( 355 ) 
o„iy effect is an increasing of the intensity of the dooblet components 
It i easily shown that also multiple reflexions from those parts of 
the wall of the capillary, which are normal to the field, will only 
modify the intensity. 
12. Experiments with absorption lines. I have made also some 
experiments on the sensitiveness of the inverse effect, asodmmflame 
placed between the perforated poles of a large do Bois-electromagnet 
being traversed by the white light of an arc-lamp. A large concave 
Rowland grating was made use of for the analysis of the light. 
The Rowland grating can be used for the study of the inverse 
magnetic effect, because the brightness of the arc-lamp more than 
compensates the feeble reflexion by the grating. 
A double quarter-wave plate and a Nicol are traversed by the beam 
By means of the mounted Nicol the right-handed circularly polarized 
component in one part of the field, the left-handed in the other can 
be made absolutely black. The corresponding position of the Nicol 
we call the zero-position. In this position not the slightest trace ot 
absorption can be seen at the places of the other components, which 
only appear after a considerable rotation of the Nicol. The positions 
of the Nicol, on either side of the zero-position, were found at which 
the earliest recognition of the absorption lines was possible. The first 
trace of the lines only became visible after a rotation of 20 from 
the zero. Hence, we conclude, that the absorption must already be 
considerable before being perceptible and that ordinary light in the 
components of the doublet can be recognized more readily by obser¬ 
vations on the direct effect. 
13. Conclusions. I think I have proved by the experiments 
described 1 that we must accept, till the contrary has been proved, 
that in the ease of line spectra the displaced') components of magne¬ 
tically subdivided lines, emit parallel to the lines of force completely 
circularly polarized light. A 
2. that the amount of ordinary light which might be emitted 
simultaneously with the circularly polarized light, 
several spectral lines less than 1 % 
spectral line. 
~^Tl expressly speak of displaced components in order to except 
which has the same position as the unmodified line. The groups of oppositely 
circularly polarized components then have a central line in common, w ic mus 
be unpolarized. The quintets A = 3788.01 and A = 8743.45 of the iron spectrum 
investigated by H. Becquebel and Deslandres and lately measured by Mrs. 
Bilderbeek in the Amsterdam laboratory, provide examples of the type. [Note 
added to the englisch translation]* 
the case of 
of the total intensity of the 
