on Magnetic Rotatory Polarization in Gases. 401 



From this it is seen that the wave-lengths deduced from 

 these observations are obviously the same as those with the 

 carbon bisulphide. The composition of the light from the 

 different images reflected varies little when coloured screens 

 are used; but with the light seen directly without screens the 

 influence of coloration from the successive images is very 

 manifest. Thus we find : — 



r the direct image . . . 



. X-. 



= 0-000571 



„ first image reflected . 



. 



0-000573 



„ second „ „ 



, 



0-000574 



„ third „ „ 



. 



0-000578 



„ fourth „ „ 





0-000590 



When the number of reflections increases, the refrangibility 

 steadily decreases. 



III. Magnetic Rotations caused by the Glasses interposed in 

 the Path of the Luminous Rays. — I have already mentioned 

 that, in consequence of the exigencies of construction, the 

 luminous rays had to pass through rather considerable masses 

 of glass, subjected to an extremely weak electro-magnetic 

 action. Nevertheless the rotatory power of solids is so great, 

 compared with that of gases, that the rotations produced by 

 these masses of glass are of the same order of magnitude as 

 the phenomena to be measured. 



A very simple means of determining all the corrections at 

 the same time was to create a vacuum in the tube, and to 

 measure the rotation obtained, under these conditions, for 

 each source of light, then to fill the tube with gas, and make 

 the experiments. 



At first I tried to work in this way; but the process pre- 

 sented numerous difficulties — amongst others, exposing the 

 glasses to the risk of fracture. When the tube is empty of 

 air, the glasses support a pressure of about 50 kilog. ; now it 

 is known that under the influence of even a feeble pressure, 

 the rotatory power of bodies diminishes very considerably. 

 The real degree of influence exercised by the glasses on the 

 rotations when the tube is full of gas is therefore not ascer- 

 tained under these conditions ; and the differences are not 

 unimportant. Besides, under the strain which they have to 

 support, the glasses present phenomena of elliptical polariza- 

 tion which completely alter the measurements. In order to 

 avoid these various inconveniences, I adopted the plan of 

 always operating at pressures very nearly equal to atmospheric 

 pressure. Operating by displacement, the tube was filled 

 with the different gases to be studied, several hundreds of 

 litres of gas being passed into it. It is thus possible to ob- 



