366 OPTICAL PROJECTION 



other into a circular orbit in the, other direction. [This may 

 be worked out with pendulums, or a pendulum slide.] The 

 analyser deals with these two circular waves always meet- 

 ing, at whatever point it intersects their common circle ; and 

 at that point the two contrary tangential motions destroy 

 each other, and the radial vibration alone is left. This radial 

 vibration will be barred by the analyser, whose plane is at 

 right angles to it ; and thus any homogeneous colour may be 

 cut off, and the field made dark, by some position of the 

 analyser. This is best shown by a sodium-light. But the 

 wave-lengths differ for different colours ; and so it comes to 

 pass that in rotating the analyser, one colour is cut off after 

 the other, and the residuals give in succession, apparently, 

 and more or less perfectly, the different colours of the spectrum. 

 Thus it happens [experiment] that when rotating the analyser, 

 instead of only two complementary colours and two colourless 

 positions, as usual with a plate of selenite, we have in suc- 

 cession all the colours (more or less) in beautiful grada- 

 tion. About one half to two waves for the first crystal 

 gives the best approximation to a complete spectrum of 

 colours. 



Using a double-image prism as analyser, and an aperture 

 with the two films in the stage of a size to give two over- 

 lapping discs, it will be seen that, through all the gradations, 

 the two discs are always complementary, and make white 

 where they overlap. 



Still further, since the direction of the circular orbits 

 depends on the relation of the rectangular planes in each 

 plate ; reversing either the foundation -colour plate, or the 

 quarter-wave plate, will reverse the order of colours which 

 appear as the analyser is rotated in any given direction. A 

 plate of mica or selenite cut in half, and mounted with one 

 half reversed, or a quarter- wave plate treated similarly, will 

 demonstrate this. In two positions of the analyser the plate 

 will appear all the same colour ; but as the analyser turns 



