AMMONIA-COBALT BASES. 5 



the salt before dropping the acid upon it. The quantity of salt to be taken for 

 analysis may vary from three to five decigrammes ; when more is used, there is apt 

 to be some loss from effervescence. In consequence of the small quantities of sub- 

 stance employed, the weighings must be as accurate as possible. In calculating the 

 weight of the cobalt from that of the sulphate, we have the advantage of determin- 

 ing one substance from another with an equivalent more than twice as high. 



In certain cases, as, for example, when phosphoric acid, chromic acid, &c, are 

 present, the above method cannot be employed. In such compounds we have found 

 it advantageous to separate the cobalt as a hydrate of the sesquioxide, by boiling 

 the salt with a solution of caustic potash, washing the precipitate thoroughly, and 

 estimating the ignited precipitate as Co 6 7 , or as metallic cobalt after reduction by 

 hydrogen. Fremy justly observes that this ignited oxide usually contains potash; 

 but an accurate result may always be obtained by washing it well with boiling 

 water after the ignition, and weighing a second time. It is remarkable that Fremy 

 asserts that cobalt may be accurately estimated in the form of sulphate, in conse- 

 quence of the stability of this salt, while the direct application of the method, as 

 we have described it above, appears to have escaped him entirely. 



Hydrogen. "We have in almost all cases determined hydrogen directly by com- 

 bustion with chromate of lead, metallic copper being placed in the anterior part of 

 the tube. In the case of the nitrates, however, an excess of hydrogen in the result 

 is almost unavoidable, because it is impossible, even with freshly reduced copper, 

 to decompose completely the great quantity of oxides of nitrogen formed during the 

 combustion. In other cases this effect is much less marked, and the hydrogen 

 determinations are at least as accurate as in ordinary organic analyses. 



Chlorine. The accurate determination of the chlorine in the ammonia-cobalt 

 salts is very difficult. Nitrate of silver, it is true, precipitates chlorine from most 

 of its combinations in these salts, but the precipitation is never complete, because 

 the chloride of silver is somewhat soluble in the ammonia-cobalt chlorides, forming 

 with them peculiar double salts. By long boiling with free nitric acid in the solu- 

 tion, nearly all the chlorine may be determined as chloride of silver, but very 

 accurate results cannot be obtained in this manner. The best method consists in 

 igniting the chloride with lime in a combustion tube, in the manner usually prac- 

 tised with organic bodies. In some cases, however, we have obtained very good 

 results by decomposing the solution of the chloride by sulphurous acid, or by boil- 

 ing the solution of the salt until it is completely decomposed, adding sulphurous or 

 nitrous-nitric acid to reduce the sesquioxide of cobalt, and then precipitating with 

 silver. The process is, however, always troublesome, and requires much time and 

 great care. 



Carbon. This element is best determined by the usual processes of organic 

 analysis. In consequence, however, of the very large quantity of oxides of nitro- 

 gen, which are always produced during the combustion of these salts, we have 

 found it very advantageous to employ a method first suggested, we believe, by 

 Winkelblech, and which consists in mixing with the oxide of copper a quantity 

 of finely divided metallic copper, in the form in which it is obtained by reducing 

 the oxide by hydrogen. In this manner the formation of the oxides of nitrogen 



