OF ARTS AND SCIKNCR^^. 48 



compares liis results witli the two t'onuiilas (old style), Co,,Cl|,ON'|,.II|j., 

 and Co,,Cl,,OoN,^II^^.. I liiid tliat his analyses agree fairly well willi 

 the t'orniida, — 



Co,(xiy,.o.ci,+Nii,ci. 



Thus we have, — 



If we suppose that the blaek salt consists at least essentially of 

 Co^(NII.,),..O.Cl4 the formation of Rose's dark green chloride of dichro- 

 cobalt is readily explained by the ecpiation. — 



Co,(NIL),.O.Cl,4-211Cl = Co.(NII,),ri„+OIL, 



The existence of such a double chloride as Co„(NII.,)|..O.Cl4-[- 

 NH^Cl is in itself not very probable, nor is it easy to see how such a 

 salt could be dissolved in concentrated sul{)luiric acid, and precipitated 

 by cldorhydric acid without change. Farther investigations are re- 

 quired to determine the constitution of the salt definitively. 



Terrell * appears to have first shown that salts of the cobaltamines 

 are formed when powerful oxidizing agents are added to ammoniacal 

 solutions of cobaltic salts. Terrell employed hypermaiiganates and 

 hypochlorites ; Brauii.f the hyperoxides of lead and manganese ; 

 Mills, + iodine, bromine, and potassic dichromate ; Blomstrand, § 

 iodine and cobaltic sulphate. On repealing these processes, I find that 

 that of Mills with potassic dichromate is by far the best for preparing 

 nitrates of roseocobalt and purpureocobalt. Blomstrand's method is 

 inconvenient upon the large scale in consequence of the insolubility of 

 the sulphato-iodide of luteocobalt, but answers well on the small scale, 

 and gives a fine lecture-table experiment. A good method of prepar- 

 ing the salts of luteocobalt in quantity is still wanting, as the chloride 

 and nitrate form valuable reagents in various aTialytieal opei-ations. 



* Zeitschrift fur Anal Chemie, v. p. 114. t Comptes Rendus, Ixii. p. 139. 

 X Phil. Mag. (4) xxxv. p. 245. § Cliemio tier Jetztzeit, p. 295. 



