601 
quartz plate, rotating the plane of polarisation through 90° be intro- 
duced before the slit of the spectroscope, then only the outer com- 
ponents of the resolved spectral line can be made to disappear. 
A second observation was made with the slit only of the spectros- 
cope. The lens of the collimator being removed the slit of the spec- 
troscope could be seen distinctly while viewing along the axis of the 
spectroscope. Looking through a calespar rhomb the slit appears 
double. With a wide slit, illuminated by the radiating tube, the two 
images exhibit the same intensity ; a narrowing of the slit gradually 
makes the image due to the vertical vibrations more brilliant than 
the other one. 
I will mention two cases in which errors may be introduced by 
the polarization impressed by the narrow slit. This happens in the 
first place in the case mentioned above of the comparison of the 
intensities of resolved components vibrating in different planes. In 
the second place when the resolution of lines originally diffuse toward 
one side of the spectrum is investigated, apparent shifts and dissym- 
metrical separations may result. Is the original spectral line diffuse 
toward the red then a decrease of the intensity of the central line 
of a triplet will cause an apparent shift relatively to the outer com- 
ponents toward the violet. The reverse will be the case if the 
original line is diffuse toward tbe violet. 
The apparent shift now under consideration has had no influence in 
the experiments concerning a change of wavelength by magnetic forces 
of the line Hg 5791, which the author’) and (independently) GMeLiN®) 
discovered at the same time. Its existence could be demonstrated 
also by the method of Fasry and Perrot, a method not Cue ay 
upon the use of a narrow slit. 
It is a favourable circumstance that a quartz plate introduced 
before the slit of the spectroscope and giving a rotation of the plane 
of polarization of 45°, eliminates at the same time as well errors 
due to the polarising action of the grating as those caused by the 
narrowness of the slit. 
1) ZEEMAN, Change of wavelength of the middle line of triplets. These Preceed- 
ings February 29, 1908, in print in the Dutch edition March 12, 1908, in the 
English March 29, 1908. 
2) P. Guerin. Uber die unsymmetrische Zerlegung der gelben Quecksilberlinie 
5790 im magnetischen Felde. Physik, Zeitschr. p. 212 (eingegangen 24 Febr. 1908) 
appeared April 1, 1908. 
