668 Polarisation at a Metallic Anode. 



the effect of convection currents and imperfect insulation 

 mentioned on p. 657; but the effect of these sources of error 

 would be to reduce rather than to increase the observed 

 polarization, and so could not account for the discrepancy. 

 For the same reason the discrepancy cannot be ascribed to 

 the neglect, which was made in the calculation of the polari- 

 zation, of the effect of the slope of potential in the anode tube 

 on the motion of the silver ions (see p. 648). An approximate 

 calculation of the effect of this shows that the error is practi- 

 cally inappreciable, and in any case it would tend to reduce 

 the observed polarization. 



On the other hand, it is possible that the value assumed 

 for the coefficient of diffusion of the silver ions in the presence 

 of the nitric acid is in excess of its true value. The presence 

 of nitric-acid ions undoubtedly reduces the equivalent con- 

 ductivity of silver nitrate below its value at infinite dilution : 

 this may be due to a diminution either in the dissociation, 

 or possibly in the ionic mobility of the silver ions. Either 

 of these effects would cause a decrease in the effective ionic 

 mobility which determines the rate at which the silver diffuses 

 away from the electrode. The assumption of too high a value 

 for the coefficient of diffusion RTU/e of the silver ions will 

 diminish the second constant 23'80 of equation (24), and will 

 so tend to counterbalance the error produced by the too high 

 value of the constant '0578. I think it is probable that 

 both constants are really in error, and that the agreement 

 found between the observed and calculated values of the 

 polarization is due to the two errors being of such amounts 

 as to approximately cancel each other. An examination of 

 the equation shows that this might occur without the shape 

 of the curve which expresses the variation of the polarization 

 with the time being appreciably affected. To produce agree- 

 ment between theory and experiment when the observed 

 value *052 is substituted for the theoretical *0578, a diminu- 

 tion of some 20 per cent, must be assumed to be produced in 

 the value of the coefficient of diffusion of the silver ions by 

 the nitric acid. Since the second constant occurs under the 

 sign of a logarithm in equation (24), the method of the 

 experiments forms a very inaccurate means of measuring the 

 coefficient of diffusion of the ions, so that only a very rough 

 estimate of the value required is possible. The amount of 

 the diminution suggested by these considerations seems 

 perhaps larger than one would expect on the ordinary theory 

 of dissociation. There are, however, no other experimental 

 determinations with which to compare it, so that there is no 

 reason to suppose it may not actually exist. 



