OF ARTS AND SCIENCES. 25 



nitrate,* instead of vice versa. This difference of procedure seemed 

 to increase the distance between the two end points, but not to 

 influence the final averaged result, f 



In most cases a slight excess of hydrobromic acid was added be- 

 fore filtering, but the amount recorded in the table always repre- 

 sents that which was equivalent to the end point of the reaction. 

 In Experiments 3, G, 12, 14, and 15, where argentic nitrate was 

 added in excess before filtration, the total amount of silver given 

 in the table signifies the sum of the amount of silver weighed out 

 and that which was added to attain the end point. The extra silver 

 is of course not counted. The agreement of Analyses 14 and 15 

 with 18 and 19 is sufficient to show the accuracy of both methods. 

 The clear yellow precipitate was washed by decantation until the 

 filtrate became wholly neutral, and was collected and weighed on a 

 Gooch crucible. The first filtrates were always passed through the 

 crucible several times, for fear that a trace of asbestos might have 

 been carried away. One of the absolutely clear filtrates containing 

 a trace of hydrobromic acid gave no sign of a reaction for silver 

 after evaporation to small bulk. 



In a number of experiments, modifications in the methods de- 

 scribed above were introduced. The most important change was 

 the one adopted in Analyses 2, 8, 14, and 18. In these four cases 

 the baric bromide was not heated at all, but the crystallized salt 

 was dissolved directly in water, t In order to determine the amount 

 of anhydrous salt which must have been present in these specimens, 

 parallel determinations of the water of crystallization were made 

 with the greatest care upon precisely similar samples. The agree- 

 ment of these results with the others is the best possible proof of 

 the accuracy of the alkaliraetric correction applied to those deter- 

 minations in which the substance had been ignited. 



In Experiments 4 and 13 the baric bromide was fused. In 

 Analyses 6, 14, 15, 18, and 19 the argentic bromide was fused, and 

 the weight of the fused salt is recorded in the final table. In the 

 first case the substance had been slightly darkened by exposure to 

 light; hence a little pure bromine vapor was added to the glass 

 tube in which the fusion was conducted, and the bromide gained 

 0.07 milligramme during the process. The other results are 

 tabulated below. 



* E. g. Exp. 14, 15, 16. 



t Compare Stas, Mem. Acad. Belg., XLIIL, Introduction, 

 t This procedure was suggested by a part of Marignac's work on the chloride 

 (loc. cit.). 



