DE. W. M. HICKS: A CEITICAL STUDY OF SPECTEAL SEEIES. 
453 
Still more striking, and as will be seen later, important, are parallel series with the 
separations discovered by Watson. In addition are found also others at 1932 behind 
F ( 2 ). They are 
(5) 15149-24 1932*22 
22103 1941*7 ±5 
25087 1922*9 ±6 
(3) 26628-53 1924± 
F(2) 
F(3) 
F(4) 
F(5) 
1429*38 (6) 18510*84 
see Note"^ 
1429*43 (6)28438*85 
1426 + 8 2997 G-5 
422*33 ( 1 ) 18933*34 
417*44 (4) 28856*29 
424 30203. 
* t (3) has a link 423 to 2^67 ± 5 and then 1432 to 25899 ± 5 suggesting a mesh in which the required 
line is wanting. Here the 1932 separation goes better with the strong displaced F(3) 24039-45, giving 
1936 ± 5. The mesh should be 
423 ± 6 2 U 67 1432 ± 
F (3), 24044 25899 
1429-42 [25474-30] 425 ±6 
t Note that 1426 + 424 = 1429-4 + 420-6 + 9. 
Any lines of the F type up to m = 4 will unfortunately lie in the ultra-violet 
beyond the observed region. F ( 4 ) should he 36690, and the largest frequency 
observed by Watson is (l) 36536. The others should he weak and in a region where 
glass apparatus would only allow strong lines to be registered. F (5) should be at 
35148*01+2^ hut is not seen. The line 35259*2 is about a v link ahead, in fact 
'y.35259*2 == 35152*4. It may, however, be noticed that 36536 above is just 154 
behind the expected F ( 4 ), so that it is the F line corresponding to the parallel F set 
above with the separation 156 (say F'). In the same series is also found 
F' (6) = ( 2 ) 34087*1 corresponding to the F'(6) = 29196.^. These are of value in 
that it gives the means of determining the limit with great exactness. Denote the 
parallel series by F'. For m = 2 using B.M.M.’s measure for F'(2) the separation is 
17081*46 + 0 —16925*43 + *05 = 156*03±*05. For m — '5 both lines have been 
measured interferentially and the separation is 28553*342 — 28397*167 = 156*175, 
correct to the second decimal place. The two separations difier by more than the 
allowable observation error, and is possibly due to the common change in sequent for 
series with different limits. In these cases in the separation with the larger m, this 
effect is very small Consequently we are justified in taking the separation as 156*17 ± 0. 
For 7n = 4 we have F ( 4 ) and F' ( 4 ) but only a L.D. line for F'(4). The separation, 
however, gives its exact value as 27009*47-156*17 = 26853*30. F' ( 4 ) is 36536*62 
±*66(d\ = * 05 ). The mean gives the limit for the F' series as 31694*96±*33 and 
consequently for F as 31851*13 ± *33, ie., ^ = *94+ *33. 
But further in the neighbourhood of calculated F (6) = [34242] are found also 
( 1 ) 34336*06, ( 3 ) 33918*08 respectively 94 ahead and 323*9 behind it. In analogy 
also are found (5) 17176*34, (3) 16757*91 respectively about the same amounts ahead of 
and behind F (2), but no other corresponding F (m) lines appear. We are justified in 
