376 4 ON LIGHT. 



tions. We have only to suppose our lines ab, b a, and 

 their parallels p Q, / ^, inclined to each other at the 

 angle in question, and of unequal length ; to divide them 

 similarly (i.e., in the same proportion) in the points i, 2, 

 3, 4, 5 and we shall obtain a set of ellipses, none of 

 which, however, can in either of the cases have its axes 

 equal, or pass into a circle, for this plain reason that 

 no circle can touch internally all the four sides of any 

 parallelogram except a rhomb. 



(153.) Conversely, a ray circularly polarized may be 

 considered as compounded of, and may (by suppressing 

 either of them and letting the other pass, through a 

 tourmaline plate) be resolved into two equal rays, each 

 of half its intensity, polarized at right-angles to each 

 other, and differing in phase by a quarter-undulation. 

 If one of them be in advance of the other by that phase- 

 difference, the rotation will be in one direction if in 

 arrear, in the other. Elliptic polarization, on the other 

 hand, when it exists, may be recognized by the possi- 

 bihty of resolving the ray so polarized into two op- 

 positely polarized, and either of unequal intensity, or, if 

 equal, differing in phase otherwise than by a quarter- 

 undulation. 



(154.) Finally, a ray polarized in any one plane may 

 be regarded as equivalent to two equal rays, circularly 

 polarized in opposite directions of rotation, and having 

 a common zero-point. 



(155.) A ray of ordinary light may be considered as a 

 confused assemblage of rays, polarized indifferently in 

 all sorts of planes. It is, therefore, a mixed phaenome- 



