the Diffusion of Gases. 225 



increased by a portion which entered mechanically ; and varied 

 from a third to a fourth part of the hydrogen, which escaped at 

 the same time. The results, therefore, oscillate, as they should 

 do, about our theoretical number. One volume air should re- 

 place 3.7947 volumes hydrogen; or the whole hydrogen, on 

 escaping from the jar, should be replaced by little more than 

 one-fourth of its bulk of air, and a very great contraction ensue. 



But it is unnecessary to detail experiments made with the 

 jar with the fissure, as with every precaution they were not pre- 

 cise, although at all times compatible with, and indeed illustra- 

 tive of, the law. Thus a sensible contraction always took place 

 in the bulk of the gaseous contents of the jar when rilled with 

 carbureted hydrogen of marshes, or with coal-gas, which, like 

 hydrogen, are lighter than air, and ought therefore to be re- 

 placed by less than equal volumes of air. With defiant gas and 

 carbonic oxide, which approach closely to the density of air, no 

 contraction was perceptible, not attributable to other causes, al- 

 though the gases as usual wholly escaped. In the case of car- 

 bonic acid, which is heavier than air, a slight, but positive, ex- 

 pansion appeared to take place, the experiment being performed 

 over mercury. 



But the same fissure or opening never allows the process of 

 diffusion to go on with the same degree of rapidity in two suc- 

 cessive experiments, principally, I believe, from its size changing 

 with variations in its condition in regard to humidity. The fis- 

 sures appear to be extremely minute, for we cannot cause either 

 air or the gas employed to flow through them mechanically, at 

 the same rate as it passes by the agency of diffusion, without the 

 application of considerable pressure. Artificial chinks, such as 

 that obtained by pressing together ground glass-plates, or in 

 phials fitted with accurately ground glass- stoppers, allow gas to 

 pass through under the slightest pressure, and do not answer for 

 the experiment. 



