JUNE 6, 1931] NATURE 855

Fine Structure in the Arc Spectra of In a list of wave-length measurements of the and . vacuwn spark of in the Schwnann region, for which I am indebted to Dr. R. J. Lang, there are Bromine.-The fine structures of a nwnber of very strong lines at 75923 (:\1317·12 A.) and 114601 bromine arc lines have been previously reported by (X872·59 A.). If these are asswned to be the above• Hori.l By asswning that the lines X:\6632, 6560, 6351, 3 1 mentioned combinations, then 686p P 1 - 686p P 1 6149 have identical structures, De Bruin 2 inferred would be 38678. A distinctive feature of the spectra that the nuclear is%. The fine structure measure• Hg I, Tl II, and Pb III is the appearance with great ments have been considerably extended, using a high intensity of lines arising from intercombinations frequency ( 15 megacycles) electrodeless discharge in between singlet and triplet terms. One would there• pure bromine vapour and a Fabry-Perot interfero• fore expect to find the wave nwnber difference meter. All the observed structures arise from the 38678 recurring a nwnber of times in the Schwnann 4p45s electron configuration, as was to be expected. region. At least five pairs of lines with this difference Although the lines employed by De Bruin have not have been found in approximately the expected identical structures, as he supposed, the value of positions. However, in spite of this apparent cor• i = has been confirmed. There is evidence that the roboration, I have been somewhat doubtful of the two of bromine (79, 81) have the same nuclear validity of the foregoing identification, because the spin. line 75923 had already been included by Lang' Iodine.-Fine structures have been previously re• in a scheme for Bi III. In a recent paper 5 by corded only in the iodine spark lines,3 but by employ• McLennan, McLay, and Crawford on the spark ing similar experimental arrangements to those used spectra of bismuth, the line 7 5923 finds no place in for bromine, fine structures have been observed in the the scheme for Bi II or for Bi III. region A4700-X8000. The arc lines are mostly regular Still more significant evidence that this line belongs quartets and sextets degrading to the violet in both 1 to Bi IV, and is in fact the resonance line 6s6s 8 0 - intensity and interval. The simple regularity of the 3 6s6p P 1, is to be found in the data given by Arvids• structures is such that they are obviously characteristic son • in a recent letter to NATURE, in which it is of only one j term in each line. The observed struc• reported that the line 75923 has been resolved into tures arise from the 5p46s electron configuration, that three components of relative intensities 6 : 5 : 5. If is, that corresponding to the 4p45s in bromine. This it is asswned that the nuclear quantwn nwnber is in agreement with the partial analysis of the 1 of bismuth is 9/2 and that the 6868 S 0 term is single, spectrum made here by S. F. Evans (unpublished). As then the theoretical relative intensities of the three 2i + 1 is the maximwn multiplicity attainable, the 3 components of the 6s68 IS0 - 686p P 1 combination existence of sextet terms proves that i is at least as calculated by the formulre given by Pauling and equal to t. which had been previously inferred from Goudsmit 7 is 6 : 5 : 4 in order of decreasing wave the absence of appreciable alternating intensities in number. This agrees very closely with the experi• the absorption band lines of I 2• mental ratio. It therefore seems very probable that Since with j less than i the full multiplicity is not this is the resonance line of Bi IV, giving the value attained, the application of the interval rule must 9·36 volts for the resonance potential. decide the value of i (in the absence of Zeeman effect Some progress has been made in finding further measurements). The best line observed is :\4862, wave nwnber regularities in Bi IV, but the spectrwn which is a quartet and thus involves j = t in the is difficult to interpret because of the j j coupling 5p46s term (this is confirmed by the analysis). The of the two electrons. The work is still proceeding intervals are 123, 103, 83 (cm.-1 x 10·3 ), that is, and it is hoped that a detailed report will be made 6 x 20·5, 5 x 20·6, 4 x 20·7. A value of i = t must be elsewhere at a later date. invoked to fit these exact ratios (6 : 5 : 4), and this STANLEY SMITH. value is supported by the interval ratios in other lines. University of Alberta, The most probable values of i for the are Edmonton, Canada, shown in the table below : April10. 1 Fowler, A.," Series in Line Spectra", p, 148 ; Fleetway Press (1g22). I . . . Bromine. Iodine. ' McLennan, J. C., McLay, A. B., Crawford, M. F., Trans. Roy. Soc., Canada, 22, p. 241; 1928. Smith, S., Proc. Nat . .Acad. Sci., 14, p. 951; Number of Protons 19 35 (37, 39) 79, 81 127 1928. ' Rao, K. R., Narayan, A. L., Rao, A. S., Indian Jour. Phys., 2, p, Nuclear Spin . t ' ' 467; 1928. Smith, S., Proc. Nat. .Acad. Sci., 14, p, 878; 1928. " ' ' Lang, R. J., Phys. Re·o., 32, p. 737; 1928. • McLennan, J. C., McLay, A. B., Crawford, M. F., Proc. Roy. Soc., No apparent regularity exists, but the large and small A, 129, p. 579 ; 1930. i values are associated with a single . Full 1 Arvidsson, G., NATURE, 126, p, 566; 1930. 7 Pauling, L., and Goudsmit, S.," Structure of Line Spectra", p. 140 details of the fine structure measurements will be and p. 2U; McGraw Hill (1930). published elsewhere. S. TOLANSKY. Physics Department, Armstrong College Impact Figures on Polished Rock Salt Surfaces. (Newcastle-on-Tyne ), Durham University, May 5. IF a small steel ball is dropped from a height of a few inches on a polished rock salt surface, the imprint 1 Hori, Mem. Coll. Sci. Kyoto, vol. 9, p. 307; 1926. ' De Bruin, NATURE, Mar. 15., 1930, vol. 125, p. 414. of the ball on the crystal surface remains as a circular • Wood and Kimura, .Astroph. Jour., vol. 46, p. 181; 1917. depression of one or two millimetres in diameter. The surface is deformed, however, over a region many times the area of the circular depression. If an The Resonance Potential of Trebly Ionised Bismuth. optical test plane (a piece of ordinary plate-glass will UsiNG the known data of the spectra Hg I,l Tl II,' do) is placed on the crystal and the surface examined and Pb III,• and extrapolating by means of the in monochromatic light, the interference pattern irregular doublet law, the predicted values of the shows a nwnber of families of ' loops ' extending wave nwnbers of the important combinations away from the imprint of the ball. - 1 6s6s IS0 - 686p •P 1 and 6868 IS0 686p P 1 of Bi IV If the crystal surface approximates a 1, 0, 0 plane, are approximately 76,000 and 115,000. the depression will be surrounded by eight sets of No. 3214, VoL. 127]

© 1931 Nature Publishing