from the Equatorial Surface of a Magnet, 175 



analyzer, it adds actually to the virtual left-handed rotation 

 of the reflected ray, or turns the plane of polarization left- 

 handedly. 



(2) The right-handed current conspires with a left-handed 

 rotation of the first Mcol. But we have seen in art. 5, that a 

 left-handed rotation of the first Nicol is neutralized (near 

 grazing incidence) by a left-handed rotation of the second 

 Mcol ; and from this we infer that a left-handed rotation of 

 the first ISTicol turns the plane of polarization of the reflected 

 ray lefthandedly. And since the right-handed current con- 

 spires with a left-handed rotation of the first Mcol, it adds to 

 the consequent left-handed rotation of the reflected ray, or turns 

 the plane of polarization lefthandedly. 



Of course this proof assumes, and the conclusion implies, 

 that the reflected light may be considered as approximately 

 plane-polarized in all the preceding instances of conspiring 

 actions, as well as in the corresponding instances of mutually 

 compensating actions, both in the optical observations (art. 5), 

 and throughout the first four experiments. 



23. It is proved thus beyond question, at least as a very 

 approximate expression of facts, that, near grazing incidence, 

 the effect of magnetization of the mirror upon a reflected ray 

 is to turn the plane of polarization through an extremely small 

 angle, in a direction contrary to that of the Amperean currents. 

 Under what conditions, on what assumptions, may this be ac- 

 cepted as the law of the action at all incidences ? To prepare 

 for a definite answer to this question, I shall first subject the 

 statement of the law to a simple transformation. 



When the vibration reflected from the unmagnetized mirror 

 is either parallel or perpendicular to the plane of reflection, the 

 effect of magnetization is to introduce a new and very small 

 component vibration in a direction perpendicular to the primi- 

 tive vibration, the sense of the new component being that 

 assumed by the primitive vibration when turned through a 

 right angle in a direction contrary to the Amperean currents. 

 And for incidences between grazing and principal, the differ- 

 ence of phases of the two components (the primitive and the 



IT 



new) is much nearer to than to -=. 



It is important to observe here, that the results obtained in 

 the fifth and sixth experiments necessitate the assumption of 

 some such law as this, even for incidences between principal 

 and normal. For the direction of the primitive vibration is in 

 those experiments exactly parallel or perpendicular to the 

 plane of reflection, and the Mcols remain constantly in their 

 initial position of pure extinction ; so that the observed effects 



