Production of Circularly Polarized Light. 521 



while the other is near the critical angle. The larger angle 

 is chosen for reasons given below. 



Now let there be two rhombs placed as shown in fig. 2, 



Fig. 2. 



which is a medial section. A beam of plane-polarized light 

 enters along CD, is reflected at D, E, F, and G, and emerges 

 along G, H, so that C, D, G, H, are collinear. There will 

 be a retardation of one component on the other by A, and 



n 



A = — if the angle of the rhomb is 75° for ordinary glass 

 v == ). Hence for such a rhomb the total gain of one 



7 IT IT 



component on the other will be 4 x -$ =47r— tt, and there- 



o 2, 



fore the emergent light will be circularly polarized *. 



Further, since C, D, G, H, are collinear, on rotating the 

 system there will be no lateral motion of the circularly 

 polarized beam, and no readjustment of successive pieces of 

 apparatus is needed. 



The fact that the relative retardation of one component on 

 the other has to be of the form mr + y, where 7 is an acute 

 angle and n odd, in order that ifr should come out real is in 

 accordance with the result first found by Lord Kelvin, that 

 the phase-differenee between the components resolved in and 

 perpendicular to a principal section (for substances at our 

 disposal) is oblique t- 



In total reflexion the phase-difference introduced between 

 the vibrations executed in and perpendicular to the plane 

 of incidence, is not entirely independent of the wave-length, 

 although nearly so, and it is best to choose the angle of 

 incidence so that the dependence of A on X, or the " colour 

 effect," as we may call it, is as small as possible. 



* According- to Lord Kelvin the component polarized parallel to the 

 plane of incidence is accelerated on the perpendicular component. See 

 Baltimore Lectures, p. 400. 



t Loc. Clt. 



Phil Mag. S. 0. Vol 21. No. 124. April 1911. 2 M 



