43: 



the same remarks are valid here as in the first case; thus it is 

 also practical here to choose the current velocity great and geo- 

 metrical dimensions small to render the quantity attained as great as 

 possible. 



All tiiese considerations have completely been confirmed by expe- 

 riment. In order to effect the separation of gas mixtures by diffusion 

 in a flowing gas in practice, it is first of all required that as a medium 

 a gas be chosen that can be easily separated from the diffusing gases. 

 This can be attained in a simple way by using a vapour as medium 

 gas, which can be condensed after having passed the place where 

 the diffusion is brought about. All the experiments made so far, 

 were carried out with water vapour of 15 to 60 cm. pressure. 

 The use of mercury vapour of lower pressure may, possibly, be still 

 more efficient; this will be further investigated. 



The chief point in the construction of apparatus for 

 carrying out the process described above, is the produc- 

 tion of a constant vapour current. When a gas passes 

 over a sufTicient distance through a cylindrical tube, a 

 current is obtained with parallel stream lines, but the 

 velocity is not constant; it decreases from the axis 

 towards the walls of the tube, as is represented in 

 Kig. 1. fig. 1. it is, however, possible to get a current of constant 

 velocity, though over a short distance only, when the gas passes 

 through a wide tube with a suddenly decreasing diameter or when 

 the gas escapes from a vessel through a small hole in the wall. 

 When in this waj' the medium gas flows 

 from a vessel A into a vessel B (fig. 2), 

 and when the gas mixture that is to be 

 separated, is admitted to the vessel B, the ^m^mf^;:- 

 case of diffusion against the gas current is il ,/?^ 



realized. The velocity of the current can 

 then always be chosen such that only the Fig. 2. 



component of the gas mixture that diffuses more rapidly, diffuses 

 against the current and reaches the vessel A, from which it can 

 be pumped off together with part of the medium gas. 



This idea was carried out experimentally as follows: the water 

 vapour generated in a vessel heated electrically, flows through «S(tig. 3) 

 into a tube closed at the bottom by a metal plate Z) of a thickness 

 of 1 m.m. This circular plate of a diameter of 28 m.m. has 30 

 holes of 1 m.m., each, distributed uniformly over its surface. 

 Through these holes the water vapour enters the vessel F, the 

 lower part of which is surrounded by a cooling jacket, so that 



