October 13, 1921] 



NATURE 



2C9 



remain in substance unaltered. At the same time 

 the elaboration of the theoretical considerations 

 sketched in this letter throws a good deal more light 

 on the interpretation of man}' details. For instance, 

 it is possible to account for the appearance in the 

 atom, with increasing atomic number, of groups of 

 new constitution in such a way that we obtain a 

 natural interpretation not only, as before, of the 

 existence of such families of elements in the periodic 

 table as those of the iron metals and the rare earths, 

 but also of the almost complete absence of any effect 

 on the Rontgen-ray spectra of the appearance of such 

 groups. This absence is explained by the fact that in 

 these families we do not witness any sudden change 

 with increasing atomic number in the total number 

 of quanta of the orbits of the electrons of certain 

 groups. On the contrary, we may be assumed to 

 witness in the appearance of each of these families 

 the completion of a group by the inclusion of further 

 electrons moving in orbits characterised by the same 

 number of quanta. This addition to the group is 

 brought about by a change in the interaction between 

 the various possible types of orbits with this number 

 of quanta caused by the alteration in the dimensions 

 of the orbital loops and in the "apparent" number 

 of quanta which may be said to characterise these 

 loops. 



I have confined myself here to these points of 

 general character. For details of the theorv and its 

 applications I must refer the reader to a paper in 

 preparation for publication by the Royal Danish 

 Academy of Science. N. Bohr. 



Copenhagen, September i6. 



The Separation of Chlorine Into Isotopes. 



In Xatlre of April 22, 1920 (vol. 105, p. 230) Harkins 

 and Broeker reported that they had obtained a partial 

 separation of the element chlorine into isotopes. Five 

 determinations made early in Februarv of that year 

 showed atomic weight increases of 0052, 0059, 0057, 

 0055, and 0053 unit of atomic weight. Assuming 

 that ordinary chlorine has an atomic w^eight equal to 

 35460, the new chlorine as separated had an atomic 

 weight equal to 35-515. 



Shortly after the publication of this report in 

 Nature the progress of the separation was stopped 

 by the death of Mr. Broeker. Soon after this an 

 entirely independent diffusion was begun bv Dr. 

 Anson Hayes and the present writer, using new 

 apparatus and a different source for the hydrogen 

 chloride. In tliis way about twenty grams of chlorine 

 which has an atomic weight higher by 00375 ""i^ 

 than that of ordinary chlorine, and eighty grams with 

 an atomic weight higher by 0034 unit, have been ob- 

 tained, together with several kilograms the atomic 

 weight of which has not been determined, but must 

 te more than 02 unit high in order to give the results 

 cited above. Thus far the effort has been to collect 

 considerable material for future work rather than to 

 get the maximum increase of atomic weight. The 

 details of this work were reported to one of the 

 American chemical journals in April of this year, but 

 publication may be delayed many months by a strike 

 of the printers. 



The atomic-weight method used gave results 

 accurate to 0003 unit of atomic weight, and con- 

 sisted in determining the amounts of acid in samples 

 of ordinarv and isotopic hydrochloric acid of the same 

 density. The pyknometer*was made in such a way 

 as to give high precision, and the thermostat was 

 regulated to 0001°. Eighteen atomic-weight deter- 

 minations were made. 



.Absence of impurities, and especially of bromine 

 NO. 271 1, VOL. 108] 



and iodine, was ensured by using only the purest 

 materials in the diffusion in an apparatus consisting 

 of glass and porous porcelain only, and by re- 

 crystallising the chlorine from water in the form of 

 sodium chloride three times and by precipitating it 

 once in the same form by passing in isotopic 

 hydrogen chloride gas. In addition to this, two frac- 

 tional distillations with potassium permanganate and 

 one fractional distillation without this substance were 

 used. 



Mr. T. H. Liggett, who worked with the present 

 writer on this problem in the year 191 7-18, has re- 

 ported that he also has secured an increase of 005 

 in the atomic weight of chlorine by using the same 

 method — the diffusion of hydrogen chloride gas. 



Thus we have definitely confirmed the separation 

 reported by Harkins and Broeker. 



WiLUAM D. Harkins. 



University of Chicago, August 28. 



The Pickering Series in O Type Stars. 



If the Pickering series, consisting of the lines 541 1, 

 4542, 4200, etc., in stellar spectra, is due to ionised 

 helium, it is known from the investigations of Fowler 

 and from Bohr's theory that there must be additional 

 lines which appear as violet components of the Balmer 

 series of hydrogen. As these components have an 

 average separation of 2 A. they should be readily seen 

 in the spectra of O type stars. Unfortunately, there 

 are two unfavourable circumstances : first, the O type 

 stars are all faint, and, secondly, the lines of both 

 the Balmer and the Pickering series are in general 

 so diffuse as to be hopelessly blended. 



A preliminar}' sur\ey was made here last vear of 

 O type stars for the purpose of selecting one or more 

 stars of reasonable brightness and with fairlv sharp 

 lines. As a result 10 Lacertae, type Oe 5, magnitude 

 49 was selected as the most suitable star for tests 

 with higher dispersion. On resuming work this vear 

 the director. Dr. J. S. Plaskett. directed attention to 

 9 Sagittae, type Oc, magnitude 6-2, as also a verv suit- 

 able star. After some preliminary- experiments, two 

 plates of 9 Sagittae and 10 Lacertae with three-prism 

 dispersion (10 A. to i mm. at H7) were secured on 

 August 12 which showed components at Hj8 and H7 at 

 the theoretical separation. With this as a start, check 

 plates were secured, and in addition high dispersion 

 plates, requiring 9-10 hours' exposure were secured 

 of the region around Ho in 10 Lacertae. The pre- 

 liminary- wave-lengths on the international scale of 

 the complete Pickering series, and also of the Balmer 

 series, are summarised in tke following table : — 



Pickering series. 



Balmer series. 



1 The wave-leng!h of Ha was assumed in order to determine the position 

 of the other compjnent. Measures of the Ha component were difficult to 

 make, partly on account of the small linear dispersion and partly because 

 of the existence of a line 6558 6, origin unknown. 



* Blend with Fowler's enhanced nitrogen line 4S!,8"S2 in 10 Lacertjt. In 

 9 Sagitts there is no component, as^ it is probably obliterated lay the 

 enhanced nitrogen line appearing a« emission. 



3 Ulend with Fowler'f enhanced nitrugen line i.200'06. 



