446 On Phenomena connected with the Polarization of Light. 
Moreover, in reducing the thickness of the sides so as to 
render them mere edges, the phenomenon is equally produced, 
and it then becomes difficult to imagine the existence of reflexions 
sufficiently numerous to cause the observed effect. 
Bodies of the most dissimilar nature, so disposed as to present: 
a narrow slit, give rise to the same phenomenon, provided the 
sides of the slit be well polished. Flint, glass, obsidian, ivory, 
fluor-spar, have in this respect afforded almost the same result. 
Having observed that the edges of the slit require to be well 
polished, which agrees with the theory that the polarization is 
due to repeated reflexions, the effect of removing this cause of 
polarization was tried, the edges of the slit being covered with 
lampblack. Under these circumstances the polarization was 
entirely destroyed, and the idea of repeated reflexions seemed 
thus to be confirmed, only, however, to be again overthrown ; 
since on restoring the polish of one side only, the other being 
still covered with lampblack, the polarization reappeared di- 
stinctly, and in this case the existence of repeated reflexions 
seemed impossible. This experiment, taken in conjunction with 
the preceding, seems to show that there is some particular cause 
for the phenomenon. 
In order to throw some light on this subject, the attempt was 
made to examine the phenomenon when the light undergoes 
total reflexion from the sides of the slit, such reflexion having, 
as is well known, no polarizing effect. 
After several unsuccessful attempts, no better method was 
discovered than that of observing the edge of a soap-bubble 
formed in a straight tube. Under these circumstances the soap- 
bubble forms a very thin film of hquid at the centre, perpendi- 
cular to the axis of the tube, and terminated by two opposite 
concave menisci. Placing this under the microscope and illu- 
minating it properly, it was easy to distinguish through the walls 
of the tube a bright line, formed by the light that had traversed 
the film, either directly, or having undergone none but total 
reflexions from its sides. 
When the bubble was thick, the light thus transmitted exhi- 
bited no sensible polarization ; but when the bubble was suffi- 
ciently thin to give rise to coloured rings by reflexion, par- 
tial polarization was invariably found present in the bright line, 
the plane of polarization being, as before, perpendicular to the 
direction of the slit. This observation was repeated in various 
ways, and always with the same result. . 
Finally, the slits in the thin layers of gold and silver men- 
tioned above, were examined with regard to the above phenomena. 
Leaves of beaten gold were at first examined with the micro- 
scope; they presented numerous rents and slits that offered no 
