34 RESEARCHES ON THE 



class of salts, and lead to the idea that sulphuric and oxalic acids may possibly be 

 capable of replacing each other in other combinations. 



OXIDE OF PTTRPUREOCOBALT. 



The oxide of Purpureocobalt, like that of Roseocobalt, appears to exist only in 

 solution. It may be prepared, either by decomposing the acid sulphate by baryta 

 water, or by digesting a solution of the chloride with oxide of silver in the cold. 

 The solution is not pure in either case, containing either sulphate of baryta or 

 chloride of silver in solution. The oxide, as thus prepared in solution, forms a 

 violet-red liquid, which absorbs carbonic acid readily from the air, and which is 

 decomposed by concentration. 



The constitution of the oxygen salts of Purpureocobalt, which we have described, 

 as well as that of the chlorplatinate of this radical, appears to us to leave no 

 reasonable doubt that the oxide is essentially biacid. According to the rule that 

 the number of equivalents of acid in a salt is equal to the number of equivalents 

 of oxygen in the base, the rational constitution of the oxide of Purpureocobalt 

 will be expressed by the formula 



5NH 3 .Co 2 0.0 2 . 



We shall develop this view more fully when occupied with the purely theoretical 

 portion of the subject, and in the second part of our memoir we shall endeavor, by 

 the analysis and description of other salts of Purpureocobalt, to throw more light 

 upon the nature of this remarkable radical. The chromates, pyrophosphate, and 

 picrate of Purpureocobalt have, in particular, occupied our attention. 



We have mentioned, in speaking of the reactions of chloride of Purpureocobalt, 

 that both the cobaltidcyanide and the ferridcyanide of potassium give precipitates 

 in its solution. The constitution, crystalline form and physical appearance of 

 these two precipitates exactly agree with those of the cobaltidcyanide and ferrid- 

 cyanide of Roseocobalt, and we have, therefore, not hesitated to identify them with 

 these last. We believe that in this case there is a conversion of Purpureocobalt 

 into Roseocobalt, since in the salts in question there are three equivalents of 

 cyanogen in the electropositive for three in the electronegative cyanide, the form- 

 ulte being as mentioned above 



5NH 3 .Co 2 Cy 3 + Co 2 Cy 3 +3HO, and 5NH 3 .Co 2 Cy 3 +Fe 2 Cy 3 +3HO. 



As Purpureocobalt is certainly biacid, its cobaltidcyanide and ferridcyanide 

 should have the formulas 



3(5NH 3 .Co 3 Cy 3 ) + 2Co 2 Cy 3 , and 3(5NH 3 .Co 2 Cy 3 )+2Fe 2 Cy 3 , 



although the frequent occurrence of basic double cyanides may render this point 

 less clear than the others which also involve the biacid character of the radical. 



