Series System in the Spectrum of Gold. 3 



The inverse D set referred to are in W. N. : — 



D ]a -19738-92 3815*52 -15923-38 (0/5) P 22 



62541 



Du -19113-51 (0/(H). 



The Zeeman effect (Z = 5/10) for the D 12 was not observed, 

 but the others definitely belong to a D (or possibly F) type. 

 "This and the corresponding sets in On, Ag are quite out of 

 step with the normal D lines. The satellite separation 62o, 

 however, reproduces itself in a very striking manner in the 

 spark linkages to which attention has been drawn in [IV. 

 p. 385]. 



In [ill. p. 405] an indirect attack on the problem was 

 made by using the value of the oun as calculated direct from 

 the atomic weight, S = 361'8(W/l00) 2 = ll06-9 and taking 

 the lines 17125, 20858 as D n (2), D 22 (2), as they were in step 

 with the undoubted corresponding D lines in Cu and Ag. 

 Their separation is 3733'43, producing therefore a satellite 

 separation of 82*11, with a weaker D 12 line too faint to have 

 been observed. The observed Zeeman pattern for 20858 is 

 0/6, which is that of a triplet D 22 line, in place of 0/5 required 

 for a doublet D 22 , whilst 17125 is undoubtedly a S 2 line, 

 although the pattern 4/8 as observed is not perfectly clear. 

 Since then 1 have found that Bydberg in his original memoir 

 had attached the same two lines also to D u and D l2 . It 

 would appear that the D n (2) and S 2 (2) are almost coincident 

 with 17125, as the supposition appears to be justified by the 

 results. 



The method adopted was to apply the laws (1) that the 

 separation 3815 was due to a displacement by a multiple of 

 the oun on the S or D limit, and (2) that the satellite 

 separation of 82 was due to a displacement by an oun multiple 

 in the D sequent. With this went the consideration that 

 the limit would fall into step with those for Cu and Ag, and 

 would therefore be of the order 30000 ±1000. The test (1) 

 applied to these gave a series of possible limits between these 

 values with A varying from 76 8 to 8 18. To each of these 

 test (2) was applied, with the result that only two of them 

 were found to satisfy the condition, viz. : — 



30819 A = 76 8 sat. sen. due to 24S 1 = 6g. 

 29465 A = 81S „ „ 28S 1 = 7S. 



The fact that 29465 is in due step with Cu = 31515, and 



B 2 



