mi. W. M. HICKS: A CRITICAL STUDY OF SPECTRAL SERIES. 
441 
C.K lines. Behind this is a line (2) 22079'34 W. {d\ = ±'05) corresponding to the -Sy 
displacement in the sequent— i.e., the modified D separation. With S(co) its 
mantissa = 804161— 2677^—26‘77d2/i +5‘35p 
= 25 (32166-44-r071^-r07d:/i + -21p). 
If this be combined with the mantissa of 29964 = 913165 = 511fi giving 
Ag = 32166‘27 —1'116^, there resnlts the equation 
•14 + -045^+-21p-r07rii.i = 0. 
This can be satisfied by ^ = 0, = 0, and p a fraction. It does not therefore heljT to 
a closer determination. With good measures it should be practicable to find dv^ 
within '05 andp a small fraction. This equation would then give the small correction 
for f and so increase considerably the degree of accuracy of and S. The particular 
point however gained is that here is found one of the fundamental d sequences 
depending on pure multiples of Ag. 
(4) They all show —5640 links when this link lands in the observed region except 
22868. Where the measures are reliable they congregate round a value 5633 + . The 
22079 of the last paragraph is 563373 above the 16445. This is a further justifica¬ 
tion of 22079 belonging to a displaced sequent. 
With 20438 as Di (l), Rydberg’s table gives D (2) as in the neighbourhood 37486 
±100. A ti sounder gives the region 23806. There is a line (lO) 2378375 (W.) 
which if linked in this way gives (2) = 3746375. The two lines m — 1, 2, and 
S ( CO) give the formula 
n= 50403-N/|m + -909601+ 
./I m ] 
with I) (3) = 43242'02. The e-sounder requires 19563. The only line in the neigh¬ 
bourhood is X = 5105 by Ramsay, who says his measurement is very rough. If we 
allow dX = 5A, the wave-number is 19583 + 20, and it may be the line sought for. 
There seem also other groups as in X. One instance is adduced in the next 
paragraph. 
I end the discussion of the RaEm spectrum by a consideration of the source 
of the 5640 separation. In X we found the conditions satisfied by oun displace¬ 
ments on the F ( oo) of 5 Aj —^ 1 , and 2 Ag + Ot^j. But here the values of the separations 
themselves seem very indeterminate. The values as arranged on p. 431 would seem 
to point to 5633, 5649 with vo about 2800 and 2820. The limit of the particular 
F series on which our whole discussion of RaEm is based is 29964‘20. As a fact, how¬ 
ever, this limit can only generate in the proper neighbourhood separations of 5646, 
2806, and the displacements are 5A2-6^i, 2A2+2(5i. The dependence on these 
3 o 2 
