OF ARTS AND SCIENCES. 247 



inference confirmed by the analysis of the chloride prepared by the 

 action of strong hydrochloric acid and alcohol on the substance, which 

 gave the following result. 



0.1334 gr. of the salt gave 0.0916 gr. of argentic chloride. 



Calculated for (CuUsljNHaCl. Found. 



Chlorine 17.27 " 16.97 



The formation of diphenylamine by this reaction would seem to 

 show that the sodium amide used by us was a mixture of NaNH.^ with 

 Na^NH, whereas Beilstein and Geuther proved that the formula of 

 their sodium amide was NaNHg- The explanation of this difference 

 probably can be found in the fact that our amide still contained an 

 excess of sodium, the flasks used by us not being made of a sufficiently 

 resistant glass to allow us to carry the reaction to an end, while in 

 their experiments all the sodium was converted into the amide. It is 

 also possible, but not probable, that the diphenylamine was formed by 

 a secondary reaction from the aniline. 



Some experiments to determine the temperature at which the re- 

 action takes place showed that some amine was obtained at 280°, and 

 the yield was not improved at 300-350° ; but that a much better 

 result was reached by heating over the free lamp than with an oil- or 

 air-bath, the products at the temperatures given above amounting re- 

 spectively to 6.8 and 5.4 per cent of the theoretical, or about one 

 half of the amount obtained by the more rapid direct heating over the 

 lamp. 



Action with Potassic Benzolmetadisulphonate. — The salt mixed with 

 about the equivalent amount of sodium amide was heated in a test- 

 tube as before ; but it was found necessary in this case to continue the 

 heating until the brown spots which formed at first had turned black, 

 and showed signs of decomposition, since, if the action was stopped 

 while they were still brown, as in the case of the monosulphonate, the 

 yield was exceedingly small. At best, the yield was not satisfactory, 

 reaching at the outside five per cent of the theory, so that we did not 

 attempt to collect enough of the tolerably unstable chloride of the 

 metaphenylenediamine for analysis, but contented ourselves with prov- 

 ing its presence by qualitative tests. For this purpose, the product of 

 the reaction was cautiously treated with water, and the aqueous solu- 

 tion extracted with ether ; one portion of the ether extract was shaken 

 with dilute sulphuric acid, and to the acid solution thus obtained a 

 very minute amount of sodic nitrite was added, which turned it yellow ; 

 another portion of the ether extract was shaken with dilute hydro- 



