240 
PROF. A. FOWLER ON SERIES LINES IN SPARK SPECTRA. 
These Mg doublets are analogous to the wider doublets of Ca, Sr, and Ba :—(l) They 
occur under precisely similar experimental conditions; (2) They show similar Zeeman 
effects (Bunge and Paschen) ; (3) The separations of the components—Mg 91'5, 
Ca 223, Sr 801, and Ba 1691—are roughly proportional to the squares of the atomic 
weights ; (4) The limits of the series follow a natural sequence, decreasing in the 
usual way as the atomic weight increases. 
The analogy, however, is incomplete. In Ca, Sr, and Ba, the first negative terms of 
the D series, (given by m = 2 in the Hicks formula) correspond, as already mentioned, 
to observed lines (Tables III. and IV.) having satellite separations identical with those 
of the doublets in the associated F series; in these cases the satellite lies on the violet 
side of the more refrangible component of the I) doublet, and not on the red side of 
the less refrangible component as in the positive terms of the same series. In Mg, on 
the other hand, the D pair given by m — 2 in the Hicks formula has a positive sign, 
and no lines have been observed near the position of the negative term (25,350 or 
about X 3940) given by m = 1 . An associated F series, by analogy, would be expected 
to have its limits near 110,830 (= 85,480 + 25,350), but there is no evidence of the 
existence of such a series. 
The absence of a negative term in the observed I) doublets of Mg suggests, as an 
alternative view, that the associated F series may be derived from the first line 
actually observed, namely X2798. In that case Lorenser’s formula would give for 
the limit of the F series : 84,893'0 —35,7287 = 49,164'3, and the Hicks formula 
85,479-8-35,7287 = 49,751'1. 
These figures at once suggest the limit 49778 of the “ 4481 ” series; and the 
available data for the Wide doublets are, in fact, consistent with the assumption that 
the agreement is exact. It would seem then that the “ 4481 ” series may be a 
Fundamental series deriving its limits from the first positive line of the D series of 
Wide doublets. In that case the D line X 2798 would be expected to have a 
satellite with the same separation as the lines of the “ 4481 ” series, namely, 
Sn — 0"99. A special search for the satellite has been made in the photographs of 
the arc in vacuo taken in the fourth order of the grating, but it has not been 
directly observed. Indirect evidence of its possible existence, however, is afforded by 
careful measurements of the intervals of the XYide pairs. The new wave-lengths are 
given on the International Scale in Table VI. It will be seen that while the pairs 
of the P and S series are in close agreement, indicating a normal separation of 91 "5, 
the interval of the D pair is not less, as would usually be the case if a satellite were 
present, but greater. This is confirmed by the recent measures of these lines by 
NachenA The measurements thus suggest that there is a satellite on the more 
refrangible side of the chief line, the separation being comparable with that in the 
“ 4481 ” series if we assume that the tabulated X 2797'989 refers to the chief line 
itself. Such a reversal of the usual position of the satellite, however, would accord 
* ‘ Zeitsch. f. Wiss. Pliotog.,’ 1913, XII., 2, p. 59. 
