on Magnetic Rotatory Polarization in Gases. 397 



only a narrow portion of the spectrum to pass, is placed be- 

 tween the source of light and the polariscope. 



In this point of view, the want of homogeneity in the 

 source of light causes only a slight perturbation in the appre- 

 ciation of the equality of tint of the two halves of the image ; 

 nevertheless an imperfect knowledge of the wave-length of 

 the luminous rays which reach the eye causes a serious un- 

 certainty, which it is most desirable to obviate. 



In fact, it usually happens, as will be shown later, that the 

 magnetic rotations of the planes of polarization of luminous 

 rays of different wave-lengths traversing a gaseous column, 

 are very nearly in the inverse ratio of the squares of their 

 wave-lengths. The slightest variation, then, in the nature of 

 the rays, of which the equality of tint is being estimated, 

 modifies the size of the corresponding magnetic rotation very 

 considerably ; and our apparatus was sufficiently sensitive to 

 demonstrate these variations. 



A little difference in the pressure of the oxygen in the 

 blowpipe causes a variation in the temperature of the incan- 

 descent lime, and consequently in the quality of the light 

 emitted; for the higher the temperature the richer the light 

 becomes in refrangible rays. At each passage through the 

 glasses, and at each reflection on the mirrors, a small quantity 

 of light is lost ; so that the colour of all the reflected images is 

 not exactly the same. In proportion as the number of re- 

 flections increases, the colour of the corresponding images ap- 

 proaches the colour emitted in large quantity by the source. 



In order to profit fully by the sensitiveness of the apparatus, 

 and to effect the measurements with the utmost precision of 

 which they were capable, it was necessary, in each of the 

 experiments, to estimate the average length of the luminous 

 rays which reached the eye. 



Measurements were taken of the rotation of the plane of 

 polarization of these rays while passing through the same 

 column of carbon bisulphide, or of some other body, such as 

 glass, with which the phenomenon was previously studied. 



For carbon bisulphide, for instance, the relations of the 

 magnetic rotations to the yellow rays D, and to rays of any 

 wave-length whatever, were known. It sufficed, then, to 

 know the magnetic rotation of the rays studied, compared 

 with the rotation of the yellow light, in order to be able to 

 deduce the approximate value of the wave-length from it. 

 I have already had occasion to apply this method in my pre- 

 vious researches*, and subsequently f to verify the accuracy 

 of the results thus obtained. 



* Annates de Chimie ct de Physique, t. xii. 1877. 



t Comptes Rendus de V Academic des Sciences, t. lxxxv. 



Phil. Mag. S. 5. Vol. 12. No. 77. Dec. 1881. 2 H 



