1010] 0)1 Colours of Sea and Sir//. 70 



approximation at auy rate, both the polarization and the colonr of the 

 Hght are easily explained. According to the simplest theory, the 

 polarization shonld be absolute and a maximum at 1)0° from the sun, 

 and the colour should be modified from that of the sun according 

 to the factor X~\ But it is easy to see that there must be com- 

 plications, even if all the particles are small and spherical. The 

 light illuminating them is not merely the direct light of the sun, 

 but also light diffused from the sky and from the earth's surface. 

 On these grounds alone the polarization must be expected to be 

 incomplete even at 90°, and the certain presence of particles 

 not small in comparison with the wave-length is another cause 

 operating in the same direction. It is rather remarkable that, as I 

 noticed in 1871, the two polarised components show much the same 

 colour. The observation is best made with a double-image prism 

 mounted near one end of a pasteboard tube, through which a suit- 

 able rectangular aperture at the other end is seen double, but with 

 the two images in close juxtaposition. AVhen this is directed to 

 a part of the sky 90° from the sun, and the tube turned until one 

 image is at its darkest, the two polarised components are exhibited 

 side by side in a manner favourable for comparison of colours. The 

 addition at the eye end of a nicol capable of rotation independently 

 of the tube, gives the means of equalising the brightnesses without 

 altering the colours. This observation, made independently by 

 Spring, is regarded by him as an objection to the. theory, and as 

 showing that the cause of the blueness and of the polarization are 

 not the same. The argument would have more weight if the colours 

 of the two components were exactly the same and under all circum- 

 stances, but I do not think that this is the case. Observations on the 

 purer sky, to be seen from great elevations, would be of interest. 

 The question is to what causes the second component is principallv 

 due. So far as it depends upon sky illumination, it would be bluer 

 than the first component. Any " residual blue " of the kind 

 described by Tyndall, and due to particles somewhat too big for the 

 simple theory, would make a contribution in the same direction. 

 On the other hand, large particles under the direct light of the sun, 

 and perhaps small ones, so far as illuminated by light from the earth, 

 would contribute a whiter light. In this way an approximate com- 

 pensation may occur, but the matter is certainly worthy of further 

 attention. 



In this connection it should be noticed that, according to the now 

 generally received electro-magnetic theory, complete polarization at 

 90° requires that the dispersing particles should behave as if 

 spherical, even although infinitely small. If the shape be elongated, 

 there would be incomplete polarization combined with similarity of 

 colour even under the simplest conditions. 



When the particles are no longer very small in comparison with 

 the wave-length, the direction of maximum polarization was found 



3 E.2 



