Series VI is closer still. 



In the following table are given the sources of the various samples of 

 silver arsenate: 



Since in the precipitation of the silver arsenate the arsenate solution was 

 always added to the silver-nitrate solution, the mother-liquor was in every case 

 either neutral or acid, but in no case permanently alkaline. Still, local accu- 

 mulations of arsenate undoubtedly exist for short periods, and precipitates 

 formed at these points may be affected by the conditions existing in the arsenate 

 solution. In the case of Samples F and G the arsenate solution is decidedly 

 alkaline, owing to hydrolysis, but less so with Samples A, D, and E, and least 

 with Samples B and C. In the case of Samples B and C, acid accumulates in the 

 solution during the precipitation of the silver arsenate. In comparing the results 

 from the different samples of silver arsenate it must be noted that occluded 

 basic salt would increase the apparent percentage of silver in the arsenate. In 

 the case of Samples F and G the conditions are most favorable for the occlusion 

 of basic salts, and these two samples actually yield a higher percentage of silver 

 than the other samples. On the other hand accumulation of acid in the solution 

 in which the precipitation of the silver arsenate is taking place is found to have 

 no tendency to promote occlusion of acid salts, since Samples B and C give 

 results agreeing closely with those of Samples A, D, and E. These two facts 

 lead to the conclusion that Samples A to E represent normal trisilver arsenate, 

 and that Samples F and G contain basic impurities. 



In order to calculate the atomic weight of arsenic from the per cent of silver 

 in silver arsenate a knowledge of the ratio of the atomic weights of silver and 

 oxygen is necessary. A slight uncertainty exists as to this ratio, hence calcula- 

 tions have been made upon the basis of two extreme values for silver, 107.880 

 and 107.870 . This has been done only with the results of Series I, II, and III, 

 since, as has been pointed out, they are probably nearer the truth than those of 

 Series IV, V, and VI. The difference between the two sets of results amounts 

 only to 0.06 per cent in the atomic weight of arsenic. 



