302 HISTORICAL INTRODUCTION TO CHEMISTRY CHAP. 



as 20 per cent, in extreme cases. The exact analytical work 

 of Berzelius and others showed that constancy of composi- 

 tion could be relied on within very narrow limits ; but the 

 suggestion was made, in 1860, by the Swiss chemist Marignac 

 (Geneva Archives, 1860, 9, 103) that small variations of 

 composition might perhaps exist, even in the most stable 

 and definite of chemical compounds. This suggestion was 

 submitted to a most stringent test by Stas, who published a 

 memoir on the subject in 1865 ( Works, I. 445-481). Stas 

 made a very careful comparison of the quantities of common 

 salt required to convert different samples of silver into 

 silver chloride. The standard was a sample of silver which 

 had been distilled with the help of an oxy-hydrogen blowpipe. 

 The relative quantities of salt were as follows : 



Method of Preparation. Salt used. 



Distilled silver 100*000 



Silver electrolysed from a cyanate solution ; metal I .^ \ 

 USCd \99'997 



f 99' 994 



Silver nitrate reduced by milk-sugar ; metal fused . -j 99*995 



1 99 '999 



Silver nitrate reduced by sulphites ; metal fused 99'997 

 Silver chloride reduced by fusing with sodium car- 

 bonate and nitrate 99 995 



Silver chloride reduced by heating with charcoal and 

 lime . 99 '991 



In these experiments ( Works, I. 466) the average deviation 

 from the mean value amounted only to 0*002 per cent. 



A similar series of experiments (Works, I. 478) was made 

 on the precipitation of silver by means of different samples 

 of ammonium chloride, both at atmospheric temperatures 

 and at 100. The results were as follows : 



Quantity used to 

 precipitate 100 parts 



Method of Preparation. of silver. 



Chloride prepared from cold solutions of hydro- ( 49 '600 



chloric acid and ammonia. Ratio determined at -I 49 '599 



ordinary temperature . . . . . . [ 49*598 



