LIGHT: COLOUR 295 



parallel flat sides, such as may be bought for a few pence 

 almost anywhere. Either is nearly filled with coarsely 

 powdered glass. For the bottle with plane sides it does not 

 matter what this glass is, if homogeneous ; with the other 

 bottles, it is best to purchase one or two others of the same 

 make, and having heated one nearly red-hot and quenched 

 it in cold water, to pound it up. The powdered glass sold 

 ready-made, generally varies too much in density for this 

 experiment. This powdered glass, in the cell or bottle, is of 

 course opaque to the beam from the lantern. The bottle is 

 now to be filled with a fluid of the same refractive index as 

 the powdered glass. This is easily prepared by mixing in the 

 required proportion benzol (which is less dense than glass) 

 and carbon disulphide (which is more so). This is done by 

 trial and error, and when the required density is obtained (if 

 the powdered glass is reasonably uniform in character) the 

 bottle appears clear, or at least transparent. This, however, 

 is not all. It has been shown how dispersion varies in 

 different substances. Owing to this, the refractive index can 

 only be brought precisely the same, for one colour of the 

 spectrum. We therefore place a rather wide slit on the 

 lantern nozzle, and focus its image on the screen ; then 

 interpose the cell filled with the glass and fluid. The image 

 will now be seen, still sharp for the colour which happens to 

 be exactly corrected let us suppose a green slit may appear 

 on the screen while the rest of the spectrum will be diffused 

 round this in a nebulous haze, of the complementary colour 

 to that of the slit. 



168. Anomalous Dispersion. The remarkably anomalous 

 proportions dispersion sometimes assumes can be illustrated 

 on the screen ; but after further experience resulting in occa- 

 sional failure, I am bound to confess that with the lime-light 

 the experiment is hazardous and difficult. I formerly described 

 the use of a prism bottle in a trough with parallel sides, but 

 this is too uncertain to be trusted, and I have abandoned it. I 



