406 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



without taking any readings, the potential was immediately lowered and 

 the series continued. The whole experiment was at 21.9°C. 



It has been shown by Jahn and Shoenrock,* by Haber f and others, 



both theoretically and experimentally, that in many cases the potential is 



a linear function of the logarithm of the polarizing current. In Figure 



2, therefore, we have plotted from the values given in the table the 



potential E in one direction, in the other the logarithms of the current 



I. The various values obtained from the table are represented by the 



points of the arrows whose direction upward or downward indicates 



whether the values were taken on the way up to the high potential or on 



the way back to the low. We see that the points obtained in the first 



part of the experiment fall very exactly upon a straight line whose 



equation is 



E=A log 1+ B, (1) 



where A and B are constants. The points on the way back from the 

 high potential lie in every case above this line. 



The latter behavior is entirely unexpected. Many cases have been 

 studied in which the current is diminished by previous high polarization 

 of the electrode, and this is obviously due to the slow disappearance of 

 the polarization. But we see that in our case the current at a low 

 potential is increased by the previous passage of a larger current at a 

 higher potential. If this phenomenon had been observed in a single in- 

 stance only, we should have attributed it to chance errors, but it occurred 

 regularly in all our experiments. It can only be explained by the 



* Zeit. phys. Chem., 16, 45 (1895). 



t Ibid., 32, 193 (1900). 



