48 



Proceedings of the Royal Jrisli Acculemy. 

 Fin lines have been fovwl having 6 mnpotu-iUs. 

 ■ Table III. 



A = 2992-10 



A = 3484-25 A = 4249-61 



III \ 

 AA , . AA AA 



I A 1 A 1 — A 



f \- , \- \- 



2 

 4 

 6 



± 2-73 s 

 ± 1-37 p 

 ± 1-31 s 



2xl£o(^) 4 

 1 C 



1 1 "4 



± 1-6-2 s 7 X -230 (^\ 1 ± 2-34 s 7 x -331 (^\ 



* 1-14 p 2 ± 1-32 p 4 



* 0-46 3 I 2 1 ! ± 0-31 s 1 



A = 4331-56 



A = 4343-48 



AX 



^ ' 



-"; - 



1 ± 2-20 s 



2 t 0-92 3 



3 ± 0-64 p 



Tx -313(1) 1 



3 I 2 



- 1 1 



• ■«-. , IS .-IS! (5) 

 ± 0-91 p 6 

 ± 0-74 s 4 



Xo two of these are duplicates. Henee there is not a series of which 

 several of these are menil>ers, since, aec-ording to Preston, aU members of the 

 same series have the same Zeeman Effect. With the diminution of the 

 number of components, the application of the Ennge rule becomes more 

 difficult : and the figiii-es given in the A-columns are less definite. For this 

 i-eason Ihe evidence for or against the truth of the mle diminishes in value as 

 the number of components of tlie lines decreases. The same is true to a still 

 gi-eater degree as regards the value of such lines in supporting or dispro^•ing 

 the Eitz rule. The first, fourth, and fifth lines may be cited in support of 

 the Eunge rule. In the second and third lines, however, the denominators of 

 the rational fractions are lai'ge, teing respectively 14 and 13. Nevertheless, 

 the differences between the observed and calculated values are much smaller 

 than the possible errors due to inaccuracy of measurement, and the mean 

 difference is for each line much smaller than the calculated mean deviation if 

 the intervals had been arbitrary.* 



Already those lines do not form good examples for testmg the llitz law ; 



' 0. ilunge, Phys. Z?jt.. ?iii, 232, 190" 



