254 MR. J. C. MAXWELL GARNETT 



Finally, KIRCHNER and ZSIGMONDY record that in a gold suspension in water 



"... A given (generally large) number of particles which diffract green light [i.e., small spheres] were 

 brought together by the addition of an electrolyte into a single particle which diffracted yellow or red 

 light with much greater intensity* than its components. With this uniting of particles occurs the change 

 in the colour of the fluid from red to blue."t 



We have already shown that, theoretically, the coagulation of the small spheres of 

 gold should produce a colour change in the fluid, from red through purple to blue ; 

 and the above quotations have indicated that coagulation accompanies the change of 

 colour. But that the coagulation takes place in the manner assumed for the purposes 

 of the theory has been shown by KIRCHNER and ZSIGMONDY, who prepared suspensions 

 of gold in gelatine, some of which preparations were red when wet, and changed to 

 blue on being dried, at the same time developing a gold-bronze reflection. J 



Now these dry blue membranes contained a number of clumps, each composed of 

 hundreds of ultra-microscopic resonators^ (small spheres) ; and these clumps were 

 comparable in size with a wave-length of light, being directly visible when examined 

 with a numerical aperture of T4 : they would therefore be capable of reflecting light. 

 Further, the change of colour to blue was most marked in those preparations in which 

 the individual clumps were most dense, || and it appears from fig. 12 below that the 

 selective absorption of red and yellow light by a gold crystallite is greater the greater 

 its density. The theoretical explanation of the change to blue requires the rays of 

 lower refrangibility to be stopped by reflections from crystallites,!! and this requirement 

 is thus satisfied. 



the absorption curve obtained from STOEKL and VANINO'K observations, we find that the observed curve 

 lies below the calculated curve, except for red light. But SroEKr, and VANINO record that the observed 

 fluid had a yellowish reflection, so that large particles (crystallites) must have been present in it ; and the 

 presence of these crystallites requires the volume // of gold, which per unit volume of the liquid is in the 

 form of small spheres, to be less than the total volume proportion /x This diminishes the absorptions 

 throughout the spectrum. But the volume proportion /*-/*' of crystallites produces absorption which is 

 much greater for the red and yellow than for the green and blue rays. The superposition of the 

 absorptions produced by /<,' and by //. - // would thus produce an absorption curve in accordance with that 

 observed. 



^ The aggregate may be supposed to be comparable in size with a wave-length of light ; the intensity 

 of the light reflected from it would thus be proportional to the square of its diameter, while the intensity 

 of the light diffracted by the small spheres is proportional only to the sixth power of their diameters. 



t KIRCHNER and ZSIGMONDY, he. tit., p. 592. 



| Loc. cit., p. 589. 



KIRCHNER and ZSIGMONDY, loc. cit., p. 576. 



|| Loc. cit., p. 577. 



U A similar explanation possibly applies to the fact that when light, transmitted through a stretched 

 membrane containing gold in suspension, is polarised in the direction of stretching, the emergent light is 

 red, but when the incident light is polarised in a perpendicular direction the colour is blue, the gold 

 clumps being comparable with a wave-length in the direction of stretching, but not in a perpendicular 

 direction. (Of. AMBRONN, 'Ber. d. math.-phys. Kl. d. k. Sachs. Gesellsch. d. Wissensch.,' December 7, 1896, 

 and AMBRONN and ZSIGMONDY, do., July 31, 1899). 



