460 
DE. W. M. HICKS : A CEITICAL STUDY OF SPECTRAL SERIES. 
to this new limit, with = 666, is 48‘30. The difference between this and the 
observed 4878 is just within our assumed possible errors on two lines. 
Again, the double line 20600’97, 02'02 apparently reproduces itself in the F group, 
as Meissner says 17081 shows also a weak component. It is also in evidence in 
another relation as we shall see immediately. 
The existence of parallel F series with Watson’s separations and also 1932 has 
already been pointed out, viz., 
• 1932 F(l) 1429 • 422 • 
If these depend like the others on D series, the same separations should be found in 
the reverse order. They do not appear with Dj, but they are seen with the stronger 
line Dg = 20600, with which also the linkage relation enters. Thus 
1428-46 
(8) 18753 419-13 
(1) 19172 
1428- 68 
1429- 73 
(2) 20600-97 
(0) 20602-02 
1925±3 
1429-451 
1428-40 J 
22526 
(4) 22030 
in which there is also a mesh between 19172 and 22030 and a link back from 18753. 
There may also be a forward 1932 linked line if L.D. has d\ = —1-1. It is striking 
that a mesh is repeated in the F (l) set. Thus 
17081-46 
1429- 38 18510-84 422*50 
1430- 24 18511-70 421-64 
18933-34 
The separation of the 20600 lines is practically repeated in the 18510. 
The parallel D set, with Dg = 20556, also shows traces of the separations, with the 
mid lines not seen, in analogy with the weak mid one just considered. Thus 
(7) 18709 417-86 [19127] 1429-42 (3) 20556-84 
The absence of the corresponding lines connected to 20551 may be due to the 
scattering of the lines by displacement. The wanting lines should be [19122*07], 
[18704*6], the former corresponding to the weak 19172 of the second D group. Now 
we find strong lines corresponding to both these at the same separations ahead and 
suitable for the same displacements, viz.. 
[19122*07] 
20*79 
(5) 19142*86 
(1) 19172-29 
20-27 
(4) 19192-56 
