26 RESEARCHES ONT HE 



Hence we have 

















Eqs. 





Theory. 



Mean. 





Found. 





Cobalt . 2 



59 



23.55 



23.56 



23.58 



2'3.57 



23.55 



2355 



Chlorine . 3 



106.5 



42.50 



42.43 



42.49 



42.31 



42.52 



42.40 



Hydrogen . 15 



15 



5.98 



6.11 



6.04 



6.19 



6.10 



6.11 



Nitrogen . 5 



70 



27.97 



28.01 



28.05 



27.93 



28.12 



28.11 



250.5 100.00 100.11 



The agreement of these analyses leaves no reasonable doubt that the true form- 

 ula of the chloride of Purpureocobalt is 5NH 3 .Co 2 Cl 3 , as first correctly determined 

 by Rogojski,'and subsequently by Gregory. Fremy gives in addition one equivalent 

 of water, while Claudet makes 16 in place of 15 equivalents of hydrogen. That 

 the salt, however, contains but 15 equivalents is clear, from the fact that free 

 nitrogen and hydrogen are found among the products of its decomposition by heat 

 in an atmosphere of carbonic acid gas, which could not be the case upon Claudet's 

 view, since we should then have the equation 



N 6 H 16 Co 2 CI 3 = 5NH 3 + 2CoCl+ HC1, 



while the presence of free nitrogen and hydrogen renders it probable that the 

 decomposition is in reality expressed by the equation 



5NH 3 .Co 2 Cl 3 =2CoCl + NH 4 Cl+3NH 3 +N+H 2 . 



We have more than once endeavored to determine the amount of gas actually 

 given off during this decomposition, with the view of verifying the equation just 

 given. In every case, however, a portion of the chloride of cobalt was reduced, 

 either b} r the free ammonia or by the hydrogen, so that much metallic cobalt was 

 found mixed with the chloride. A neutral solution of the chloride of Purpureo- 

 cobalt is readily decomposed by boiling, a dark-brown precipitate, probably of the 

 hydrated magnetic oxide, being thrown down, while the solution becomes brown- 

 yellow, and contains chloride of ammonium and chloride of Luteocobalt, ammonia 

 being at the same time given off. The quantity of chloride of Luteocobalt which 

 is thus formed is always very small, being very much less than one equivalent 

 for two equivalents of the chloride of Purpureocobalt. 



On the other hand, a solution of chloride of Purpureocobalt may be boiled for a 

 very long time with concentrated chlorhydric acid without decomposition, and this 

 stability in the presence of acids is one of the most remarkable peculiarities of the 

 whole class of ammonia-cobalt salts. 



Chlorhydric acid and the alkajine chlorides precipitate chloride of Purpureo- 

 cobalt from its solutions almost completely, slowly in the cold, but instantly on 

 boiling. Ignited in a current of hydrogen, the salt yields metallic cobalt as a gray 

 spongy mass. Heated in an open crucible, the salt fuses and swells up, giving off 

 abundant vapors of chloride of ammonium and ammonia, while pure chloride of 

 cobalt remains in lavender-blue scales. In some cases, however, this is mixed with 

 metallic cobalt, while in others, in which the ignition takes place with free access 

 of air, brilliant iron-black octahedra are formed, which are the anhydrous magnetic 

 oxide of cobalt, Co 3 4 . The red gas arising from the action of nitric acid upon 



