538 SCIENCE PROGRESS 



experimental work by the present Lord Rayleigh (then R. J. 

 Strutt), pubhshed in the P.R.S. A 94 (191 8), verified this within 

 the Hmits of his experimental error ; but Cabannes in the 

 Annales de Physiqjie, Tome xv (1920), showed that there are 

 discrepancies not covered by experimental error. Strutt and 

 Cabannes also noticed that the polarisation of the scattered 

 light was imperfect, and the late Lord Rayleigh himself offered 

 an explanation of this, viz. that the molecules are not spherically 

 symmetrical, but have three principal axes of symmetry 

 oriented at random [Phil. Mag., 35 (191 8)]. Mathematical 

 reasoning shows that the scattered light should be proportional 

 not to (/i, - i)"/NX* merely, but to 



(^-i)VNV.(6 + 6/,)/(6-7p) 



where p is given by 



p = (A^+B^+C^-AB-BC-CA)/[3(A^+B^+C0 + 2(AB + BC+CA)] 



In this A, B, C are three parameters characteristic of the 

 three axes of the molecule and the wave-length. The work of 

 Cabannes lends support to this formula. 



In his first paper Lord Rayleigh stated quite clearly that 

 his theory of molecular scattering is not applicable in the case 

 of highly condensed media such as dense vapours, liquids, and 

 solids ; here the molecules possess only a restricted freedom of 

 movement ; their distribution cannot be regarded as a simple 

 random arrangement. One cannot equate the energy scattered 

 by a volume of liquid or solid to the sum of the energies 

 scattered by individual molecules, since the phases of the 

 secondary wavelets are no longer uncorrelated and interference 

 must take place. In fact Strutt has shown \P.R.S., A 95 (191 9)] 

 that liquid ether scatters much less light than it should in com- 

 parison with the vapour on consideration of relative density 

 alone. A theory of scattering for liquids and solids has been 

 supplied by Smoluchowski and Einstein [Ann. der Physik, 25 

 (1908) and 33 (1910)], based on the so-called " theory of fluctua- 

 tions." Scattering, in their view, is not due to individual 

 particles, but to small local variations of density arising from 

 the thermal agitation of the molecules. By reasoning based 

 on statistical mechanics and thermodynamics it can be shown 

 the mean square of the fluctuations of the density /j of a sub- 

 stance occupying volume V is RTyS/o'/NV, where N is the 

 number of molecules in a gram-molecule and /3 is the compressi- 

 bility. At right-angles to the incident light the intensity of 

 scattered light is given by 



(Je)V^VV2r»V 



at distance r from the volume V, where Ae is the fluctuation 



