ATOMIC WEIGHTS 257 



Eamsay and Aston/ in their paper npon the atomic weight of l)or()n, 

 suffo-est that Abrahall's bromide mav have contained hvdrobromic acid, 

 which would fully account for the low result obtained. They them- 

 selves adopt two distinct methods, the first one being the time-honored 

 determination of water in crystallized borax. The latter was prepared 

 from pure boric acid and pure sodium hydroxide. Eesults as follows, 

 reduced to a vacuum : 



Mean, 47.1677, ± .0086 



Hence B = 10.942. 



The second method adopted by Eamsay and Aston was to distill anhy- 

 drous borax with hydrochloric acid and methyl alcohol, both scrupu- 

 lously pure, thereby converting it into sodium chloride. The operation 

 was conducted in a glass flask, and in the first series of determinations 

 ordinary soft glass was used. This, however, was somewhat attacked, 

 so that the sodium chloride contained silica; hence oxygen in the ma- 

 terial of the flask had been replaced by chlorine, thereby increasing its 

 weight and lowering the apparent atomic weight of boron. In a second 

 series flasks of hard combustion tubing were taken, and the error, though 

 not absolutely avoided, was reduced to a very small amount. Both series 

 are subjoined, together with the percentage of chloride formed; but the 

 weights, given by the authors to seven decimal places, are only quoted to 

 the nearest tenth milligramme. They are reduced to a vacuum standard : 



1 Journ. Chem. Soc, 63, 211. 1893. 



