1904.] RICHARDS — THE ATOMIC WEIGHT OF NITROGEN. 117 



Stas himself concluded from these results that nitrogen was 

 almost certainly higher than 14.03, and probably about 14. 045, 

 basing his conclusion upon a somewhat doubtful application of the 

 theory of least squares.^ 



In spite of the great care taken by Stas in this unusually extended 

 investigation, it is of course not impossible that small constant 

 errors might have existed in parts of the work. Stas was by no 

 means infallible ; his long oversight of the solubility of argentic 

 chloride, the uncertainty concerning the amount of oxygen occluded 

 by his silver, and his frequent use of glass vessels somewhat attacked 

 by his reagents for long-continued operations, being among the 

 evidences that he too was mortal. Nevertheless, it is true that Stas 

 was more precise than any one who preceded him ; and his results 

 cannot be overthrown without much conclusive experimental 

 evidence. 



Three years ago the accuracy of one of these series" of experi- 

 ments made by Stas was impugned by Alexander Scott, namely, 

 the series in which ammonic bromide was compared with silver. 

 The atomic weight is computed from the result of these experi- 

 ments by subtracting the weight of the bromine precipitated as 

 silver bromide from the weight of the ammonium bromide, in order 

 to find the weight of ammonium present. Because the bromine is 

 equivalent to the silver, the ammonium previously united to the 

 bromine must also be equivalent, and upon assuming the atomic 

 weight of silver to be 107.93, ^^^ molecular weight of ammonium 

 is easily found to be 18.078. Subtracting from this four times the 

 atomic weight of hydrogen, that of nitrogen remains — namely, 

 14.048 — because ammonium consists solely of nitrogen and 

 hydrogen. 



Now Scott contended, with plausibility, that this particular spe- 

 cimen of ammonium contained impurities, substances other than 

 nitrogen and hydrogen, because Stas admits that his bromide was 

 not perfectly colorless. 



These impurities would all be included in the estimate of the 



for nitrogen (14.01 to 14.02), a discrepancy which it is not easy to explain, 

 unless the chloride was precipitated from a solution so concentrated as to occlude 

 nitrate. The lack of details in his description makes it impossible to decide 

 this question. 



1 Stas, Untersuchungen (Aronstein), p. 322 (1S67). 



2/. Chem. Soc. Trans. ,79, 147 (1901). 



