DR. W. M. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES. 433 



The material at disposal then goes to show that the S separations are near 5371, 

 2649 for some definite limit and 5374, 2652 for another whose limit mantissa is one 

 oun less. 



Arguing from analogy with the successive spectra of A, Kr, X we should expect to 

 find in the observed region only lines corresponding to D (l) and 8(2). The D (2) 

 lines would be considerably shorter than the last observed line n = 33259. The S (l) 

 would be in reversed order with Sj (l) near 42100 or S 3 (l) near 34000, the first 

 absorbed by the glass apparatus used. We should expect D(l) lines up to the 

 longest observed (n = 14166) with Dn(l) lines down to at least n = 21000, and 

 showing the same kind of modified separations as in previous cases, and taken 

 account of above. S! (2) should be about 25600 with S 2 (2) about 31270, just on the 

 observed boundary and S 3 (2) quite beyond about 33900. This absence of S sepa- 

 rations is the reason why it was so difficult above to obtain accurate values of i/,, v 2 for 

 a definite limit. By themselves therefore the material is hopelessly inadequate to 

 determine the S(oo) limit, the values of A 1( A 2 , or of the various links. We have 

 only five possible or three probable choices for S to 5 significant figures, and also the 

 value of one d ( 1 ) sequent correct to a few decimals, with estimates of the F and S 

 triplet separations. The only method of attack then seems to be an indirect one, to 

 tabulate the sets of lines giving the triplet separations, to try to distinguish between 

 those related to D and S systems, to obtain as close a value as possible of the c,, to 

 determine some of the satellite separations, and from these last to attempt to find the 

 corresponding F series with the same constant separations. These F series ought in 

 each set to consist of several orders (m) at least, as the F ( oo ) all lie in the observed 

 although badly observed region. The observed separations and the values of i/j, v., 

 combined with the approximate value of the oun may enable a determination of the 

 important constants S ( o), A,, A 2 to be arrived at. 



We shall take then 5371 +dv l 2649 + cZi^ and a set about 3 larger for the values of 

 I/D i/ 2 , the two sets belonging to two limits, relatively displaced by an oun, and both 

 giving the same values of A,, A 2 . In the above the dv^ dv 2 will probably not be 

 greater numerically than 1. The calculations will be made with the 5371 set and the 

 conditions applied dv l = 1, &c., and dv l = x 1, dv 2 = y+ 1 where x, y are the changes 

 produced by the oun displacement. We start the first approximation by taking 

 S ( oo ) = 50500 + where may amount to several hundreds and in which x, y are of 

 the order 47, 2 '2. The denominators of the S 1( S 2 , S 3 limits with their differences 

 calculated from these are 



1-473697-14-591^ 



72624-2'053ff 12'54 eh, 



1-401073-12-538 (t+dvj 



32076 -'839 



1-368995-11-699 



3 N 2 



