Pal 
The Magnetic Separation of Lines 7 
and several new lines, most of which are weak. The wave-lengths 
of these lines are determined by their distances from lines already 
known, and the Angstrom scale values determined from known 
second order iron lines. Constant use of this iron calibration with 
zircon and Yttrium lines, along with Exner’s and Hascek’s tables, 
leads me to think that the new wave-lengths may be reiied upon 
to within 0.1 Angstrom unit. The intensities for the red spectrum 
are given, and show the comparative intensities for that part of 
the spectrum. When one value only for intensity is given it refers 
to the perpendicular component, and the omitted parallel com- 
' ponent has about double the intensity. Exceptions to this are 
noted under some. lines. 
In Table Ba, are given the values of AA/A* for lines observed 
by Runge and Paschen and remeasured by the author for com- 
parison of accuracy obtainable. Barium has been found the 
poorest substance of the four studied for such a comparison. 
In Table Ba,, the line 6675.3 has its components in the ratio 
of +.635 (0, I, 3); the line 5997.4 may be represented by +.55 
(1, 2, 3); the line 5971.9 by +.30 (2, 5); and the line 4580 pos- 
Sibly by ==.10°(4, 8, 15). 
‘Table Ba, contains a list of triplets arranged in groups with 
like separation. The omitted parallel component is in the posi- 
tion of the line without field, i. e. with zero separation. Groups 
I, III, and IV stand in the simple ratio of 3-2-1. The separation 
in group II differs from group III by 1/11. The separation in 
group III Professor Runge has designated the normal triplet (see 
later in VIII, Comparison of Substances) “a”; or these groups 
are 30/2, a,a/2, and 12a/11. Groups I, Ili, [V.are represented 
in one single line 5997.4 excepting the zero component. Group I 
is also represented in 4166, group II in 5854, and group III also 
in 4554. | | 
In Table Ba, are the remaining lines of barium, and under 
“remarks” are indicated lines of several components which have 
the value of the triplet. It will be seen from these tables that the 
triplets look like lines of several components with some compo- 
nents suppressed. Or, what is more significant, the magnitude of 
the separations reduees to a few in number and recurs in different 
= 
/ 
