250 MR. J. C. MAXWELL GARNETT 



In fig. 2 the graph of nY/X for glass (v = I'd), shown by a broken line in fig. 1, is 

 again represented ; but here on such a scale as to have the same ordinate at X = '589 

 as that of the graphs of the observed absorptions of Au (A) and Au (B). The 

 calculated and observed curves resemble one another in having a minimum in the red 

 and a maximum in the green, although the calculated maximum occurs at about 

 X = '550, while the observed maximum falls at X "= '533. Also both calculated and 

 observed absorptions fall from green to blue, while the dotted curve in fig. 1 shows 

 that the absorptions produced by molecularly divided gold will increase from green 

 to blue, having a maximum at about X = '475. These results then, so far as they go, 

 are in accordance with the suggestion, put forward in the former memoir, that 

 the colouring agent of gold ruby glass consists primarily of diffused spheres of gold, 

 although some discrete molecules may also be present. The following is the evidence 

 which has accumulated to show that a gold ruby glass contains minute spheres ot 

 gold, many to a wave-length of light, and that it is to these small gold spheres that 

 the pink colour of the glass is primarily due : - 



(1) There are particles, presumably of gold, visible in all specimens of gold ruby glass in which the 



colour has been developed.* 



(2) Whenever these particles are of diameter less than 10~ 5 centim. they are spherical in shape.t 



(3) SIEDENTOPF and ZSIGMONDY statej : " It is only in the case of ruby glasses that the particles are 



so dense that they cannot be fully separated under the microscope." In other words, whenever 

 there are many small spheres to a wave-length of light, the glass is ruby. 



(4) We have just seen that, within the limits of experimental error, this ruby colour is that which 



would be produced by small spheres (but not by molecules) of gold, many to a wave-length, 

 embedded in the glass. 



(5) The polarisation of the cone of light emitted by the particles in the path of a beam of white light 



traversing any of the three ruby glasses examined by SIEDENTOPF and ZsiGMONDY is that 

 which would be possessed by the cone of light emitted by small spheres of metal embedded in 

 the glass. Further, the colour of the cone of light in the case of these three glasses was green, 

 while it has been shown in the former paper|| that the intensity of light of wave-length X 

 emitted by small spheres of gold embedded in the glass is proportional to (a 2 + 4y8 a )/X 4 , and, 

 according to Table I. above, this expression for gold spheres in a glass of refractive index 

 v = 1 -56 has a maximum in the neighbourhood of X = -560, i.e., in the yellowish-green. 



We conclude then that the colouring agent of gold ruby glass is metallic gold,! the 

 major portion of which is in the form of small spheres. 



The irregular blue and purple which often appear, instead of the ruby at which the 

 glass manufacturer aims, can be explained as indicated in the appendix to the former 



* SIEDENTOPF and ZSIGMONDY, ' Ann. der Phys.,' January, 1903. 

 t Cf. former paper, 'Phil. Trans.,' A, 1904, p. 391. 

 I Loc. tit., p. 27. 



See their table reproduced at p. 397 of the former paper and discussion following it on pp. 398-401 

 ('Phil. Trans.,' A, 1904). 

 || Loc. cit., p. 400. 

 U Not aurous oxide, as stated in the text-books on glass making. 



