33t 
values of the refractive indices cannot be relied upon in the violet 
and ultra-violet region. 
Â 
V 
X 
Y 
5,75 
0,999505 
0,510 
2395 
5,807 
999093 
517 
2603 
5,827 
998862 
522 
2491 
5,843 
998492 
529 
2489 
5,850 
998172 
534 
2589 
5,858 
997711 
542 
2629 
5,867 
996929 
556 
2604 
5,875 
995410 
581 
2663 
5,882 
9908 
633 
3126 
5,885 
9860 
681 
3254 
5,8866 
9770 
726 
3955 
5,916 
1,002972 
0,434 
2156 
5.942 
1532 
469 
2396 
5,960 
1164 
477 
2485 
5,977 
0934 
482 
2626 
6,013 
0658 
487 
2492 
6,055 
0523 
490 
2649 
6,137 
0335 
493 
2500 
6,200 
0291 
495 
2696 
If the values of Y are plotted against those of X as rectangular 
coordinates, the points so defined will not lie on a straight line. The 
discrepancies are large and (from a detailed study of the data) 
I find it impossible to deduce any further conclusion except that 
probably ß > 0, i. e. that a 2 > cq. If this be confirmed it would seem 
safe to infer that N 2 ^> N v The matter is thus seen to be of con¬ 
siderable interest; but to settle the question further experimental data 
are required. 
§ 15. In what precedes we have frequently alluded to the diffi¬ 
culties which occur when we attempt to treat bodies of probably 
intricate atomic structure as containing only one class of movable 
electrons. No further apology, therefore, is needed for giving here 
some further considerations which rest entirely on the basis of this 
simplifying hypothesis. The results to which we shall arrive may 
