Spectrum Lines for Monochromatic Illumination. 527 



At some points only D x appeared, at others only D 2 , while 

 at others both D 1 and D 2 were found. 



The polarizing nicol is of course placed with one of its 

 diagonals making an angle of 45° with the vertical. A 

 photograph of this phenomenon is reproduced on PL VII. 

 fig. 1. The best separation of D L and D 2 was given by step 

 no. 6, and as each plate was 4*7 mm. in thickness, the total 

 thickness was 4'7 x 6 or 28 '2 mm. The seventh step showed 

 both T>i and D 2 , as the total thickness here happened to be 

 that giving circular polarization for both wave-lengths* 

 A slight inclination of a plate of this thickness would 

 cause it to transmit D x or D 2 only, by changing the length 

 of the optical path in quartz. 



In practice we may use a plate anywhere between 25 and 

 40 mm. in thickness. The best thickness is 32 mm., which 

 gives us the maximum intensity for either sodium line when 

 the other is cut off. With a plate of say 25 mm. in thick- 

 ness Dx can be completely extinguished, but the transmitted 

 light (D 2 ) will not be as bright as when a plate of the 

 correct thickness is used. 



If the echelon is illuminated with white light, the con- 

 tinuous spectrum transmitted by each step is furrowed by 

 black bands, which represent wave-lengths of light vibrating 

 parallel to the long diagonal of the analysing nicol. The 

 distance between the bands decreases with the number of 

 plates which operate at each step. A photograph of these 

 bands with the D lines superposed is reproduced on PI. VII. 

 fig. 4. 



For a thickness equal to 32 mm. o the distance between adja- 

 cent bright and black bands is 6 Angstrom units. If a dif- 

 ferent thickness is employed, and a black band brought into 

 •coincidence with D l5 D 2 will lie a little to one side of the 

 -centre of the adjacent bright band and its intensity will be 

 less than if the correct thickness is employed. 



The extinguishing of one of the D lines can be shown 

 with a natural uncut crystal of quartz, if the surfaces are 

 fairly good. The crystal is to be placed between crossed 

 nicols, utilizing two opposed surfaces, which are separated 

 by a distance of two or three centimetres. 



An image of the crystal is projected upon the slit of the 

 spectroscope, and D : will be found absent at certain points, 

 D 2 absent at others. It is another matter if a large amount 

 of light is to be used, as in experiments upon fluorescence, 

 for in this case we must use a large block of uniform thick- 

 ness free from all traces of cri/stalline irregularities. Brazilian 



2N2 



