1820.] " -D'". Thomson on Arsenic. 89 



finding the composition of any of them inconsistent with the 

 notion that an atom of arsenic weighs 14-5. 1 have not yet 

 satisfied myself sufficiently respecting the constitution of the two 

 varieties of arseniate of ammonia which are known to exist. But 

 from the experiments which I h.ive made, I may venture to say, 

 that they will not be found to militate against the number 14-5 

 as the equivalent for arsenic acid. U[)on the whole then the 

 prescHt state of our knowledge seems to make the choice of that 

 number preferable to 7-23, its half, which I adopted in the tifih 

 edition of my System of Chemistry. It has the advantage of 

 destroying the fractional parts of oxygen, which, on the suppo- 

 sition that 7-25 is the weight of an atom of arsenic, enter into 

 the composition of arsenious and arsenic acid. Vi e may, there- 

 fore, fix upon the following numbers as representing the weight 

 and composition of arsenic, arsenioiis and arsenic acid : 



Arsenic, weight of its atom, 9'5. 



Arsenious acid composed of 1 atom arsenic + 3 atoms oxygen 



= 12-5 weight. 

 Arsenic acid composed of 1 atom arsenic + 5 atoms oxygen 



= 14-5 weight. 



Article II. 



Experiments to determine the Composition of ilifferent inorganic 

 Bodies which serve as a Basis io the Ca/cu/ations relative to the 

 Theory of Chcvdcal Proportions.-''' By J. Beizelius. 



In a memoir published several years ago, I stated some 

 experiments by which I endeavoured to obtain results sufficiently 

 exact to serve for data, by means of which the composition of 

 various compounds might be calculated with more certainty than 

 they could be determined by analysis itself. As in these expe- 

 riments I employed methods, such that the results depended as 

 little as possible on the dexterity of the experimenter, 1 enter- 

 tained some hopes of attaining my object ; but 1 met with so 

 many difficulties inseparable from that enterprize that none of 

 my results could be considered as normal. 



After six years of continual experiments on this subject, and 

 after having acquired much experience and liaving discovered 

 several improvements in the analvticul methods, I resolved upon 

 resuming these researches, winch 1 consider as of the greatest 

 importance. My object has been not to obtain results which are 

 absolutely exact, which I consider as only to be ol)tained by 

 accident, but to approach as near accuracy as cliemical aualysi* 



* TianaUled from Afliaudlingar i Fysik, <S(c. v. 319. 



