Magnetic Rotary Disl>ersion of Solutions 5 



but once, the images produced by the rays that have passed 

 through it, three, five, etc., times, become relatively bright, since 

 the rotation is proportional to the length of the liquid traversed. 

 An indistinctness results. These reflected rays were eliminated 

 from the field of the observing telescope by making the ends of 

 the stationary cell slightly prismatic. 



With such a sensitive polarizing system all glass, even when 

 quite thin, shows more or less double refraction, and the elimina- 

 tion of this becomes a serious problem. By examining a great 

 many microscope cover-glasses a sufficient number of optically 

 good ones was obtained. The thinner the glass the freer it is 

 likely to be from strains ; but if it is less than 0.2 mm. in thick- 

 ness the pressure of liquid against it will produce not only a lens 

 effect and destroy the adjustment of the optical system, but also 

 make the glass doubly refractive. 



In constructing the tubes a cover-glass with a diameter 2 mm. 

 greater than that of the tube was laid on a level surface and the 

 tube placed on it. Without touching tube or cover-glass, either 

 beeswax or a mixture of fish glue and glycerine, depending on 

 the solvent, was flowed around them. In this way tubes were 

 finally obtained whose end-plates showed practically no double 

 refraction. The glass on the sides of the cell containing the 

 tubes being 1.5 mm. thick, holes were bored and windows of 

 cover-glass were mounted in the manner just described. 



The analyzing nicol could be rapidly rotated when necessary. 

 The setting, however, was made by means of an accurate screw 

 carrying a drum graduated to 0.005°. ^ tenth of the scale divi- 

 sions representing this quantity could be readily estimated by the 

 unaided eye. 



With the apparatus as described and the magnet excited, suc- 

 cessive settings for different tubes could be made in a few sec- 

 onds, and the field was neither distorted nor contained any ex- 

 traneous light. Successive settings on a non -absorbing solution 

 could be made whose extreme values difi"ered from each other 

 by less than 0.007° ! and with the sensibility diminished to give 

 sufficient light to read through the absorption band itself of the 

 liquids studied, the error was less than 0.01°. 



269 



