Tlie Homogeneity of Helium. 



317 



repeated diffusion which possessed respectively diffusion rates corre- 

 sponding to the densities 2*133 and 1*874. It was there pointed out 

 that these densities are not correct (although their ratio is probably 

 not wrong), owing to the curious fact that the rate of diffusion of 

 helium is too rapid for its density, i.e., it does not follow Graham's 

 law of the inverse square root of the densities. These samples of 

 gas also differed in refractivity, and the difference was approxi- 

 mately proportional to the difference in density. 



Towards the end of the paper, the conjecture was hazarded that 

 it was not beyond the bounds of possibility that the systematic dif- 

 fusion of what we are accustomed to regard as a homogeneous gas, 

 for example, nitrogen, might conceivably sift light molecules from 

 heavy molecules. It is true that the fineness of the lines of the 

 spectrum would offer an argument in favour of the uniformity of 

 molecular weight ; but still it is never advisable to assume any 

 physical theory without submitting it to rigorous proof. And it was 

 thought possible that the fractional diffusion to which helium had 

 been subjected might have had the result of effecting such a separa- 

 tion ; a separation, not of chemical species, but of molecular magni- 

 tude. The other and more ordinary explanation of the splitting of 

 helium into fractions of different density is that helium must be 

 regarded as a mixture of two gases, one lighter than the other. 



Since the publication of the paper mentioned, Dr. A. Hagenbach 

 has confirmed the possibility of separating helium into portions of 

 two densities by diffusion ; and the differences in density were 

 practically the same as those observed in the laboratory of Univer- 

 sity College.* 



These experiments were made with somewhat over 200 c.c. of gas ; 

 but it was decided to make experiments of a similar kind, on a much 

 larger quantity of helium. 



An apparatus was therefore constructed, similar in principle to the 

 one previously employed, but on a much larger scale. The main 

 features are shown in the illustration on p. 208 of the paper 

 previously alluded to ; but on account of the large amount of gas 

 diffused, it was not practicable to collect it in tubes. Instead, there- 

 fore, of the bent tube EN" of the former apparatus, the tube con- 

 nected with the stopcock E was continued horizontally, and by 

 means of six vertical branches it communicated with six gas reser- 

 voirs, each furnished with a two-way stopcock. It was possible 

 with this means to cause gas from any one of the reservoirs to enter 

 the diffusion apparatus A. In order to be able to collect the gas in 

 any desired reservoir, the delivery tube of the Topler pump F 

 delivered gas into a jar somewhat similar to that shown at J, but 

 provided with a vertical branch, which was bent horizontally some 

 * < Wied. Ann.,' vol. 60, p. 124. 



