GAS BATTERY. 259 



be used as the electrolyte, without damaging the apparatus or 

 introducing foreign matter. 



In the experiments I am about to describe the results were 

 generally tested by chemical action, as manifested by the 

 electrolysis, either of iodide of potassium or of water. I had 

 at my disposal a highly sensitive astatic galvanometer ; but I 

 found such slight local actions disturb it, that a range of test 

 experiments was in each case necessary to eliminate the true 

 battery action from the accidental currents ; and, with all the 

 pains that could be bestowed upon it, the results were less 

 definite and trustworthy than those obtained with the iodide. 



I may here state also that, although with the battery 

 described in my original paper, when charged with oxygen, 

 hydrogen, and dilute sulphuric acid, I could not succeed in 

 perceptibly decomposing water with less than twenty-six cells, 

 yet the new arrangements, from their superiority in size and 

 construction, were capable, when charged with the same 

 gases and electrolyte, of decomposing water with four cells ; 

 and a single cell would decompose iodide of potassium. 



Experiment i. Ten cells charged to a given mark on the 

 tube with dilute sulphuric acid, sp. gr. 1*2, oxygen and hydro- 

 gen, were arranged in circuit with an interposed voltameter,* 

 as in figs. 6 and 7, and allowed to remain so for thirty-six hours. 

 At the end of that time 2'i cubic inches of mixed gas were 

 evolved in the voltameter; the liquid had risen in each of the 

 hydrogen tubes of the battery to the extent of 1*5 cubic inch, 

 and in the oxygen tubes 07 cubic inch, equalling altogether 

 2*2 cubic inches ; there was therefore O'l cubic inch more of 

 hydrogen absorbed in the battery tubes than was evolved in 

 the voltameter. 



This experiment was several times repeated, with the same 

 general results. I give some of them in the annexed table. 



* These experiments were made with the battery fig. i, though for more 

 cloorly showing the volumes of the gases the second form is represented in 

 figs. 6 and 7. The voltameter employed on this occasion had electrodes of fine 

 platinum wire a quarter of an inch long. From the nature of the gas battery it 

 is difficult to know the efficient surface of the plates. In ordinary batteries I have 

 found, and stated some time ago, that for quantitative effects the electrodes 

 should be of the same size as the battery plates. 



S 2 



