328 BELL SYSTEM TECHNICAL JOURNAL 



cases now somewhat surpasses one part in ten thousand, which is 

 about as good as the chemical methods can offer. 



Many of the physical evaluations have been in beautiful agreement 

 with the best-approved of the chemical, reflecting honor on both; 

 but there have been striking temporary exceptions, with ultimate 

 results very surprising to anyone brought up in the tradition that 

 chemical atomic weights stand for the ne plus ultra in accuracy. The 

 weights of krypton and xenon were formerly given as 130.2 and 82.9; 

 Aston evaluated them as 83.77 ± 0.02 and 131.27 ± 0.04; within a 

 year (1931) redeterminations of the densities of these gases (perhaps it 

 would be justice to call this a physical rather than a chemical method) 

 resulted in 83.7 and 131.3. Among the elements for which the 

 analysis of isotopes has lately given a value markedly different from 

 the accepted chemical atomic weight, are osmium, selenium, scandium 

 and caesium. It will be interesting to see what happens to these 

 values in the tables of atomic weights. 



I left the story of the discovery of H^ to the end, so as to make 

 earlier mention of several things on which it depended. Apparently 

 it was the joint result of two independent predictions. First, the 

 ratio of the masses of the atoms H^ and O^^ agrees remarkably with 

 the ratio of the chemical atomic weights of hydrogen and oxygen: 

 both are certainly between 1.0077 and 1.0078. This agreement seems 

 wonderful testimony to the accuracy of the measurements of physicists 

 and chemists; but it turns out to be a mere coincidence. Such 

 testimony it indeed would be, if H^ and O'® were the sole isotopes of 

 their respective elements; but from the moment when O^^ and O^* 

 were discovered, it could be taken as meaning one thing only (short of 

 actual errors in the work): it could be taken only as meaning that 

 there is an extra isotope (or more than one) of hydrogen, more massive 

 than H^ This idea came first to Birge and Menzel, who proceeded to 

 compute in what ratio of abundances H^ and H^ must stand in order to 

 produce the agreement in question, if H^ be the only extra isotope. 

 The result must depend, of course, on the ratios of the abundances of 

 O^^ and O^* to that of O^^. For these ratios the estimates (made from 

 band-spectra, excepting for a preliminary one by Aston) are not in 

 very good accord. At the time of the prediction of Birge and Menzel, 

 they indicated a ratio of 4500 to 1 for the abundances of H^ and H^ in 

 ordinary hydrogen. Second, the diagram of Fig. 6 shows a recurring 

 uniformity, a stepwise pattern, in the broken line connecting the 

 successive dots from Z = 3 to Z = 8. If isotopes H^, H^ and He^ 

 exist, then this pattern extends uninterrupted down to Z = 1. 



These were the ideas which brought about the discovery of H^ by 



