PROCEEDINGS OF THE AMERICAN ACADEMY 



A cursory glance at the list will show a lamentable lack of con- 

 sistency in the results of even a single method in different hands. 

 The only ratio which seemed capable of yielding apjiroximate re- 

 sults was the ratio of baric chloride to metallic silver, and here the 

 variations in the atomic weight of barium amounted to nearly three 

 tenths of a unit. The question whether the errors were due to me- 

 chanical defects of analysis, or to admixture of foreign substances, 



* In Clarke's original treatise 137.007 is apparently misprinted for 137.07 

 (Smithsonian Misc. Coll., Vol. XXVII. p. 63). 



