S2 Berzeliiis's Experiments to determine the Composition [Feb. 



In the first 39" 146 of oxygen gas 



second 39' 150 



third 3y-1.50. 



fourth 39-149 



These experiments vary only in the ten-thousandth parts, and 

 two of them even agree in them. We may, therefore, con;>ider 

 it as exceedingly near the truth that lUO parts of oxymuriate of 

 potash, when pure and dry, give out 39"l') parts of oxygen 

 gas. 



To be able to make use of this result, it is necessary to know 

 the exact composition of nuuiate of potash. I took, therefore, 

 muriate obtained by these experiments, in which the best 

 reagents could not detect the least excess of alkah, and I decom- 

 posed it by a solution of crystallized nitrate of silver : lU grammes 

 of muriate of potash gave in two experiments 19*24 grammes of 

 fused muriate of silver; but it was necessary to know with the 

 same degree of precision how much nmriate of silver is obtained 

 from a given weight of silver. DiH'erent chemists have made 

 experiments on the subject, but the results do not always agree 

 with each other. Those who have come nearest the truth I 

 consider as the following. Wenzel found that 100 silver gave 

 131*4 of muriate of silver; Davy, 132"5.; Bucholz, Rose, Marcet, 

 and Gay-Lussac, 133"3 ; and finally, in my former experiments 

 already described, I found from 132-7 to 132' 75 ; for the degree 

 of accuracy requisite in experiments which are to serve as a basis 

 to our calculations, the diti'erence between 132'5 and 133*3 is 

 too much. Even the difference between 132*7 and 132*75 has 

 a very sensible influence on the results deduced from it. It was, 

 therefore, necessary to examine if the results of my old experi- 

 ments were exact ; and in that case which of the two was nearer 

 the truth. 



1. 1 dissolved silver purified with the requisite care in pure 

 nitric acid in an inclined phial, and the liquid was evaporated to 

 .dryness in the phial to get rid of all excess of acid. The nitrate 

 of silver was then dissolved in water, and the clear solution was 

 poured into 'a solution of sal-ammoniac .. The precipitate was 

 collected on a filter, well washed, dried, and then fused upon a 

 watch glass of a known weight. A current of sulphuretted 

 hydrogen gas was then passed through the filtered liquid, but no 

 trace of silver could be discovered. Twenty grammes of silver 

 gave in this way 26*54 grammes of muriate of silver ; that is to 

 say, 100 of metal gave 132*7 of muriate. This experiment can 

 only err by a loss, since any excess of weight is impossible. 



2. To compare this result with that of an experiment in which 

 no loss could be sustained, and in which consequently the error 

 must be in excess, I dissolved pure silver in an inclined phial, 

 evaporated the solution to dryness, redissolved the nitrate, in 

 water, and added to it in the same phial pure muriatic 





