gg I'KOrEKTlKS OK OPTICAL (iKATlNOS. 



sist of about twenty apertures. As more slits are uncovered, 

 it will be seen that the spectra, although their centres remain 

 unchanged in position, become narrower and sharper, and the 

 colours purer, and no longer confined to the elges. Also, in 

 the dark regions between the spectra, now wider than before, 

 secondary maxima appear. These are much narrower and 

 fainter than the spectra themselves. It will be found that 

 if N is the number of apertures exposed, the number of sec- 

 ondary maxima is N — 2. They can readily be counted, for 

 N = 3. 4 and 5, but after that, become rather too faint and 

 close together. On replacing the white light of the lamp by 

 a sodium flame, it will be seen that the spectra are now quite 

 narrow lines. This was, of course, indicated by the purity 

 of the colours in the white light specti'a. 



Draw curves as in section 5, Figure V., and dimension 

 them tc show the positions of the centx-es of the spectra, the 

 width cf the dark bands separating them, and the positions 

 of the secondary maxima. Verify that the positions of the 

 spectra have not changed on increasing the number of 

 apertures. 



SECTION 8. THEORY OF SECTION 7. 



Fcr simplicity, the discussion will be confined in the 

 first instance to illumination by homogeneous light. As in 

 section 6, the visibility curves from the different apertures 

 are superimposed, and the illumination at any point is I'mited 

 by the ordinate of the visibility curve. 



Consider first a grating with a large number, N, of lines. 

 Let E be the total width of the grating. It may be looked on 

 as an aperture of width E divided into N elements. Figure 

 III, of section 4, represents this case as well as the case of a 

 slit. The scale i?, however, different. The whole discission 

 of section 4 will also be seen to hold, but instead of there be- 

 ing an infinite n imber of elements supplying 'ight, there is 

 a large finite number. E is much larger than e, and there- 

 fore the distances from P of corresponding parts of the 

 diffraction patter.n are much smaller, and instead of a broad 

 central band, there will be a fairly narrow central line 

 flanked by secondary maxima, the whole phenomenon being 

 confined to a very narrow region near P. The Tirst minimum 

 occurs where •! f X /K or calling d f (the very small angle 

 which separates the maximum from the minimum) •l'' it dccurs 



I'. 



