Value of the Silver Oun. 307 



Satellite separation for 23S 1 = 72*07. 



Change per oun displ.= 3' 13. 



This gives the set 



1645-62- -28 £ S06M60+K 59643-58 + 2-28 S 



1717-69 30644'60+Z 59571=51+2-28 \ 



2566-05 31565-30+Z 60564-01 +2-28? 



1645, 1717 are beyond Paschen's furthest ultra red. He 

 gives lines at 2502*37, 2506*25. These are numerically the 

 combinations d% (3) — d 2 (4:), d x (3) — <ii(4), bat we should 

 scarcely expect to find combinations of such high order to be 

 strong enough to register themselves in this region. Their 

 separation is that of one oun displacement on dQ?). More- 

 over, 2566*05 — 2506*25 = 59*80, whilst a displacement of 

 19 o\ produces 59*47. The observed may therefore be a dis- 

 placed D line. The corresponding D line would be 60623*08. 

 This is outside the observed ultra-violet region. We cannot 

 test for the existence of the D(l) lines by sounding as in the 

 case of Au, since the links are not sufficiently long to reach. 

 The test by searching for d(l\—d(m) combination lines 

 requires d 2 (l) -d 2 (2) = 16625, d 2 (l)-d Q ('d) =22085. The 

 former is outside K. R.'s measures, and only just within 

 Eder and Valenta's. The latter give a line at 16559*20 

 which as d l (l) — d 1 (2) would make D n (l) = 1731*86 or a 

 satellite separation of 86*24 equivalent to a displacement of 

 28o\=7 6\ the same as in Au. The combination 22085 

 comes in a gap in the spectrum with band spectra, and the 

 line may possibly have escaped observation. Thus the means 

 for applying an observational test for the existence of the 

 D (1) lines are not at disposal. On the contrarv we can 

 apply the test for the D. The D n is found at (2) 59569*91 (3). 

 The D 22 is outside the observed region bv Handke, but by 

 sounding with the e link we find (4) 56795*59 . e = 60566*59 (3) . 

 These are entered in the table. The weaker line J) h2 is not 

 observed, but (1)59687*24(14 o\) gives the value in the table 

 which is practically exactly the calculated value. There is 

 then good evidence for the existence of the D set based on 

 34 A, and this is o-ood also for the validity of the D. The 

 apparent exception in Ag from the multiple A law is thus 

 explained in the same way as in Au. 



The evidence from the whole of the D material points dis- 

 tinctly to a negative value of f. The only numerical data 



