282 EXPERIMENTAL INVESTIGATIONS. 



first result did follow : upon taking out the stoppers the morn- 

 ing after the battery had been charged, the liquid rose in the 

 air-tube one-fifth of the gaseous volume. I now closed it 

 again and examined it three days afterwards ; a very curious 

 effect had taken place ; the volume of the gas in the air- tube, 

 which had previously contracted, had now increased, and it 

 continued slowly increasing day after day. I at first believed 

 that the nitrogen was decomposed, but after many conjectures 

 and experiments found that the increase was due to the addi- 

 tion of hydrogen. On repeating the experiment with nitrogen 

 instead of air the same effect took place, but of course with- 

 out the previous contraction. I now returned to battery fig. 

 4; several of these .cells charged, some with atmospheric air 

 and hydrogen, and others with nitrogen and hydrogen, did not 

 exhibit the effect, though suffered to remain six weeks, each 

 in closed circuit, 



To ascertain whether the vacuum formed by the abstrac- 

 tion of oxygen from the liquid had anything to do with the 

 above effect, a central narrow tube, open at both ends, was 

 substituted for the stopper in the battery fig. 8 ; the hydrogen 

 was still evolved. Not to detail a tedious set of test experi- 

 ments, I at length found that two points were essential to 

 obtaining the effect with certainty ; first, the exclusion of 

 any notable quantity of, atmospheric air from solution ; and 

 secondly, great purity in the hydrogen. In the former case, 

 when the hydrogen could find oxygen to combine with, it 

 was not evolved ; in the latter, there would be mixed or 

 rather diluted gas on both sides, and the forces would be ba- 

 lanced ; thus I have never succeeded in obtaining the effect 

 in the open battery, fig. 4 with hydrogen obtained in the 

 ordinary way from granulated zinc or iron filings, but have 

 sometimes succeeded with hydrogen procured by electro- 

 lysis. In the battery, fig. 8, I have succeeded in producing 

 the effect, but in a feeble degree, from hydrogen obtained 

 in the common way, but have never failed with hydrogen 

 obtained by electrolysis. Oxygen of the greatest purity, 

 voltaically associated with nitrogen, does not produce a 

 similar effect. The above unexpected results render it 

 necessary, in order to ensure accuracy in the eudiometric 



