DE. W. M. HICKS: A CEITICAL STUDY OF SPECTRAL SERIES. 457 



Furthermore the observed separation 4775 requires, on this new limit, a displace- 

 ment 664 + 13'9d' 1 . With di>i = 'II we have A t the same as before, i.e., the 47'86 is 

 the proper v l separation of the new limit. The v 2 proper to this limit should be 19'15 

 giving S 3 = 27885 '90. This is not seen, but the it-linked line is possibly given by 

 L.D.'s line 27995'54-106'8 = 2798874. The actual S 2 line appears split up into 

 the additional displaced lines, each (l) 27858'26, (l) 27868'82, (l) 27873'63 from 

 observed S 2 . As a parallel tt-link to the first we find also L.D., 27964'2 + 4 106'8 

 = 27857'44 for actual 58'26. The oun is too small to decide whether the displace- 

 ments are produced in the limit or the sequent. 



The limit for this new set, 52112'20, is more in step with the progression of the 

 other gases, viz. : 



Ne. A. Kr. X. RaEm. 

 52112 51731 51651 51025 50403 



It is possible to get an estimate of the u.v links although there is no means 

 apparent at present of getting the exact value of s ( 1 ). Either of our S groups gives 

 the same value for s(2), viz., 53081'24-28787'86 or 52112'20-27818'89 = 24293'3. 

 The denominator of this is 2'1248, so that the n + a./2 of s(2) is '1248. We can get 

 an estimate of the value of s(l) from the law that (a+^-A 1 )/(/ i c-|-|-A 1 ) is about '2 in 

 the other periodic groups. The corresponding values in this group are A, '189 ; Kr, 

 '198 ; X, '222. If the ratio '18 is taken for Ne n + a. comes to about '1139 with s (l) 

 about 89000. These are only rough estimates, but the values of u.v will only alter 

 slowly with considerable changes in s (l). The values calculated from this 89000 + ^ 

 with Aj = 666 are 



u= 10678 + '0017 v= 10G'86 + '0018 



This shows that although the value of s (l) may be extremely rongh, those of n.r 

 may be relied on within a few decimals. 



One is inclined to regard the weak S set as that which is analogous to those 

 determined in the other gases, and which certainly belong to the blue spectrum, and 

 that in Neon, which shows only one spectrum, it is composite, and the blue not 

 strongly developed. The question naturally arises whether there is any evidence of 

 an unstable or weak D set with the same limits, i.e., about 969 behind the former D. 

 There is a set in this neighbourhood 



(0) 20173'82 42-86 (2) 20216'68 



of the right order of inverted intensities and a small i/ 1} possibly a case of the common 

 D triplet modification. With the limit 52112'20 the mantissa of 20173 is 853094, 

 which is 373 greater than that of the old d. This is within error limits of 26S = 376 

 This may be explained by the supposition that the d sequence is not affected by the 



3 Q 2 





