22 RESEARCHES UPON ATOMIC WEIGHTS. 



solution until reduction of the bromine was complete. The solution was boiled, 

 after mechanical separation of the greater part of the free sulphur and bromide 

 of sulphur, and was filtered. Iodine was eliminated by boiling the hydrobromic 

 acid with several small portions of potassium permanganate and rejecting the 

 bromine set free. By heating the remainder of the hydrobromic acid with an 

 excess of permanganate, over half of the bromine was obtained in the free state. 

 The process of reduction with hydrogen sulphide and oxidation with perman- 

 ganate was then repeated with the resulting bromine, and the final product was 

 redistilled shortly before use. 



SILVER. 



One sample of silver was purified especially for this research. The processes 

 to which it was subjected consisted first of precipitation as chloride from nitric- 

 acid solution and reduction with invert sugar and sodium hydroxide. After 

 fusion with a blowpipe on a crucible of the purest lime the metallic buttons 

 were freed from surface impurities by scrubbing with moist sand and etching 

 with nitric acid. Next the buttons were dissolved in nitric acid and the solu- 

 tion was reduced with ammonium formate.^ The precipitated silver was 

 thoroughly washed and again fused on a lime crucible. The final process of 

 purification consisted in electrolyzing the silver as described on page 6. The 

 electrolytic crystals were fused in a current of hydrogen on a lime boat, and 

 the buttons, after cleansing with nitric acid and drying at 200°, were cut into 

 fragments of convenient size with a clean chisel and anvil. Then they were again 

 treated with fresh portions of dilute nitric acid until free from iron, washed, 

 dried, and finally heated to about 400° in a vacuum. This silver was employed 

 in analyses 4 to 8. 



In the first three analyses a mixture of two specimens of silver was em- 

 ployed, both of which had already been used in an investigation upon the 

 atomic weight of iodine by one of us.^ One was prepared from silver nitrate 

 which had been seven times recrystallized from nitric acid, five times recrys- 

 tallized from water, and finally precipitated with ammonium formate. The 

 other was precipitated once as silver chloride, electrolyzed once, and finally 

 reduced with ammonium formate. 



DRYING OF CADMIUM BROMIDE FOR ANALYSIS. 



The method of analysis was essentially that usually employed in this 

 laboratory for the analysis of metallic haUdes. Weighed portions of the 

 bromide, after fusion in nitrogen and hydrobromic-acid gases, were first titrated 

 against weighed portions of silver. Then the precipitated silver bromide was 

 collected and weighed. 



' Richards: Pub. Car. Inst. No. 28, p. 19 (1905); Jotir. Amer. Chem. Soc, 27, 475; Ze//. 

 anorg. Chem., 47, 72. 



2 Baxter: Proc. Amer. Acad.. 41, 79 (1905); Jour. Amer. Chem. Soc. ,27, 881; Zeit. anorg 

 Chem., 46, 42. (See page 108.) 



