600 
the ratio present in the light as it is emitted by tne source. In some 
eases the observer sees only a faint central component and two 
intense outer Components, whereas the true ratio is just the reverse. 
In order to obtain in the image the true ratio of the intensities I 
suggested to introduce before the slit of the spectroscope a quartz 
plate of such a thickness, that the incident light is rotated through 
an angle of 45°. *) 
Besides the mentioned polarizing effect of the grating there is a 
second cause tending to make the ratio of the intensities of com- 
ponents of different direction of vibration in the image different 
from that corresponding to the constitution of the emitted light. 
I mean the polarization impressed upon light which traverses fine 
slits. Since Fizeavu ®) this effect is well-known, but the errors which 
may ensue from it in investigating spectral lines magnetically resolved 
have not yet been pointed out. 
The following simple experiment is easily made. A vacuum tube 
charged with mercury is placed in a horizontal magnetic field. The 
emitted light is analysed by means of a spectroscope securing great 
illumination and high resolving power. The slit must be under the 
control of the observer at the eye-piece. The two yellow mercury 
lines, which are resolved into triplets or the green mercury line, 
which splits into three groups each of three lines may be observed. 
If the slit is rather wide then the central components of the yellow 
triplets may have twice the intensities of the outer ones; the three 
groups of the green mercury line have about the same integral inten- 
sity if not wholly resolved. If the slit (made of platinoid) be narrowed 
gradually, the intensity of all components decreases, but that of the 
central component or group more than that of the outer ones. At last 
the central components of the triplets and even the middle group of 
the brilliant green line can be made to disappear entirely *), whereas 
the outer components remain visible. From these observations we 
cannot but conclude that the vibrations perpendicular to the slit at 
last hardly traverse the narrow slit. 
The correciness of this explanation may be inferred from the fact 
that the rato of the intensities changes gradually during the narrowing 
of the shit. 
The view may be controlled by the following observations. If a 
) Le. p. 291. 
2) Fizeau. Ann. de Chim. et de Phys. Vol. 63 p. 385. 1861. 
8) This extreme case involves the use of an exceptionally narrow slit rarely 
employed in practice. 
