138 PROCEEDINGS OF THE AMERICAN ACADEMY. 



should yield yet higher values, and hence that the pi of < uprous 



Ball really causes the deposit to be too heavy. 



The results of all these varied operations assign a wide range of 

 apparent values to the atomic weight of copper. In order to present the 

 matter in a nutshell, a table is given below. 



| 68.80 

 Uncorrected results of Rayleigh, Gray, Shaw, Vanni, etc. . to 



( 50 

 63.47 



525 

 63.547 

 63.573 

 63.615 



Large plates in cupric solutions, 20° (present research, i 



. . . . .. Q° . . ( . 



Small - " " " " 



.Mnliiun plates, in cuprous solutions, 0° " " 



4t >. ». IS QQ° (( (t 



Corrected results from cupric solutions 



Atomic weight of copper by chemical processes, . 



Discrepancy . 



63.563 



0.041 



Tt has already been stated that the corrected value is probably a little 

 too high, although it is still nearly 0.07 of one per cent below the chem- 

 ical value. This discrepancy, taken in connection with the inexpli- 

 cable variations in the results of parallel determinations with the silver 

 voltameter (see page 135), led us to suspect this instrument long before 

 the results given above were completed. Accordingly, many of the 

 electrolyses which we have been discussing were also used as a means 

 of testing the accuracy of the silver voltameter. The overlapping inves- 

 tigations have been separated in the description only in order to make a 

 very complex matter less obscure to the reader. Tin- outcome of the 

 experiments upon the silver voltameter is explained below. 



II. Tiik Silver Voltameter. 



The inference to be drawn from the preceding work is that the amount 

 of silver deposited in the silver voltameter, as described by Lord Rayleigh 

 and Mrs. Sidgwick* in their classical paper, may be too great by nearly 

 0.1 per cent. In spite of the great care taken by these investigators, 

 such a deviation from the exact value is not impossible ; and indeed 

 Lord Rayleigh does not claim for this work an accuracy over 0.1 per 

 cent. 



The first possible cause of error to be investigated was the possible 

 occlu>ion (or rather inclusion) of electrolyte by the crystalline deposit. 



* Phil. Trans., 2. Ill (1884). 



