286 Mr. W. C. Dampier Whetham. [Nov. 24, 



When the solutions are of different specific resistance, there will 

 be a discontinuity of potential gradient at the boundary and a conse- 

 quent electrification. This can be got rid of entirely by using solu- 

 tions of the same specific resistance, and in all cases the effect on the 

 velocity of the boundary is non-reversible, and can be approximately 

 eliminated by reversing the current and taking the mean value of 

 the velocity. If the velocity is found to be the same in opposite 

 directions when the current is reversed, the effect must be neg- 

 ligible. 



The first solutions used were those of copper and ammonium 

 chlorides dissolved in aqueous ammonia. The copper solution is a 

 deep blue ; the other colourless. Their strength was O18 gram- 

 equivalent per litre. The mean velocity when the current was 

 passing upward came out 0"0406 cm. per minute, and when passing 

 downwards 0*0441 cm. per minute. In each case the junction moved 

 with the current. The potential gradient was 2 '73 volts per cm., 

 which gives as the specific ionic velocity 



G'00026 cm. a second. 

 Kohlrausch gives for infinite dilution 



0-00031 cm. a second. 



Solutions of potassium permanganate and potassium chloride were 

 taken to show the motion of the acid radicles, and, as it should, the 

 junction moved against the current. If we assume that the disappearance 

 of the red colour can only occur where permanganate is replaced by 

 chloride, the motion of the junction can be taken as an indication of the 

 velocity of chlorine. The result for solutions of 0*046 gram-equiva- 

 lent per litre was 0'00057 cm. per sec., and for solutions of about 

 one-tenth this strength 0*00059 cm. per sec. Kohlrausch gives 

 '00053 cm. per sec. as the specific ionic velocity of chlorine. 



The success of these preliminary experiments led me to attempt to 

 improve the method. I investigated one pair of salts with exactly the 

 same specific resistance for the same strength, though in cases where 

 this condition is nearly fulfilled (such as that of copper and 

 ammonium chlorides), the error thus introduced can be shown to be 

 negligible. 



The direct estimation of the potential gradient is unsatisfactory, 

 but if we measure the specific resistance of the solution (r), the area 

 of the junction tube (A), and the current (7), we can calculate the 

 specific ionic velocity (v^) from the observed velocity of the boundary 

 (v), for it is easy to show that 



vA. 



