on i\ 



Magnetic Rotatory Polarization in G 



ases. 



323 



of 14°, a rotation was obtained which, corrected by that of the 

 air, was found to be 42 / '41 ; this at 0° would give 43 / '17. 

 It is, however, useless to attempt to estimate the magnetic 

 intensities at a greater distance from the bobbins ; for a 

 column of 0'25 metre, on issuing from the first bobbin, gives, 

 at a temperature of 15°, 40 /, 6 (say 41/-37 at 0°), a rotation 

 which is nearly equal to that of the 0*50-metre tube. 



Finally, the variation in intensity with the distance was 

 studied by observing the rotation of a heavy piece of flint glass, 

 in our previous researches designated as glass No. 1 (Fell). 

 This piece was 50*028 millim. long. It was placed endwise in 

 three positions : the first was such that one of the faces coin- 

 cided with the exterior surface of the bobbin. The following 

 numbers, corrected for the rotation of the air, were thus ob- 

 tained : — 



Distance of one of 



the faces of the 



piece of flint glass 



from the first 



bobbin. 



Magnetic rotations. 



Of the flint glass. 



Of an equal thick- 

 ness of bisulphide 

 of carbon. 



millim. 

 1st 



2nd 50 



3rd 100 



3135 



14-34 



6-73 



23-49 

 10-74 



5-00 



Total 





39-23 





In the third column, the rotations of an equal thickness of 

 carbon disulphide have been calculated. The numbers thus 

 obtained give facilities for verification, since their sum ought 

 to represent the rotation of a column of 0*15 metre of carbon 

 bisulphide. Now, experiment has given 39 / for the rotation of 

 0'15 metre of carbon bisulphide under these conditions ; and 

 the preceding amount is 39'"2. This verification is as satis- 

 factory as it can be. By means of these data it has been pos- 

 sible to construct a curve which gives with great accuracy the 

 rotation of a definite thickness of carbon bisulphide at any dis- 

 tance from the bobbin. We have used this curve, as will 

 be shown, for the determination of various corrections which 

 might affect the direct measurements. 



In short, it is clear that, supposing a luminous ray traverses 

 a column of infinite length of carbon bisulphide submitted to 

 the action of the six electromagnetic bobbins previously de- 

 scribed, it will be enough, in order to take into account the 



