6 RESEARCHES ON THE 



may be completely prevented. Great care must, however, be taken when it is 

 wished to determine hydrogen at the same time with carbon, because, copper re- 

 duced from the oxide by hydrogen, always contains water, which it is difficult to 

 separate. 



Nitrogen. No element has presented such difficulties as nitrogen. "We have 

 found it impossible to obtain results within two or three per cent, of the truth by 

 employing the old methods of analysis, that of Dumas for instance. The quantity 

 of nitric oxide formed during the combustion is surprising, and it is absolutely im- 

 possible to get rid of it by means of ignited metallic copper, placed in front of the 

 combustion tube. Will and Varrentrapp's method with soda lime is inapplicable, 

 because one equivalent of ammonia is always decomposed by the equivalent of oxy- 

 gen set free in the reduction of sesquioxide to protoxide of cobalt. Good results 

 could not be obtained by boiling the salts with caustic alkalies, collecting the 

 ammonia in chlorhydric acid, and determining it by bichloride of platinum. Even 

 after the reduction of the sesquioxide of cobalt to protoxide by means of sulphur- 

 ous acid, this method was found unreliable. The improvements made by Simpson 

 in the absolute determination of nitrogen by volume at last furnished us with a 

 reliable process ; and nearly all the analyses in this memoir were executed by 

 his method. The improvement introduced by Simpson consists essentially in mix- 

 ing oxide of mercury with the oxide of copper employed to effect the combustion. 

 The vapor of metallic mercury completely decomposes the oxide of nitrogen, and 

 any excess of free oxygen is absorbed by means of metallic copper. By this 

 method we have analyzed most of our compounds without special difficulty, though 

 we have often found it necessary to employ a much larger proportion of oxide of 

 mercury than is recommended by Simpson. One class of ammonia-cobalt bases 

 have, however, been the source of frequent analytical failures, and of great loss of 

 time and material. We refer to the salts of Xanthocobalt, a base containing deut- 

 oxide of nitrogen, and giving off this gas at a gentle heat, below that at which 

 oxide of mercury is decomposed. Simpson's method has not always been found 

 accurate, since even when a very large amount of oxide of mercury is employed 

 there is frequently much nitric oxide in the nitrogen collected for measurement. 

 In many cases the simple admixture of a large proportion of metallic copper with 

 the oxide, as recommended by Wmkelblech, has been found to give most excellent 

 results. It is proper also to state here that, in consequence of difficulties in 

 obtaining proper apparatus with which European chemists do not have to contend, 

 we have, in the majority of cases, measured the volume of nitrogen in the old way, 

 using, however, very accurately graduated tubes for collection, and correcting with 

 great care for temperature and pressure. "We have also found it advantageous to 

 operate upon quantities of substance sufficient to yield at least two hundred cubic 

 centimetres of gas, since in this way the error of reading becomes extremely small. 



Sulphuric Acid. This acid cannot be accurately determined in the ammonia- 

 cobalt salts by direct precipitation with chloride of barium. In almost all cases, a 

 great apparent excess of acid is obtained, and this may amount to five per cent., 

 even when the sulphate of baryta appears to have been completely washed. We 

 have, therefore, in all cases preferred to decompose the salt to be analyzed, by 



