( 92 ) 



It will also be assumed that the properties of tlie image of' the 

 source of light, as far at least as we are concernerl with them in 

 really observable phenomena, are the same as those of the source 

 itself, so that the latter may be substituted for the image. Finally, 

 we suppose that the entire luminous motion in the aether is developed 

 by means of Fourier's theorem into simple harmonic motions ; when 

 the total luminous motions in S and >S" are each other's images, 

 the same will evidently be true of those parts of the luminous 

 motions, which have a determinate period T — ^ or rather periods 

 between two definite limits 2' and T -^ d T. 



§ 3. Let ^ be a straight line, drawn from any point in the source 

 of light parallel to the lines of force, and let L denote the luminous 

 motion iviih a definite period T. existing at a distant point of Q. 

 By taking the image of the whole system, relatively to a plane 

 parallel to the line Q, it is easily seen that the image L' is exactly 

 the luminous motion, that would exist in the point considered, if, 

 the source of light remaining unchanged, the direction of the field 

 were reversed. Hence L' may very well difter from L, but, in all 

 observable properties, L' must remain unchanged, if the reflecting 

 plane be turned around the line Q as axis. 



Whence it follows, that, if all vibrations of L are resolved parallel 

 to a line i?, perpendicular to Q, the intensity produced by the com- 

 ponents must be independent of the direction of R. Indeed, Ri and 

 i?2 being two lines perpendicular to Q, and Tr\ and 1,2 the inten- 

 sities corresponding to them in the manner indicated, we may give 

 to the reflecting plane two positions F^ and P^ in such a way 

 that the image of R^, relatively to Pj, coincides with that of ^2) 

 relatively to Pj. Indicating by R' the direction of these coinciding 

 images and by iV the intensity corresponding to this direction of 

 vibration in L' — this quantity remaining the same, as was remarked 

 above, for every position of the reflecting plane — we may write 



Iri = Tr- and Ir2— J'r' ', heUCC I,.i = Jr2- 



In this way we come to the conclusion that the light propagated 

 along the lines of force, and having a definite period T, or, in other 

 words, occupying a definite place in the spectrum, cannot be polarised 

 plane or elliptically, neither com|)letely, nor partially. It can only 

 be unpolarised, or circularly polarised ; in the latter case the pola- 

 rization can be partial as well as complete. 



The light would be unpolarised, if an influence of the magnetic 

 field did not exist at all. As far as we know, the components of the 

 doublets seen along the lines of force are completely circularly pola- 



