1917] on The Complexity of the Chemical Elements 133 



the mixture as well as the separate atomic weights are both unknown. 

 New ideals emerge and old ones are resuscitated bv this develop- 

 ment. There may be after all a very simple numerical relation 

 between the true atomic weights. The view that seems most pro- 

 bably true at present is that while hydrogen and helium may be the 

 ultimate constituents of matter in the Proutian sense, and the atomic 

 weights therefore approximate multiples of that of hydrogen, small 

 deviations, such as exist between the atomic weights of these two 

 constituent elements themselves, may be due to the manner in which 

 the atom is constituted, in accordance with the principle of mutual 

 electro-magnetic mass, developed by Silberstein and others. The 

 electro -magnetic mass of two charges in juxtaposition would not be 

 the exact sum of the masses when the claarges are separated. The 

 atomic w'eight of hydrogen is 1'0(»78 in terms of that of helium 

 as 3*1)9, and that the latter is not exactly four times the former may 

 be the expression of this effect. Harkins and Wilson have recently 

 gone into the question with some thoroughness, and the conclusion 

 of most interest in the present connection, which appears to emerge, 

 is in favour of regarding most of the effect to occur in the formation 

 of helium from hydrogen, and very little in subsequent aggregations 

 of the helium. In the region of the radio-elements, where we have 

 abundant examples of the expulsion of helium atoms as a-particles, 

 it seems as if we could almost safely neglect this effect altogether. 

 Thus radium has the atomic weight almost exactly 226, and the 

 ultimate product almost exactly 2o6, showing that in 5 a- and 

 4 /5-ray changes the mean effect is nil, and the atomic weights are 

 moreover integers in terms of oxygen as 16, or helium 4. It is true 

 that the atomic weights of both thorium and uranium are between 

 U • 1 and • 2 greater than exact integers, but it is difficult to be sure 

 that this difference is real. 



When, among the light elements, we come across a clear case of 

 large departure from the integral value, such as magnesium 24*32 

 and chlorine 35 "46, we may reasonably suspect the elements to be a 

 mixture of isotopes. If this is true for chlorine, it suggests a most 

 undesirable feature in the modern practice of determining atomic 

 weights. More and more the one method has come to be relied 

 upon : the preparation of the chloride of the element and the com- 

 parison of its w^eight with that of the silver necessary to combine 

 with the chlorine, and with the weight of the silver chloride formed. 



Almost the only practical method, and that a very laborious and 

 imperfect one, which may be expected to resolve a mixture of 

 isotopes, is by long-continued fractional gaseous diffusion, which is 

 likely to be the more effective the lower the atomic weight. Assume, 

 for example, chlorine were a mixture of isotopes of separat'^. atomic 

 weights 34 and 36, or 35 and 36. The 34 isotope would diffuse 

 some 3 per cent faster than the 36, and the 35 some 1*5 per cent 

 faster. 



