Determination of Atomic Weights. 189 



possible in glass flasks specially constructed. This sal-ammoniac 

 sublimed without leaving a black residue. (2) Sulphate of am- 

 monium was heated with concentrated sulphuric acid until decom- 

 position commenced, nitric acid added so as to decolorize it, and 

 it was then treated with hydrate of lime in a bath of common 

 salt. The subsequent treatment was the same as the above. 

 (3) Nitrate of potassium prepared by Stromeyer's method was 

 heated with zinc and iron wire. The zinc had been freed from 

 charcoal by fusion with soda and saltpetre, or with litharge. 

 The iron wire had been heated in a current of air and then re- 

 duced with hydrogen. This ammonia, as well as that prepared 

 by methods (1) and (2), has simply a penetrating smell, and not 

 an unpleasant one, like the ammonia of commerce. The sal-am- 

 moniac prepared from the ammonia obtained by method (3) was 

 sublimed in vacuo in a long glass tube at as low a temperature 

 as possible. The sal-ammoniac used in the rest of the experi- 

 ments was always freed from an admixture of ammonia by being 

 heated until vapour began to form. 



Of all these kinds of sal-ammoniac definite weights were mixed 

 with silver solution at the ordinary temperature or at 100°. 

 For 100 parts of silver there were always 49*592-49*602 

 parts of NH 4 C1, the mean being 49*597. If sal-ammoniac and 

 silver were weighed in atomic weights (in round numbers 14 and 

 108), there was always an excess of silver in the solution, which 

 frequently amounted to one hundred times as much as the sources 

 of error in the method. Hence neither temperature nor pressure 

 have any influence on the composition of sal-ammoniac. 



To prove the unchangeability of the relative weights of the 

 elements which form chemical combinations, the author reduced 

 chlorate, bromate, and iodate of silver by sulphurous acid. In 

 the chloride obtained, chlorine and silver were present in exactly 

 the same ratio as in chlorate of silver. 



Iodate of silver prepared by precipitating iodate of potassium 

 with nitrate of silver is impure. It always contains an admix- 

 ture of nitrate of silver w T hich is not to be removed by washing. 

 Only in case of precipitation with sulphate or hyposulphate of 

 silver is a chemically pure preparation obtained. Reduction 

 with sulphurous acid was effected under complete exclusion of air, 

 and only with freshly prepared acid. Pure sulphurous acid 

 produces in a solution of nitrate or sulphate of silver a white 

 precipitate of sulphite of silver. But if an acid be used which 

 has been exposed to the light, the precipitate is grey, owing to 

 the admixture of sulphide of silver. Such an acid is totally un- 

 fitted for reducing iodate of silver. 



Bromate of silver must not be prepared with nitrate of silver. 

 Bromate of potassium is precipitated with sulphate or hypos ul- 



