RICHARDS AND CUSHMAN. — ATOMIC WEIGHT OF NICKEL. 345 



both gravimetrically and volumetrically.* He wisely regulated his opera- 

 tions in such a way as to avoid the use of alkalis ; but in his iirdor 

 to escape this danger he encountered others as serious. It is highly 

 unlikely that the electrolytic nickel, dried at only 50° while adhering 

 to the dish, could have been free from impurities, as we have already 

 shown in the experimental part of our paper. Indeed, he confessed in a 

 later paper f that the electrolysis of cobalt at any rate is very far from 

 being as accurate a process as it is sometimes supposed to be. The 

 report of the acid reaction of the chlorides of these metals would have 

 had more significance if the indicator had been named, for the salts 

 destroy the magenta of phenol phthalein only because they remove the 

 hydroxyl ions from the solution. Both chlorides are perfectly neutral to 

 methyl orange. A loss of chlorine during the drying of either chloride 

 would of course raise the observed atomic weight ; and while in the 

 case of nickel the loss was so small as not to have produced a visible 

 cloudiness, no proof is offered that no loss took place. In the case of 

 the cobalt " eine gewisse, aber so schwache Triibung, dass sie, wie man 

 zu sagen pflegt, nicht ' blank ' erschien " involved a loss of as much as 

 one per cent of material, and it is well possible that a smaller but still 

 important amount of basic salt may have escaped notice in the case of 

 the nickel. Our own experience with the halides of both metals con- 

 vinces us that it is quite impossible to obtain them pure and dry by evap- 

 orating to dryness in moist air. Turning now to the determination of 

 the chlorine in the salts, we find other grave flaws. No account was 

 taken of the solubility of argentic chloride in the gravimetric work, and 

 several milligrams must have been washed away by the hot dilute nitric 

 acid used as a washing fluid. One is surprised, too, to find that the anti- 

 quated process of burning the filter was adopted, instead of Gooch's 

 admirable substitute. In the volumetric work again the solubility of 

 argentic chloride was overlooked, although it produces a most injurious 

 effect on the method of Volhard.t It is indeed surprising to see so emi- 

 nent a chemist using volumetric methods at all in this way, for every one 

 knows the difficulty of obtaining results of a very high grade of accuracy 

 by their aid. In this laboratory the burette is only called into use when 

 at least ninety-nine per cent of the material has been weighed out, and 

 then only a few cubic centimeters of a very dilute solution are added to 



* Zeitschr. Anorg. Chem., IV. 10, 1893. 

 t Zeitschr. Anorg. Chem., VIII. 4. 

 t These Proceedings, XXVI. 34, and XXIX. 67. 



