Light in a Diffraction Grating Spectrum, 397 



images of the filament appear in different parts of the first 

 -order spectra. Sometimes these images are nearly black, and 

 sometimes they are far brighter than the rest of the spectrum. 

 On mounting the grating on the table of a spectrometer I 

 was astounded to find that under certain conditions the drop 

 from maximum illumination to minimum, a drop certainly of 

 from 10 to 1, occurred within a range of wave-lengths not 

 greater than the distance between the sodium lines. In other 

 icords, this grating at a certain angle of incidence icill show 

 one of the D lines, and not the other. 



Setting the grating at nearly normal incidence, a bright 

 narrow line appeared in the yellow, and a slightly broader 

 dark line showed up in the green. On decreasing the angle 

 •of incidence these lines approached one another, one travelling 

 up the spectrum, the other down. At an incidence angle of 

 a few minutes they came in contact presenting an appearance 

 very similar to one of the shaded lines in the spectrum of a 

 Nova. On decreasing; the angle of incidence to zero, the 

 lines fused producing uniform illumination at the spot. 



When the light is incident on the opposite side of the 

 normal from the spectrum we find the red and orange 

 •extremelv brilliant ud to a certain w T ave-leno-th, wdiere the 

 intensity suddenly drops almost to zero, the fall occurring, 

 as I have said, within a range not greater than the distance 

 between the D lines. A change of wave-length of 1/1000 is 

 then sufficient to cause the illumination in the spectrum to 

 change from a maximum to a minimum. 



The theorv of the diffraction-orating- as it stands at the 

 present time, appeared to me to be wholly inadequate to 

 explain this most extraordinary distribution of light, and 1 

 accordingly endeavoured to find out if possible the necessary 

 modifications which must be introduced. 



The ordinary theory shows that under certain conditions 

 (square groove and normal incidence for example) the directly 

 reflected light, or central image, may have certain wave- 

 lengths wdiolly absent and appear strongly coloured in 

 consequence. Coloured central images have been studied 

 experimentally by Quincke, and Rayleigh has treated them 

 theoreticallv for transmission-gratings, and Rowland for 

 gratings acting by reflexion. 



In studying the colours of these central images I have 

 found that when the plane of polarization is parallel to the 

 groove the colour is quite different from what it is when 

 the plane is at right angles to the groove. The polarizing 

 power of gratings has been experimentally investigated by 

 AVien and Rubens, but to the best of my knowledge their 



