508 [March 26, 



pound; sometimes it is only the mean density of the products of 

 decomposition. Some of the best-known substances, such as hydrated 

 sulphurous acid, ammoniacal salts, and pentachloride of phosphorus, 

 suffer decomposition when they are vaporized, and thus have an 

 apparent vapour- density, which is in reality nothing more than the 

 mean density of the products of their decomposition. 



"We recognize such cases in which the apparent is not the real 

 vapour-density by making a diffusion-analysis of the vapours. 

 This method of solving questions of the kind was proposed by one 

 of us two years ago*. In carrying it out practically, it was resolved 

 from the first not to diffuse through a porous diaphragm, but to 

 recur to Graham's original method, namely, to let our vapours dif- 

 fuse through a simple aperture or through a short tube. 



Independently of the experimental difficulties attending the use 

 of a porous diaphragm at high temperatures, there is a fatal objec- 

 tion to it, founded upon the inconclusiveness of the results obtained 

 in such a way. 



Our resolve to avoid porous substances was not by any means 

 shaken by Pebal's memoirf on the diffusion of chloride-of-ammo- 

 nium vapour through asbestos ; for what is more likely than that a 

 finely-divided silicate (a salt of an acid of indefinite capacity of 

 saturation) should decompose ammoniacal salts at elevated tem- 

 peratures ? 



The apparatus used in our experiments is of the simplest kind. 

 It is represented in the drawing, and consists of two glass flasks, the 

 necks of which do not fit air-tight : the narrow tube proceeding from 

 the upper one is fused to the flask. The lower flask is for the re- 

 ception of the vapour to be operated upon ; the upper flask is for 

 the atmosphere into which the vapour is to be diffused. The atmo- 



* Playfair and Wanklyn on Vapour-densities, Transactions of Boy. Soc. of 

 Edinburgh, 1861, vol. xxii. part 3. p. 458. In this paper it was proposed to 

 extend to vapours what had previously been applied to gases. One of the 

 earliest, perhaps the earliest example of a precise diffusion- analysis of a gas was 

 given by Frankland in his research upon the isolation of ethyl (see Quart. 

 Journ. Chem. Soc. vol. ii. p. 285, 1850). After describing his diffusion-apparatus 

 and its use in the case of ethyl, Frankland proceeds, " This method might in 

 almost every case be employed with advantage to determine whether or not any 

 specimen of gas be simple or mixed." 



t Ann. de Chim. et de Phys. January 18G3. 



