202 M. Stas on the Determination of Atomic Weights. 



vescence. "When all the gas was expelled, the flask was inclined, 

 and the nitric acid distilled off at as low a temperature as possible. 

 To prevent the salt from creeping, the flask was surrounded by a 

 cylinder of tin plate, so that only part of the bottom covered 

 with liquid was directly heated. The dry residue was now slowly 

 heated to melting in a current of dry air, and by rapid cooling 

 suddenly made to solidify. The flask was now weighed and treated 

 a second time in the same manner. A careful examination of 

 the water in the flask, as well as of the nitric acid distilled off, 

 showed that no appreciable particles had been carried over. 



Chloride of potassium. — The salt investigated was prepared in 

 three ways : — 



(a) Nitre was recrystallized ten times, and decomposed by 

 hydrochloric acid. 100 parts of this chloride of potassium 

 gave 135 '643 parts of nitrate of potass. 



(b) Chlorate of potassium was decomposed by ignition in a 

 platinum retort. As the chloride of potassium contained traces 

 of platinum and of silica, it was ignited with sal-ammoniac, dis- 

 solved, and the evaporated residue fused. 100 parts of chloride 

 of potassium gave 135*638 of nitrate of potass. In another 

 preparation the fused salt was twice dissolved, again evaporated 

 and fused. 100 parts of chloride of potassium then furnished 

 135-647 of nitrate of potass. 



(c) Chloride of potassium was prepared from platino- chloride 

 of potassium. Five pounds of platinum were dissolved in aqua 

 regia and mixed in the dark with lime, then, after being acidified, 

 precipitated by chloride of potassium, and the fully washed pre- 

 cipitate ignited with carbonate of potassium and sodium. The 

 washed platinum was a second time dissolved, and again pre- 

 cipitated by chloride of potassium. The precipitate, washed and 

 boiled with water, was divided into three parts by partial solution. 

 At first one- third of the whole mass was dissolved, then one half 

 of the residue, while, finally, an undissolved residue remained. 

 All three portions, dried at 200°, were reduced in a current of 

 hydrogen, whereby water was given off, which thus only com- 

 pletely escapes when the salt is decomposed. Hence this salt is 

 not suited for determining the atomic weight of platinum. 

 The chloride of potassium washed out was evaporated in a 

 platinum retort and fused. Of the three portions of chloride of 

 potassium 100 parts gave, 135*649, 135645, 135*640, and 

 135*655 parts of nitrate of potassium. 



Chloride of potassium (a) examined in the spectrum-apparatus 

 gave a very distinct sodium reaction. A direct analysis made with 

 4 grammes of the salt, showed, however, that the quantity of so- 

 dium did not amount to more than from ^ to — of a milligramme. 

 Chloride of potassium (b) gave also the sodium-line, as also did 



