126 PROCEEDINGS OP THE AMERICAN ACADEMY. 



(additional run of three months) the presence of arsenic in sediment 

 and solution was confirmed qualitatively by different reagents, and I 

 assume that the precipitate of arsenious sulphide was not weighable. 

 In the third test (ten months additional run) the sediment gave 

 0.8 mgr. of sulphide (0.64 mgr. oxide) while the solution gave a " weak 

 reaction." Here again the precipitate seems to have been too small 

 to weigh. In the succeeding determinations the attempt to estimate 

 gravimetrically was abandoned, and the Berzelius-Marsh method was 

 used. Eighteen months had elapsed during which time the action had 

 apparently been at its height, as after sixteen months more the sedi- 

 ment in the fourth test gave no arsenic and the solution an amount 

 much less than the preceding. The rest of the tests £jave diminishing 

 amounts from the solutions, while none of the sediments, except the 

 last, contained any arsenic. The silver solutions were precipitated 

 with hydrochloric acid, filtered, and the filtrate, after evaporation 

 with sulphuric acid, was added to the Marsh apparatus. 



Unfortunately, the quantitative estimation of arsenic by the Berze- 

 lius-Marsh method was unknown to Hamberg, but I have placed the 

 most accurate estimate possible on the mirrors from his description of 

 them, and tabulate the results of the absorption tests on the opposite 

 page. 



In the interval between the third and fourth tests, the argentic 

 nitrate solution was one day replaced by a tube containing fused 

 calcic chloride, and the air led through a glowing tube for eight 

 hours. No deposit of arsenic was found. From January (15?), 

 1882, to March 28, 1882, a second argentic nitrate solution was placed 

 after the first, and, on examination at the same time, contained no ar- 

 senic. After this, although a second solution was used, there is no 

 record of its having been tested beyond the statement that the sedi- 

 ment formed was very slight. From July 14, 1885, to October 28, 

 1885, the argentic nitrate was replaced by a tube containing nitric 

 acid (strength not given), and for most of the time there was an 

 argentic nitrate solution after it. This acid yielded over a gram of 

 ammonic nitrate which was not arsenical. The argentic nitrate solu- 

 tion was kept in a month longer and gave only a faint deposit. This 

 was the last test. 



Making allowance for the second test there were found from 2 to 

 3 mgr. of arsenic as arsenious oxide in the air passed into the argentic 

 nitrate, if we assume that each deposition of an arsenical mirror was 

 carried to completion. This is, however, only from 0.44 to 0.66c£, of 

 the 449 mgr. which are supposed to have escaped as a gaseous com- 



