254 Mr GRAHAM on the Law of 



tract T^th of its bulk by standing over water, without agitation, 

 for forty-eight hours, and from the loss of the denser portion of 

 it. But in the case of this gas, the experiment should succeed 

 over brine, which absorbs much less of the gas than water does. 



The process of diffusion may be managed so as to demonstrate 

 relations in density. The short upper tubes of two diffusion- 

 bulbs, not closed by plaster, but open, were connected by means 

 of thick caoutchouc adopters, with the two ends of a short piece 

 of straight tube, in which there was a diaphragm of plaster, yth 

 of an inch in thickness, and equidistant from either end of the 

 tube. The apparatus being proved air-tight, and the plug in a 

 proper condition for diffusion, one of the diffusion-bulbs was 

 filled with nitrogen gas, and the other with carbonic oxide, and 

 the bulbs placed upright in separate contiguous glasses contain- 

 ing water. The quantity of gas in each was carefully observed 

 at the beginning of the experiment, and after the expiry of 

 twenty-four hours, when it was found to be identically the same 

 as at first ; at least, if a contraction or expansion took place, it was 

 the same in both bulbs, and therefore entirely due to changes in 

 temperature or pressure. Now, the gases were found by analysis 

 to be uniformly diffused through both bulbs; so that nitrogen and 

 carbonic oxide are of the same density, or at least do not differ 

 more than y^th part, which was the limit of observation in the 

 case of these experiments. It appears, also, that inequality of 

 density is not an essential requisite in diffusion. 



I had occasion to remark, more than once, a singular accident 

 to the stucco plugs. After being disused for some days or weeks, 

 and left in the interval exposed to the air, which might be either 

 dry or damp at the time, the plugs occasionally, on a new trial, 

 did not permit diffusion to take place through their pores, at 

 least immediately. Hydrogen, however, always opened a passage 

 in the course of two or three minutes, and then the diffusion 

 proceeded as rapidly as ever. Carbureted hydrogen, and the 

 other gases, often required a longer period. A slight heat re- 



