130 PROCEEDINGS OF THE AMERICAN ACADEMY. 



flask was placed an empty Kempff washing bottle (B), reaching just 

 below the stopper and serving as a safeguard against back pressure 

 as well as to catch any of the arsenical mixture that might be mechan- 

 ically carried up. Through the rubber stopper of the first washing 

 bottle was passed a second (C), containing about 50 c. c. of the same 

 argentic nitrate solution to absorb the volatile compound. This bottle 

 was connected with the right-angled tube of a side-neck test-tube (D), 

 similar in arrangement to A, and serving as a guard against contami- 

 nation of the silver solution in C from this end. To the tubulus of D 

 was attached the water-pump. The flask was kept at 30° for 36 

 days and the fermentation was marked, air being drawn through 

 nearly all the time. At the end there was no perceptible change in 

 the solution in the bottle C. This solution was tested as follows. A 

 clean new evaporating dish was heated with about 3 c. c. strong sul- 

 phuric acid, and the acid after cooling was diluted with about eight 

 parts of water and introduced into the Marsh reduction flask.* No 

 mirror appeared in 45 minutes, at the end of which time was added 

 the solution obtained from the argentic nitrate. This had been heated 

 to boiling, and precipitated with hydrochloric acid, the filtrate being 

 evaporated with the addition of a few drops of nitric and sulphuric 

 acids until it fumed strongly. Water was then added, and the cold 

 solution put into the reduction flask. No mirror appeared although 

 the action continued for 45 minutes, thus showing no absorption by 

 the silver solution of any volatile compound of arsenic from the 

 fermenting mixture. 



Experiment 2. — In the following two experiments the dilution 

 was made very great, in case this condition should affect the forma- 

 tion of the volatile compound. The arrangement of the flask and 

 rear guard tube (A) was as in Exp. 1. Next to the flask was 

 placed a short straight tube filled with fused calcic chloride (B), 

 and to this was connected a washing bottle (C) with 2°f argentic 

 nitrate. Diluted syrup solution and yeast were placed in the flask, 

 and to it were added 10 c. c. of a standard solution of arsenious oxide 

 equivalent to 10 mgr. As 2 3 . The apparatus closed at A was left for 

 nine days at the ordinary temperature. Fermentation set in at once 

 and continued until the end. The silver solution was somewhat 

 darkened, probably by the action of light and some volatile organic 

 compound. Air was drawn through the system until the air in the 



* The method used in all the experiments was the modification of the 

 Berzelius-Marsh described by me in these Proceedings, XXVI. 24. 



