EPIPOLIC DISPERSION OF LIGHT. 151 



gone some qualitative modification which might be considered as rendering it spe- 

 cifically difFeient from the incident beam. It cannot be the mere tinge of colour 

 which the loss of so small a portion of blue light has given to it. There is still plenty 

 of blue light left, and the experiment on sulphate of manganese proves that a mere 

 absorption of a much larger proportion of the blue rays has not the same effect. 

 Moreover the portion of light dispersed traverses the solution of quinine with perfect 

 facility, proving that no peculiar absorptive power is exercised by that medium on 

 these rays ; nor indeed would the separation of such rays by dispersion at the surface 

 in any way tinge the medium itself with a complementary tint, but only the residual 

 beam. 



I come now to the examination of the dispersed portion of the light. As just 

 remarked, when once dispersed it is freely transmitted. The epipolic colour is seen 

 as well, in a long test-cylinder filled with the solution, at the bottom of the tube 

 as at the top, when viewed by an eye situated in its axis, supposed vertical. If all 

 light be cut off from the tube by a sheet of black paper rolled round it, except from 

 the lowest inch of its length, that inch is seen to gleam with quite as intense a colour 

 as when the uppermost inch only is so exposed. 



I have already had occasion to remark that the epipolic tint is a compound one. 

 To obtain a pure ray for prismatic analysis, a cylindrical glass jar with perpendicular 

 sides was partly filled with the quiniferous liquid and placed in a strong light, the 

 whole anterior side being coated with black paper rising somewhat above the level 

 of the liquid. The eye was then placed in such a position, below that level, that the 

 visual ray proceeding from it would suffer total reflexion at the under surface. For 

 comparison, a similar vessel of water, similarly shaded, was placed beside it. The 

 surface of this, so viewed from below, was of course perfectly black, no ray from 

 above being able so to penetrate it as to reach the eye. Not so the quiniferous solu- 

 tion. In this the under surface was wholly visible, of a fine blue colour, considerably 

 deeper in tint than in the ordinary mode of viewing it, though not of so rich and 

 saturated a character as the epipolic blue of the fluor. It was, however, much more 

 luminous, and being thus completely purified from all possible admixture of regularly 

 refracted or reflected light, was well-adapted for prismatic analysis. 



By raising the eye exactly to the horizontal level of the surface of the liquid, 

 the whole of that surface became of course foreshortened into a narrow blue line. 

 And in this situation it became perfectly evident that this line was not a mere elon- 

 gated ellipse, the perspective representation of the circular area of the surface, but a 

 very narrow parallelogram, having a breadth of about a fiftieth of an inch, of a vivid 

 and nearly uniform blue colour over its whole breadth. This proves that the epipolic 

 dispersion takes place within the liquid, and almost wholly within a distance not 

 exceeding one-fiftieth of an inch from the surface. I say almost wholly ; for when 

 a sunbeam was directed downwards on the surface, by total reflexion from the base 

 of a prism, a feeble blue gleam was observed to extend downwards below this vivid 



