A few words may now be said about the modification of the above for lecture demonstrations. As only a moderate degree of accuracy is required, the source of light (a slit illuminated with sunlight, or any other powerful source) is brought slightly nearer the cylinder, so that the image is cast on a distant screen.
Effect of temperature.—Half the height of the cylinder is filled with cold liquid a circular piece of mica being placed above. Hot liquid is now slowly and cautiously poured over mica. When the cylinder is rotated, the light from the lower half would be the first to undergo total reflection, and this totally-reflected image may be received on a second suitably-placed screen. Light would, however, be still transmitted through the hot portion of the liquid. There would thus be a transference of one-half of the image from one screen to the other. On further rotation the missing portion would join its other half on the second screen. A slight rotation in one direction or the other would produce corresponding transference of the images from one screen to the other.
Different indices for the different rays.—A spectrum is formed by allowing light which forms the image to pass through a carbon bisulphide prism. As the rotation of the cylinder is continued, different portions of the spectrum would be totally reflected in succession, and would appear on the second screen, the spectra on the two screens being complementary.
Conclusion.—My object has been to get a fairly accurate determination of the index and its variations with an improvised and inexpensive apparatus, which can easily be set up. Even with the simple apparatus