212 PROCEEDINGS OF THE AMERICAN ACADLMY 



Each aiinlytical result is combined with every other result above, 

 in order that a better idea may be obtained of the variations of the 

 figures. The combination of the averages of tlie three series of anal- 

 yses would of course have given the same final average in a much 

 more simple manner. Of the tabulated experiments, No. 23 is the 

 least reliable, for the reasons before given ; and rejecting values VII. 

 and X. upon this ground the total average would become G3.648, with 

 a "probable error" of about ±0.002. 



The importance and value of the method of analysis, involving as 

 it docs the weighing of the silver used as well as of the argentic bro- 

 mide formed, must be at once evident. Not only does the method 

 give two wholly distinct classes of results for the atomic weight of 

 copper, but besides this the relative weights of silver and bromine 

 afford the sharpest possible test of the purity of the materials and the 

 accuracy of the work. This method was first used by Professor Cooke 

 in verifying the atomic weight of antimony, and he strongly insisted 

 on the strength of the evidence thus furnished;* although the point 

 has been strangely overlooked by commentators. 



Percentage of Silver in Argentic Bromide. 



From Experiments 20 and 23 = 57.454 



" " 21 " 24 = 57.448 



" " 22 " 25 = 57.447 



Average = 57.450 — 



Average rejecting Exp. 23 as before = 57.448 — 

 According to Stas = 57.445 



Previous determinations made in this Laboratory by various experi- 

 menters have given values ranging from 57.442 to 57.450. 



The agreement of the present value of the atomic weight of copper 

 with the result which I previously published in these Proceedings f is 

 a rather remarkable one, especially since the methods of work were 

 wholly different in the two oases. The very slight difference between 

 the preliminary and final results of this paper has already been ex- 

 plained, so that all the experiments which have been tried in this 

 laboratory point to precisely the same figure. 



* These Proceedings, xvii. 19; also Huntington on Atomic Weight of Cad- 

 mium, these Proceedings, xvii. 28. 

 t Vol. x.xiii. p. 180. 



