136 THE ATOMIC WEIGHTS. 



the mean, would make the atomic weight of copper 63.087, 

 dr .222. This figure has so high a probable error that we 

 need not consider it further. 



The results obtained by Dumas * are wholly unavailable. 

 Indeed, he does not even publish them in detail. He merely 

 says that he reduced copper oxide, and also effected the 

 synthesis of the subsulphide, but without getting figures 

 which were wholly concordant. He puts Cu = 63.5. 



Latest of all, and probably the best also, we have the de- 

 terminations by Hampe.f First, he attempted to estimate 

 the atomic weight of copper by the quantity of silver which 

 the pure metal could precipitate from its solutions. This 

 attempt failed to give satisfactory results, and he fell back 

 upon the old method of reducing the oxide. From ten to 

 twenty grammes of material were taken in each experiment, 

 and the weights were reduced to a vacuum standard : 



20.3260 grm. CuO gave 16.2279 grm. Cu. 79.838 per cent. 



20.68851 " 16.51669 " 79-835 " 



10.10793 " 8.06926 " 79-831 " 



Mean, 79.8347, zh -0013 



Hampe also determined the quantity of copper in the 

 anhydrous sulphate, CuSO^ . From 40 to 45 grammes of the 

 salt were taken at a time, the metal was thrown down by 

 electrolysis, and the weights were all corrected. I subjoin 

 the results : 



40.40300 grm. CuSO^ gave 16.04958 grm. Cu. 39.724 per cent. 

 44.64280 " 17-73466 " 39-726 " 



Mean,' 39.725, =b .0007 



We now have four series of experiments upon copper 

 oxide, as follows : 



Berzelius 79-823, ±.0020 



Erdmann and Marchand 79.8645, dz .0038 



Millon and Commaille 79-778?, zfc .0043 



Hampe 79.8347, ± .0013 



General mean 79.830, ^b -ooio 



*Ann. d. Chim. at Phys., (3,) 55, 129. 

 f Fresenius' Zeitschrift, 13, 352. 



