278 Messrs. Gee and Holden on the Change of 



the first direction, evolution of gas at B begins, and the 

 streaks simultaneously cease to be visible. The streaks are to 

 be seen streaming from one electrode only at a time, and 

 always descend from that palladium electrode which first 

 serves as kathode and then as anode. 



When platinum electrodes are substituted for the palladium 

 ones, we have not been successful in obtaining the streaks; it 

 is, however, only necessary that one of the electrodes should 

 be palladium in order to obtain the streaks. 



An experiment was made in order to ascertain the part 

 which the reversal of the current plays in the formation of 

 the streaks. The current was sent in one direction through 

 the electrolyte, the palladium electrodes then taken out, 

 washed, and their surface cleaned with glass-paper. On now 

 reversing the current and replacing the electrodes, the streaks 

 are quite as evident as if the electrodes had not been taken out. 

 But if, on taking the electrodes out, they are heated to redness 

 instead of merely having their surface cleaned, on replacing 

 them and reversing the current the streaks are not seen, and 

 gas appears immediately from the anode. We concluded 

 from this experiment that the function of the current in the 

 first direction was to fill with hydrogen that electrode at 

 which the streaks appear in the second direction of the cur- 

 rent ; and that the reason why, after a time, the streaks ceased 

 to be visible, and simultaneously gas began to be evolved at 

 the anode, was that by this time all the occluded hydrogen 

 had been used up. It was therefore thought probable that 

 the streaks were formed by a combination of the occluded 

 hydrogen with the ion (either S0 4 or some of its components, 

 such as oxygen) which is liberated at the anode in the second 

 direction of the current. The theory that the streaks were 

 composed of concentrated sulphuric acid seemed open to great 

 objections ; for we know that concentration takes place at the 

 anode in dilute sulphuric acid without a reversal of the cur- 

 rent, and also does not depend on the use of palladium elec- 

 trodes ; but the streaks are, as is shown above, only to be 

 obtained after a reversal of the current, and are not to be 

 obtained with platinum electrodes. It was therefore decided 

 to test, as far as practicable, other possible combinations, such 

 as hydroxyl (the streaks cannot be composed of water, for 

 they are denser than the surrounding electrolyte), a salt of 

 palladium, &c. 



Testing for Hydroxyl. 

 The palladium electrodes used were rectangular strips, about 

 2 centim. long and *3 centim. broad, placed vertically in 10- 



