156 The Right Hon. Lord Rayleigh [May 7, 



particles are not ginall enough. Another is that they are not 

 spherical — that is to say, that it is not a matter of indifference which 

 way they are presented to the primary beam. The latter alternative 

 may be illustrated by considering an extreme case — namely, what we 

 may call a needle-like molecule, capable of vibrating only in one 

 direction fixed within it. Evidently such a molecule when obliquely 

 situated will have a component vibration parallel to the direction of 

 the incident light. 



From the experimental fact that there is such a component, we may 

 infer that the molecules of air are not in the optical sense spherical. 

 Experiments on various gases have shown a characteristic departure 

 from complete polarisation, different for each gas. Much effort has 

 been spent on determining the exact amount for each, and it is hoped 

 that the numbers obtained will form valuable material in the future 

 for investigating the structure of atoms and molecules. 



Polarisation of the Night Sky. 



"We have seen that the polarisation of the daylight sky is one of 

 the most conclusive proofs that its light is due to scattering by 

 small particles. "What of the sky at night ? Some of you will 

 perhaps be inclined to reply that the sky at night is dark, and that 

 the question whether its light is polarised does not arise. It is, 

 however, by no means the case that the sky on a clear night is 

 absolutely dark, as anyone may readily prove by holding his hand 

 with outstretched fingers against the sky. The fingers will appear 

 dark against the sky as » luminous background. 



The light is no doubt very faint, but I thought it would be 

 practicable to test whether it was appreciably polarised or not. For 

 this purpose what is called a Savart polariscope was used. Time will 

 not allow us to consider the rather complex theory of this apparatus ; 

 it must suffice to say that if the light which falls upon it contains 

 even a small part which is polarised, alternately bright and dark 

 bands are produced, which further show colour due to the composite 

 nature of white light. These bands are clearest when the incident 

 light is completely polarised, as you now see them projected on the 

 screen. But they can still be seen when the polarisation is but 

 slight. I will illustrate this by removing the polarising Nicol uhich 

 I have been using, and substituting a single glass plate, through 

 which the incident light passes. If I incline this plate, so as to 

 polarise a small fraction of the light, you see the bands, faint but 

 sufficiently distinct. In examining the light of the night sky, a 

 photographic plate is substituted for the paper screen I have been 

 using to-night, and the apparatus is designed for the utmost economy 

 of light. With two hours' exposure a definite image of the sky was 

 obtained, with the stars superposed upon it. The Savart bands 

 could be seen, but they were very faint compared with what 



