A REVISION OF THE ATOMIC WEIGHT OF MANGANESE. . 39 



mide which occasionally escaped the crucible, was collected from the filtrate 

 and wash-waters upon a small filter, the ash of which was treated -^dth nitric 

 and hydrobromic acids before weighing. Although the filtrates and first wash- 

 waters were essentially free from dissolved silver bromide, the subsequent wash- 

 waters usually contained a trace of silver bromide. The amount of dissolved 

 salt was determined with the nephelometer by comparison with standard bro- 

 mide solutions. 



Several difficulties were met in carrying out the analyses. In the first place it 

 proved difiicult to wash the platinum boat absolutely clean. When rinsed with 

 cold water only, and dried at 100°, the weight was in many cases a few hun- 

 dredths of a milligram greater than before fusion of the bromide. Ignition to 

 redness of the boat thus treated then produced a slight loss in weight. Rinsing 

 with hot water reduced the gain in weight of the boat after drying but did not 

 wholly prevent a sUght loss on ignition. The cause of the variation was not 

 discovered, hence it seems safer in the calculations to use the weight of the boat 

 after drying. The total variation is so slight, however, that it scarcely affects 

 the final result. 



Two other difficulties arose from the fact that when a solution of a manganous 

 salt, even as dilute as the filtrates from analyses, is filtered through filter-paper, 

 in spite of long-continued washing a small amount of manganese is tenaciously 

 retained by the paper. This was discovered from the fact that the asbestos 

 residues always contained manganese. In analyses 29 to 31 it was found 

 possible to eliminate the manganese completely by washing the filter finally with 

 5 per cent hydrobromic acid. In two cases the residues were analyzed for 

 manganese and were found to contain 0.00023 and 0.00057 gm. of Mna O4 

 respectively. 



The average of these two quantities is, however, larger than the total residue 

 in some cases, hence this value can not be used to correct the previous analyses. 

 In order to determine accurately the proper correction for this error, a solution 

 containing manganous nitrate in the porportion in which it was usually con- 

 tained in the filtrate of an analysis was passed through filter-papers and the fil- 

 ters were then washed as thoroughly as possible with water. The ash of these 

 papers invariably contained manganese, the weights of manganic oxide in sev- 

 eral experiments being found to be 0.00018, o.oooii, 0.00006, 0.00018 and 

 0.00005 gm. with an average of 0.00012 gm. This quantity was subtracted 

 from the weight of asbestos shreds in all cases except analyses 29 to 31, where 

 the paper was washed with hydrobromic acid. 



The residue obtained by the filtration of the manganous bromide proved to 

 contain manganese and to be free from detectable amounts of platinum and 

 silica. Probably this insoluble residue consisted chiefly of oxides of mangan- 

 ese, although prolonged fusion in hydrobromic acid failed to reduce materially 

 the proportion of insoluble matter. The discovery of adsorption in the case 

 of manganous nitrate led to the suspicion that at least a portion of this residue 



