( 291 ) 



was strongly polarized. The vertical vibrations were strongly 

 preponderating. 



The influence of a rotation of the plane of polarisation of the 

 yellow mercury light on the distribution of intensities in the triplet 

 was then examined. The plane of polarization was rotated by 

 means of quartz plates with faces perpendicular to the axis placed 

 in front of the slit. I had at my disposal two small plates of 2.15 

 resp 4.17 mm. thick. According to Gumlich l ) the rotation for mercurj 

 light of wavelength 5769 in a quartz plate 1 mm. thick, is at 

 t — 20° 22°. 718 and hence the rotation in my plates amounted to 



22.72 X 2.15 = 48°.90 en 22.72 X *-17 = 94°.7. 



The change in the distribution of light is at once apparent. In 

 figure 3 the outer components are hardly visible. The negative 

 reproduced corresponds to the case in which the plate rotating the 

 plane of polarization 94°. 7 is in front of the slit. 



Figure 2 corresponds to the case in which the incident vibrations 

 are inclined at about 45° to the slit. It may be remarked that in 

 this case the real distribution of intensities between the components, 

 as existing in the emitted light, is observed. 



Vertical and horizontal vibrations now being equally present in 

 each of the components, and hence the circumstances as to vibrations 

 being the same for the three components, the polarization by the 

 "rating is eliminated. 



The distribution of light in figure 2 is certainly not in contradiction 

 with equation (1) and eye observation seems to confirm it numerically 

 also. Of course a photographic reproduction is not sufficient for a 

 comparison of intensities and a numerical test must be reserved for 

 a future paper. 



For an estimation of the real ratio of intensities of the components 

 of a divided spectral line henceforth care must be taken that for the 

 region of the spectrum under review the vibrations in the incident 

 light are inclined at an angle of 45° C. to the slit. 



If in the case of a complicated division of a spectral line some 

 components are weak, it will sometimes be possible to streng- 

 then these components by placing a quartz plate of suitable thickness 

 in front of the slit. This will be feasible in all cases in which the 

 incident vibrations are not those most favoured by the grating. 



Of course also with other spectroscopes this device does apply 



i) Gummch. Wied. Ann. Bd. 64 p. 333. 1898. 



19* 



