RICHARDS AND CUSHMAN. — ATOMIC WEIGHT OP NICKEL. 103 



neutral ; it was next dissolved in strong hot hydrochloric acid, and after 

 the separated sulphur had been removed, the solution was evaporated to 

 dryness, and the residue was taken up with water. The material was 

 now considered fairly free from its usual impurities, with the single ex- 

 ception of cobalt. 



In commenting on the work of an early experimenter upon the atomic 

 weights of nickel and cobalt, Clarke has objected that " his results are 

 entitled to no especial weight at present, since it cannot be certain from 

 any evidence recorded that the oxide of either metal was absolutely free 

 from traces of the other." * Since the two metals have atomic weights 

 only differing at the outside by half a unit, " traces " of one in any prep- 

 aration of the other metal could not alone furnish a reason for invalidat- 

 ing the results. Nevertheless, for our purpose it seemed desirable to 

 prepare nickel as nearly free from cobalt as possible. In order to attain 

 this end with any degree of certainty, it is obvious that a qualitative test 

 must be found that should show with sufficient accuracy the presence or 

 absence of cobalt. Winkler t has recommended a test for which he 

 claimed greater accuracy than the better known method with potassium 

 nitrite. The moderately dilute solution of nickel is treated with am- 

 monia until a clear blue color is obtained and then one or two drops of 

 potassic permanganate are added. If no cobalt is present, the blue solu- 

 tion of nickel takes on a purple tinge ; whereas cobalt, if present, reduces 

 the permanganate. Winkler does not state the dilution of the perman- 

 ganate solution, or how much should be added, although manifestly the 

 degree of refinement of the test depends on these points. The perman- 

 ganate solution which worked well with us contained one gram of the 

 salt in a litre. To the dilute solution of nickel to be tested, contained in 

 a color-comparison apparatus, enough ammonia is added to render the 

 solution a light sky-blue, and then one tenth of a cubic centimeter of 

 the permanganate is dropped in. Under these conditions, the mixture 

 appears decidedly purplish in hue, if cobalt is absent. Of course the test 

 is of value only in the absence of any foreign substances having either a 

 reducing or an oxidizing action on permanganate. We have found it 

 possible by this method to detect one part of cobalt in twenty-five hun- 

 dred parts of nickel, an amount of impurity which could cause a final 

 error in the atomic weight of only one part in five hundred thousand. 



Anthon's process | for eliminating cobalt was adopted for the purifica- 



* Recalculation, 1897, p. 291. t See Dammer, Anorg. Chem., III. 490. 



t Zeitschr. Anal. Chein., VI. 20. 



