278 MR. J. C. MAXWELL GAKNETT 



shows that, for a film of such thinness, the intensity T of the transmitted light is 

 greatest for red light.* This red colour has been seen both by FARADAY! and by 

 BEILBY| in parts of their green films. FARADAY says the red colour was extremely 

 faint but appeared to have an objective reality, while BEILBY describes the effect as 

 that of " an irregular film of pink jelly." 



It appears that extremely thin films of gold are, by surface tension, drawn up into 

 green patches, leaving larger areas covered by an almost transparent, but faintly red, 

 film. The effect on the unaided eye is that of a transparent green. 



The silver leaf used by BEILBY was over 300 p,^ thick. It therefore comes well 

 inside the range for which the analysis of 10 applies. Now Table VIII. , or fig. 9, 

 shows that, for amorphous silver of normal specific gravity (//. = 1), /c/X is least for 

 the more refrangible rays. Again, Table IX., or fig. 11, shows that, for p. = 1, M is 

 greatest for the same rays. It follows therefore, from equation (23) above, that, on 

 both these accounts, the light transmitted by silver leaf should be blue ; and, in fact, 

 silver leaf transmits a deep blue light. The approximately equal values of the 

 reflecting power, R , shown in Table IX., or fig. 13, for /A = 1, correspond to the 

 almost colourless reflection from polished normal silver. 



Consider now the colour changes which, according to figs. 8 to 13, deduced from 

 the calculations of 10, should accompany a diminution in the density of gold and 

 silver films from its normal value (/u. = 1) to zero (p. = 0). This diminution of density 

 may be conceived either as an increase of the distance between adjacent molecules or 

 as due to the aggregation of groups of neighbouring molecules into small spheres. 

 For geometrical considerations show that so soon as two spheres form adjacent to one 

 another in an otherwise amorphous mass of metal, the density of the mass must begin 

 to diminish. And it has been shown that the calculations in question are applicable 

 whether the metal is in small spheres or in an amorphous state, and thus when it is 

 partly in the one condition and partly in the other. 



Taking first the case of gold, it appears from figs. 8 and 10, in conjunction with 

 equation (23), that, as /x begins to diminish from unity, the absorptions of red and 

 yellow light increase rapidly, owing both to the increase of HK/\ and to the decrease 

 in M . Meanwhile, owing to the decrease of the ratio of M (green) to M (blue) and 

 to the increase of (wic/X) (green) (w/c/X) (blue), the relative intensity of green to blue 

 in the transmitted beam diminishes. Thus the first effect is to make the transmitted 

 light bluer, and this effect continues until //. = about 75. As p continues to diminish 

 below this value, the absorption of red rapidly decreases until, at /u, = '68, in a very 

 thick film, the absorption of red has become as small as that of blue. The film is 



* Cf. Table IV., p. 406 of former paper, and Table VIII. above, 

 t Loc. cit., p. 400. 

 I Loc. cit., p. 40. 



When exp ( - iird . WK/A) is the dominating factor in T . The corresponding value of /* is less in 

 thinner films for which M is very important. 



