the Spectrum of Copper. 469 



be regarded as a displaced limit of 12^£ (producing 51*80) 

 we find a corresponding p at (In) 32739*77 (\= 3053-52). 

 The two give a mean limit of 28334*41. The shift 51-80 

 would then indicate F 1 (oo) = 28386'21. 



m=5. The line 25328*06 with f x (5) =3059*2 gives 

 F 1 (oo) = 28387*26 or with the Y x (co ) = 28382*96 of m=3 

 gives /i (5) = 3054*90. This has a mantissa 991768 and much 

 more in line with the others than 5059. 



m = 6. F x (6) should be close to S 2 (4) = 26135*66. That 

 in the table as F 2 (6) gives f a = 2246*43. 



m=7. The F 2 gives f 2 — 1719*16, with denominator 

 7*990 much more in step than the 1715 9 determined from 

 the very uncertain data of d'—f above. The f\ should 

 be practically equal to f 2 . In this case the F : (7) would 

 require F^oo ) =28385*10. 



m = 8. The F, gives / 3 = 1357*28 (mant. =*990). The 

 same value for /i in F : requires Fj(qo ) = 28382*70. 



m — $. With a mantissa 990 for f\, it is 1099*6 requiring 

 F^ =28383-0. 



7n — 10. The lines given in the table cannot be a normal 

 F set because they require a /'sequent =889*87. This has 

 a denominator 11*101. The calculated value for F 2 with 

 denominator 10*990 is 27502*21. The normal set are thus 

 18*18 less than the observed. As 4|8 on the limit shifts 

 it through 18*03, the observed will be ( — 4^8) F (10) nume- 

 rically if not in reality. 



The combined discussion gives F x (co ), F 2 (oo ) as very 

 close to 28382*9, 28410*0. 



The inverse D of Cu, Ag, Au. — The inverse D set seem 

 to stand by themselves, but they form a characteristic 

 feature in all the elements of this group. For this reason 

 it is desirable to consider them together in a comparative 

 study. The lines are, in* wave numbers (R.A.) : 



Cu. Ag. Au. 



D T2 . (8) 17289-22 ^ (4r) 18206-55^1 (4) 15923*38 | 



D n , (1)17439-40 248-44 (2r) 18744-15 1920-59 (4) 19113\*31 13815*54 

 D 12 . (8) 17537-66 J (lr) 18947*14 J (2) 19738*92 J 



The Cu measures above are deduced from Hasbach's obser- 

 vations, who gives the same satellite separation as Aretz. 

 This separation cannot be in error by more than *1. The 

 Ag numbers are from Kayser aud Eunge, as Raspar's are 

 clearly in error giving a doublet v = 919*87 ; the gold also 



