on Magnetic Rotatory Polarization in Gases. 321 



was important to note carefully the variation in intensity with 

 the distance, and the precise value of this intensity at different 

 points. 



In fact, I found that it was necessary to interpose in the 

 course of the luminous rays plates of glass about 0*005 

 metre in thickness. Now, these glasses have a magnetic 

 rotatory power which, for the yellow light, is three thousand 

 times that of air. It is, then, conceivable that, in spite of the 

 feeble magnetic intensity to which they are subjected, these 

 plates may lead to corrections which it is necessary to ascer- 

 tain as accurately as possible. 



In order to effect these various determinations, I caused a 

 copper tube 0*50 metre long to be constructed, and placed it 

 in the centre of the different bobbins. The exact position of 

 one of the extremities of the tube was fixed by a small reading- 

 microscope; the tube was then slipped along so as to bring 

 the other end into precisely the same place as that occupied by 

 the first. The magnetic rotation of the column of carbon- 

 bisulphide was ascertained in each position; and the sum of the 

 rotations obtained was equal to the rotation of a single column 

 having the same length as the solenoid. 



The magnetic rotation was measured by the yellow light of 

 soda. 



I have stated above that the magnetic intensity varied in a 

 continuous manner, in consequence of the variations in the 

 intensity of the electric current. In all the measurements I 

 have reduced the results to what they would have been if the 

 current had been constant and given in the sine-compass a 

 deviation of 24°. 



The continual elevation of the temperature in the interior 

 of the bobbins during the passage of the very intense electric 

 current employed, introduces a most important correction 

 into the measurements. This elevation of the temperature 

 decreases the magnetic rotatory power of the carbon bisulphide, 

 first by reducing its density, secondly by a direct effect inde- 

 pendent of dilatation. This question has been studied with 

 various liquids by M. de la Rive, and with carbon bisulphide 

 by M. Bichat, who, to express the rotation of this body at 

 various temperatures, has given the formula 



l-0'00104*-0-000014* 2 . 



If dilatation alone intervened, it would give 



1 



1 + 0-0011398* + 0-000001370 1 2 

 Up to about 12° the two expressions do not differ by a unit 



