346 Prof. A. Schuster and Mr. A. W. Crossley. [Jan. 21, 



Ag 2 O 2 ) then shoot out rapidly and form a bridge across the electro- 

 lyte. 



We have not observed similar effects when the silver plates were 

 perforated, and the platinum wires which passed under the silver 

 plate were everywhere in metallic contact with it. We used filter- 

 paper to cover the anodes, and followed generally Lord Rayleigh's 

 instructions regarding the conduct of the experiment. The platinum 

 basins were in some experiments first cleaned out with sand, but 

 often this was not done. They were then washed with (1) con- 

 centrated nitric acid, (2) strong caustic soda, (3) tap water, (4) 

 distilled water. They were dried roughly with a clean silk hand- 

 kerchief and heated over a Bunsen flame. After an hour's cooling 

 they were weighed. The deposits of silver were washed three or 

 four times with distilled water, and allowed to stand under water 

 for a night ; they were then again washed several times and dried in 

 an air-bath at first at 100 C. ; the temperature was finally raised to 

 160 for ten minutes. After an hour's cooling, the final weighings 

 were taken. 



In a large number of experiments it almost certainly happens that 

 some anomalous results are obtained, either through insufficient wash- 

 ing or through loss of small quantities of silver. We give, without 

 exception, the result of each experiment, and think that on the whole 

 they show a remarkable consistency in the indications of the silver 

 voltameters. The effects we investigate are the differences in the 

 deposit of less than one part in a thousand, and the possibility of 

 investigating these differences is a proof that the electrolysis of 

 silver nitrate can safely be trusted to that degree of accuracy. 



The only serious source of error against which we had to guard was 

 the prevention of leakage in the leads between the two voltameters. 



That our results can in no way be attributed to such leakage is 

 shown by the fact that the voltameters were used in the same position 

 with the bell-jar exhausted or full of air. When the jar was full of 

 air, the difference in the deposit disappeared, except for the small 

 anomaly due to the different sizes of the basins. 



When the jar is exhausted, it might be thought that a film of 

 moisture could condense outside the platinum bowl, owing to the 

 cooling due to evaporation in an atmosphere saturated with vapour. 

 If such a film were to a certain extent to short-circuit the bowl, a 

 smaller deposit would be formed in vacuo ; but our effect is an 

 increase, not a diminution, of the deposit. Our leads were all care- 

 fully insulated, and as the resistance of the voltameters was never 

 more than 1 ohm, there is no difficulty in avoiding leakage to the 

 extent required. 



For the sake of clearness, we do not give our results in the order 

 in which they were obtained, but the numbers attached to each 



