Polarization at a Metallic Anode. G67 



The agreement between the observed and calculated 

 polarizations is a< close as in the case of the silver-nitrate 



cells. 



Discussion of Results. 



The foregoing results show that the formula? obtained 

 represent the polarizations throughout with considerable 

 accuracy, both as regards the absolute magnitude of the 

 polarization and its rate of variation under very varied 

 conditions of polarizing current. This may be taken as 

 establishing the general correctness of the theory, at any rate 

 as applied to the anodic polarization of the class of cell con- 

 sidered. But although a formula of the type of equation (22) 

 is undoubtedly correct in its main features, I think it is 

 possible that the numerical agreement found is to a certain 

 extent a spurious one. For one thing, it turns out to be a 



closer one than we have any right to anticipate. The value 



-p>T 



•0578 volt of the factor ..... x 10~ 8 , on which the calcu- 



lation of the polarization in the silver-nitrate cells is based, is 

 invariably taken in all theoretical work on concentration-cells 

 a- accurately representing the E.M.F. of a concentration-cell 

 the concentration of the metallic ions of which at one elec- 

 trode is ten times that at the other ; yet in actual practice 

 the number obtained is always appreciably smaller. Thus a 

 direct measurement of the E.M.F. of the cell 



Ae 



•002 n. AgNO, -05 n. AgNO. 



?5 



2n. HN0 3 2n. HNO 



Ag 



gave the value '074 volt at 19° C, which corresponds to a 

 value of the factor of "052. There is no reason to suppose 

 that this diminution in the factor us actually measured below 

 its theoretical value does not hold in the polarization experi- 

 ments, so that one would be justified in expecting that the 

 observed polarizations would be some 10 per cent, lower than 

 those calculated on the basis of the theoretical value for the 

 factor. From this point of view, the agreement found means 

 really a discrepancy of about 10 per cent., an amount which 

 considerably exceeds the experimental error. I made an 

 estimate of the experimental error by repeating two of the 

 experiment* under somewhat altered conditions, and found 

 results consistent with the first to under -002 volt, which was 

 equivalent to an error of about '.) per cent. This method of 

 estimating the error, it ia true, does not necessarily include 



