64 PROCEEDINGS OF THE AMERICAN ACADEMY. 



cooled in the reducing flame. Particular 25ains were taken to prevent 

 the absorption of oxygen, and the button did not show the slightest trace 

 of having contained this gas. From this portion wire was prepared of 

 various thicknesses, by means of a draw plate ; and the weights of given 

 lengths of these wires were determined, so that small weights could be 

 made with considerable accuracy. Of course the wire was treated in the 

 same fashion as the rest of the silver, in order to remove any iron which 

 might be present on the surface. 



The third and fourth samples of silver were prepared in the same man- 

 ner as the second, the starting point being the pure residues left from the 

 analyses made with previous samples. No qualitative nor quantitative 

 difference could be observed between any of these preparations of silver. 

 Fused upon sugar charcoal, they melted to a clear globule free from any 

 film, — a fact which in itself, according to Stas, is an excellent test of the 

 purity of silver, — and all gave practically the same results in later 

 determinations. 



All water used was redistilled with potassic permangajiate, some of it 

 being condensed in a platinum condenser and some of it by means of a 

 tube of pure block tin, which was carefully tested in order to prove the 

 absence of an impurity of lead. Considerable quantities upon evaporation 

 in platinum left a scarcely appreciable residue, there being apparently no 

 difference between the water condensed in tin and that in platinum.* 

 The water was prepared as short a time as possible before being used, 

 and was carefully kept in a suitable bottle fitted with a siphon, air being 

 admitted to the bottle through a filter of cotton wool. It was carefully 

 tested for chlorine by means of the nephelometer from time to time. 



The sulphuric acid used for the preliminary drying of the gases 

 was the usual " chemically pure " acid of the Laboratory, of a specific 

 gravity of about 1.83. For the final drying this acid was boiled down in 

 platinum. 



Weighing. 



The balance used was a long-armed Becker, sensitive to about one 

 thirtieth of a milligram with the largest load that it was required to carry 

 during the investigation, while the weights were a good set of gold plated 

 ones, which were kept in the balance case under a glass cover. These 

 weights were very carefully compared with one another, and all weigh- 

 ings were, of course, reduced to the vacuum standard. The specific 



* See These Proceedings, XXVI. 249; XXX. 380. 



