30 Prof. W. M. Hicks on the 



and because the Zeeman pattern for 17125 is that of a S 2 

 line. It gives with the known S(x) a denominator 2*545. 

 We can now proceed in one of two ways, (1) either by using 

 oll6t) = P(co) as s(l) or (2) by looking for a 3815 doublet 

 in the region given by a denominator 3'54. Method (1), 

 however, never gives a formula reproducing well other lines. 

 Here it prophesies a line Si('d) at 20879. There is only one 

 line in this neighbourhood, viz. (6)20858 already adopted as 

 D 22 (2) and showing no companion at 3815 ahead. There is, 

 however, in the 351 neighbourhood one doublet which 

 satisfies the condition, although their Zeeman patterns, so far 

 as they can be deduced from Hartmann's measures, are in- 

 determinate. They are (3)20776-50 3815-57 (6)21592-07. 

 The resulting formula for Sj is 



■036108 l* 



*/{ 



29169'85-NM m + *563932 



m 



The lines as indicated through this formula are given in 



•t 



the following table : — 



S. S. 



m = 2. 

 Change per oun = 4*808. 

 (4)13309-98 29471-77 (1)45633*56 +2-08p 



381556 



(6; 17125*54 



wi=3- 



Change per oim = 1*723. 



381557 381545 



r (4??) 41960*21 ( - \0L ) = 77*40 

 (10)24592-07 SOWS (41977-83) {\ Xn)mro .^ {10s \' )=78 .. 2l 



m = i. 



Change per oun = *814. 



(^)(1)27987*13 = 27974*23 33285-13 (5)38596*04 



These wave numbers are 31' 61 less than the calculated S 2 , 

 but they give a very exact ^(S + S) and therefore indicate 

 displaced s(4). A displacement of 42 8 l on sequent gives 

 34-19. O-C=-'05. 



