38 RESEARCHES UPON ATOMIC WEIGHTS. 



the weighing-bottle in which it was originally weighed, and the stopper was in- 

 serted without an instant's exposure of the salt to moisture, by means of the 

 bottling apparatus described on page 9. The weighing-bottle was then al- 

 lowed to stand in a desiccator near the balance case for some time before it was 

 weighed. 



METHOD OF ANALYSIS. 



Next the boat was transferred to a flask and the salt was dissolved in about 

 300 c.c. of the purest water. The weighing-bottle was rinsed and the rinsings 

 were added to the solution. Then the solution was filtered into the glass stop- 

 pered precipitating flask through a tiny filter to collect a trace of insoluble 

 matter, and the filter-paper and residue were ignited in a weighed porcelain 

 crucible. 



From the weight of manganous bromide very nearly the requisite quantity 

 of pure silver could be calculated. This silver was weighed out and after solu- 

 tion in nitric acid as described on page 12, and dilution until not stronger than 

 I per cent, the solution was slowly added with constant stirring to the i per 

 cent solution of manganous bromide in the precipitating flask. After having 

 been shaken for some time, the solution was allowed to stand several days, one 

 week in the case of analyses 14 and 15, with occasional shaking, until the super- 

 natant liquid was clear. 30 c.c. portions of the solution were then tested with 

 hundredth-normal solutions of silver nitrate and sodium bromide in the nephel- 

 ometer for excess of bromide or silver, and, if necessary, either standard silver 

 nitrate or sodium bromide solution was added, and the process of shaking and 

 testing repeated, until the amounts of bromide and silver in the solution were 

 equivalent. If the solution was perfectly clear when tested, and contained no 

 considerable excess of bromide or silver, the test solutions were discarded, since 

 they contained only negligible amounts of silver bromide, otherwise they were 

 returned to the flask and a correction was applied for the silver bromide thus 

 introduced. 



As soon as the exact end-point of the titration had been found, about 4 eg. 

 of silver nitrate in excess were added to precipitate dissolved silver bromide 

 and the solution was again shaken and allowed to stand until clear. The pre- 

 cipitate of silver bromide was collected upon a weighed Gooch crucible after it 

 had been washed with water by decantation about ten times. Next it was heated 

 for several hours at 140°, then for 2 hours at about 230° in an electric air-bath, 

 and, after it had cooled in a desiccator, it was weighed. In order to determine 

 how much moisture was retained by the precipitate, in each case it was trans- 

 ferred as completely as possible to a clean porcelain crucible and weighed; then 

 the salt was fused by heating the small crucible, contained in a large covered 

 crucible, and again weighed. The fused salt was light yellow as a rule, showing 

 that no appreciable reduction had taken place. The asbestos mechanically de- 

 tached from the Gooch crucible together with a minute quantity of silver bro- 



