132 Professor Frederick Soddy [May 18, 



bringing before you, I belieye, some of his exceedingly interesting 

 work on pleochroic balos, from which he has grounds for the conchi- 

 sion that the accepted period of uranium may be too long. But 

 since I obtained, for the period of ionium, a minimum valae two- 

 thirds of that estimated by Meyer from the atomic weight, it is 

 difficult to belieye that the accepted period of uranium can haye been 

 oyerestimated by more than 50 per cent of the real period. The 

 matter could be pushed to a further conclusion if it were found 

 possible to estimate the percentage of thorium in the thorium-ionium 

 preparation, a piece of work that ought not to be beyond the resources 

 of radio-chemical analysis. This would then constitute a check on 

 the period of uranium as w^ell as on that of ionium. Such a direct 

 check would be of considerable importance in the determination of 

 geological ages. 



The period of ionium enables us to calculate the ratio, between 

 the weights of ionium and uranium in pitchblende, as 17 "4 to 10^, 

 and the doctrine of the non- separability of isotopes leads directly to 

 the ratio, between the thorium and uranium in the mineral, as 41*7 

 to 10^ This quantity of thorium is, unfortunately, too small for 

 direct estimation. Otherwise it would be possible to deyise a yery 

 strict test of the degree of non-separability. As it is, the work is 

 sufficiently convincing. Thirty tons of a mineral containing a 

 majority of the known elements in detectable amount, in the hands 

 of one whose researches in the most difficult field of chemical 

 separation are world-renowned, yield a preparation of the order of 

 one-millionth of the weight of the mineral, which cannot be distin- 

 guished from pure thorium in its chemical character. Anyone could 

 tell in the dark that it was not pure thorium, for its a-actiyity is 

 30,000 times greater than that of thorium. This is then submitted 

 to that particular series of purifications designed to give the purest 

 possible thorium for an atomic weight determination, and it emerges 

 without any separation of the ionium, luit with a spectrum identical 

 with that of a control specimen of thorium similarly purified. The 

 complete absence of impurities in the spectrum show that the chemical 

 work has been yery effectiyely done, and the atomic weight shows 

 that it must contain 3o per cent by weight of the isotope ionium, a 

 result which agrees with its a-activity and the now known period of 

 the latter. 



Determinatiox of Atomic AYeights. 



The results enumerated thus prove that the atomic weight can no 

 longer be regarded as a natural constant, or the chemically pure 

 element as a homogeneous type of matter. The latter may be, and 

 doubtless often is, a mixture of isotopes varying in atomic weight 

 over a small number of units, and the former then has no exact 

 physical significance, being a mean value in which the proportions of 



