140 On Spectra formed hy the passage of 



disappearing at the other. To produce this effect, the "band- 

 producing " film should be set with its neutral axis at 45^, and the 

 circularly polarizing film superposed with its neutral axis in the 

 plane of primitive polarization. If the axis of either film is turned 

 thi'ough 90^, the motion of the bands is reversed ; i.e. if the bands 

 formerly moved from left to right, they now move from right to 

 left. If the two films are both placed with their axes at 45° to 

 the plane of polarization, the only effect of the circularly polarizing 

 film is to alter the position of all the bands by a corresponding 

 amount {i.e. to increase or diminish their refrangibihty), without 

 aflecting the nature of their motions.* 



A very pleasing and beautiful variety of the foregoing ex- 

 periments may be obtained by using a double-image prism as the 

 analyzer instead of the Nicol's prism. Two spectra formed re- 

 spectively by the ordinary and extraordinary ray are thus obtained, 

 which by rotating the double-image prism may be made to lie 

 parallel to one another, or be partially superposed at pleasui-e, while 

 by turning the polarizing prism the spectra can be made of any 

 desired relative intensity. Suppose that we adjust the two prisms 

 with their axes at right angles, and interpose the selenite used in 

 the first experiment, which gave a band alternately in the red and 

 in the green, we get now two spectra parallel to one another, the 

 band in the red of the one occurring simultaneously with the band 

 in the green of the other. The two bands are thus seen to be 

 strictly complementary, for the band in the red of the one spectrum 

 appears, attains its maximum, and vanishes simultaneously with 

 the similar changes of the band in the green of the other spectrum. 

 This coincident appearance of the bands, moreover, is independent 

 of the inclination of the axis of the selenite to the plane of 

 polarization, the only efi'ect of a change in which is to increase or 

 diminish the maximum intensity of both bands alike, a result which, 

 as has been noticed, does not hold with regard to the alternation of 

 the two bands in the same spectrum. f (Fig. 2, Plate XCV.) 



When the two prisms are placed with their axes parallel, so 

 that the two images of the sht are seen alongside one another, and 

 consequently the two spectra partially superposed and different 

 colours mixed, the appearance of the bands is very striking. A 

 band occurring in either spectrum is no longer black, but of the 

 colour of that part of the other sj)ectrum which coincides with it. 

 The appearance is, in fact, as if a stripe had been cut out of the 

 one spectrum through which the colour of the other spectrum is 

 seen, while on either side of the band we have in striking contrast 



* The effect of circular!)' polarizing the light before it pnsses through the 

 selenite, is simply thut the occurrence of the bands is irrespective of the inclina- 

 tion of the axis of the selenite to the plane of primitive polarization, and depends 

 solely on the position of the analyzer. 



t See note on p. 138. 



