72 



PROCEEDINGS OP THE AMERICAN ACADEMY. 



SERIES IV. 



and three wholly distinct samples of silver were used in this series, we 

 may conclude that all ordinary accidental errors had been eliminated ; 

 and in a critical discussion of the result wer may limit ourselves to the 

 consideration of the possible constant errors of the process. 



The most serious objection to the method is, of course, the possible 

 retention of water, of magnesic oxycliloride, or of ammouic chloride by 

 the magnesic salt. 



With regard to the first two difficulties, it need only be said that the 

 gases used for drying the magnesic chloride were as dry as present pos- 

 sibilities permit them to be made. Tlie phosphorus pentoxide in the 

 last drying tube showed no trace of liquefaction at the close of the 

 research, but seemed to be as light and powdery as at first, in spite of 

 the fact that several hundred litres of gas had been passed over it. Any 

 trace of oxygen, as well as of aqueous vapor, was excluded from the hot 

 salt; for the hydrochloric acid gas was replaced by nitrogen, and this 

 was driven out in its turn by dry air only after the tube had cooled. A 

 means of proving absolutely that no water remained does not exist ; but 

 it is extremely hard to see how water could have gained access to the 

 carefully guarded magnesic chloride. 



The fact that every sample of magnesic chloride used in the last series 

 gave an absolutely clear an'd transparent solution in water is additional 

 evidence of much weight ; for a very small trace of oxychloride would 



