OF ARTS AND SCIENCES. 



293 



V. Summary op Results, 1887 to 1891. 

 The Atomic Weight of Copper. 



Of these results five — namely, I., II,, III., V., and VI. — are 

 incomparably more trustworthy than the others. Their computation 

 involves only very accurately determined elements, and they are least 

 affected by a given change in the molecular weights of the standards 

 of reference. Their experimental errors are far smaller than those 

 of any of the others, and they involve the simplest and most direct 

 processes, and the minimum number of necessary corrections. They 

 are in no case computed by the objectionable method of difference. 

 Finally, each one of the five is wholly independent of the weight of 

 any copper compound whatsoever. These reasons constitute a suffi- 

 cient ground for the separation of such results from the others. 



