Series System in the Spectrum of Gold. 17 



I omit reference to them, as for the present purpose — that of 

 illustration of method and the more exact determination of 

 the gold oun — the next set in which /(I) depends on 

 N/(8A) 2 is preferable and is sufficient. 



Case II. /(1)=N/(8A) 2 . Here f(l) = 30017*06 - *251 .? 

 with the oun displacement producing a change of 11*040. 

 With the d[l) value F(oo), viz., 26734*86 - '237 x, the 

 calculated values of the set for m = l are 



F x -3531*20 56502*92 F 1? 



F 2 -328220 56751*92- -488^? F 2 . 



For the F lines only (1)56744*03 + 3'22p is observed. It 

 should be normally a strong line. There is a strong line at 

 (5)56721*50 separated by 22*53, and as 2B 1 in the sequent 

 produces 22*09 it is probably ¥ 2 (1){2B 1 ). The F x should 

 be 56495 (not seen). But there is again a strong line 

 (5)56561*09, 66 ahead. This may be either 68 x (66*27) on 

 the sequent or the 7 S L (64*0) on the limit (similar to Case I.). 

 The only interest in these at present is that if displaced F 

 lines, the weakness of the normal ones is explained. 



For evidence of the F lines we must again have recourse 

 to sounding. The u links are too small to reach. The v links 

 land close to a doublet with the proper separation, viz. : 



(In) 18217*03 + *15p 250-53 (3) 18467*56 + *15p, 



with again several lines intensity (In) near 18217, suggesting 

 other displaced lines with consequent weakness. Subtracting 

 v gives as the F lines 3283, 3533*51. Treating the series as 

 N/(9A)-N(8 A) gives— with the observed F 2 — 



-3533*51 [56493*52] 



25051 25051 



-3283*00 26730-52 + 1*6;> 56744*03 + 3*22p. 



In the absence of observation in the ultra red this set 

 cannot be considered as conclusive, although it satisfies the 

 conditions. If it be taken as re;d, the observed 56744*03 is 

 the value calculated above and should give the value of ar*. 

 That is 



56744*03 + 3*22^ = 56751*92- -488^, 



whence x= 16*16 — 6*60/>. 

 This gives a closer value to A as depending on the D, F data. 

 If it is the same as from S(oo) and v, ,v<10 or p should be 

 about 1 — or the observed measure of the AY, L should be 

 "1 A less. The possible error is certainly greater than this. 

 Provisionally we will take p=l, a? = 9*5 + .r, where now ,v = 

 Phil. Mag. S. 6. Vol, 38. No. 223. July 1919. C 



