248 PROCEEDINGS OF THE AMERICAN ACADEMY 



Acids. — The chemically pure hydrochloric and nitric acids were 

 each distilled three times successively in a platinum still. Large 

 quantities evaporated in open dishes left unweighable and scarcely 

 visible residues. The insignificant amount of the residues was a proof 

 of the purity of the air, as well as of the acids. Neither acid gave a 

 test for iron, and the nitric acid contained no chlorine even at the 

 beginning of the operations. 



Since the platinum still was very large, it was found more con- 

 venient to distil sulphuric acid from a small hard glass retort heated 

 by a ring burner, immediately before it was needed. After three 

 such distillations, a preparation was obtained which left no weighable 

 residue upon the evaporation of any quantity used in the work. 



All sulphuric acid used for drying was boiled with ammonic 

 sulphate. 



In order to keep the air of the laboratory quite pure, during the 

 greater part of the time neither volatile acids nor ammonia were 

 allowed in the room devoted to the investigation, and all available 

 precautions were taken against dust. 



Indicators. — As is well known, the salts of most of the heavier 

 metals are acid toward phenol phthalein and neutral toward methyl 

 orange. The most important consequence of this relation has already 

 been pointed out. When the former indicator was used, the slightly 

 acidified solution was of course first freed from carbonic anhydride 

 by long continued heating upon the steam bath, and the end point was 

 determined with pure caustic alkali. It is a noteworthy fact, that 

 methyl orange is useless in very concentrated solutions of sodic sul- 

 phate, the color change becoming apparent only upon dilution. This 

 indicator is naturally less serviceable in the presence of a strong color 

 like that of a copper salt. It is less sensitive with cupric sulphate 

 than with cupric bromide. In such cases as these, colorimetric com- 

 parison can alone afford accurate results ; but even here the effect of 

 a personal equation must be more or less perceptible. 



Since the amount of either indicator used in any one case was not 

 more than the thirtieth of a milligram, it could not have seriously 

 influenced subsequent operations with the solution. 



Water. — The distilled water taken hot from the tin-lined condenser 

 around a steam-drying oven contained no ammonia discoverable by 

 Nessler's reagent, and in the first crude experiments this water was 

 used without further treatment. After having been once more dis- 



