226 Mr Purvis, The influence of a strong magnetic field on the 
Within the limits of experimental error the ratio of these 
numbers is as 1 : 2. The tendency of Sb, Pb, Bi and Sn to produce 
reversals of the lines, and the diffuseness and closeness of the 
constituents of lines of high refrangibility, interfere with very 
exact measurements ; so that although the constituents of the 
lines Au 2676, Sb 2668, and Bi 2989 are not unlike the above 
quadruplets, the values of d\/\ 2 are not so strictly comparable. 
It is also of some interest to call attention to a relationship 
between the constituents of certain of the triplets vibrating 
perpendicular to the lines of force and those of the quadruplets 
also vibrating perpendicular. For example, the mean values of 
d\/ A, 2 for the constituents of the triplets of Pb 4062, 3639, 3575, 
2833 ; Sn 3175, 3033, 3009, 2840, 2706 ; Sb 3029 ; Bi 4121 ; Au 
3898, 2913 is + 2 , 02s: — 2 - 02s, and these numbers are comparable 
with the mean values of the corresponding lines of the quad- 
ruplets of Cu, A g, Au, Sb and Pb, which are + 2'08s: — 2'07s. In 
connection herewith attention may be drawn to the relationship 
between the constituents of the quintuplet of Sb 2877 (p. 221), and 
the corresponding ones of the quadruplets. 
General Residts. 
The more important results of the preceding observations 
are : — 
(1) If the metals Pb, Sn, Sb, Bi and Au are considered sepa- 
rately, there are lines comparable with each other in belonging to the 
same type : i.e. they have the same number of constituents, the 
ratio of the intensities is the same, and also the polarisations and 
values of d\j\". The comparison is particularly noticeable amongst 
the quadruplets. It is also observable amongst the triplets, 
although, of course, the comparison cannot be so rigid. 
(2) If these elements are compared with one another, there are 
lines dividing into four and three which are similarly comparable. 
And this relationship is again well marked amongst lines of the 
various metals which become quadruplets. 
(3) There are lines in the spectra of Au, Sb and Pb which are 
also comparable with certain series lines of Ag and Cu. Attention 
is called above to lines in the principal and subordinate series of 
Na, Cu, Ag, A1,T1, Mg, Cu, Sr and Ba which belong to the same type 
as described by Runge and Paschen. And one of the outstanding 
facts of the present investigation is that there are lines in the 
spectra of Au, Pb and Sb also belonging to this type. The lines of 
the latter metals have not hitherto been correlated with any law 
of series: and the results show that there are lines of widely 
