440 



these parts tog ether with some water-vapour leave the appara- 

 tus eacli through a very narrow 

 capillary. The water vapour is 

 removed by freezing it out. The 

 sepai'ated pai't is received in a 

 vessel, the rest of the gas mixture, 

 however, is again led back to 

 the apparatus by means of a 

 circulation pump. 



The vapour current is con- 

 troled by regulating the curi-ent 

 in the heating spiral wound on 

 W, and the temperatures in K^ 

 and K,. The latter is effected 

 in such a way that the water 

 flowing through the cooling 

 jackets with accurately con- 

 stant velocity is beforehand led 

 through a copper tube, sun-ound- 

 ed by a heating coil, so that 

 the tempeiature of the water 

 depends on the current passing 

 through this heating coil. The 

 check on the curient velocity 

 is made possible by the capil- 

 laries between U and 6\, and 

 between if and C\, these causing 

 a difference of pressure between 

 W and Ci resp. C^ that is in direct 

 ratio to the current velocity in // resp. J/. This difference of pressure 

 can be measured by the difference of level between the condensed 

 water in C, resp. C, and the water in W. Neither the absolute 

 value of the current velocity nor the temperature of the water 

 in /{"i and K^ need be known ; when the level of the water 

 in the two tubes with regard to the level in W is such, that 

 the unmixing of the gas mixture is satisfactory, the heating current 

 need only be regulated so, that this position is maintained. 



It is not necessary to keep the temperature, and with it the 

 density of the vapour, accurately constant, for both the current 

 velocity corresponding to a given difference of pressui-e between the 

 ends of the capillary tube, and the diffusion constants of the diffusing 

 gases are approximately inversely proportional to the density of the 



Fig. 5. 



