238 ME. J. C. MAXWELL GARNETT 



From observations on gold and copper ruby glasses, it has been shown* that the 

 first stage in the formation of a crystal of those metals is the small sphere ; and from 

 observations on the growth of sulphur crystals in CS 2 , VoGELSANGt arrived at the 

 conclusion that the small sphere is always the first stage in the formation of a crystal. 

 He remarked, however, that it is by no means necessary that each of the small 

 spheres, formed as crystallisation commences, should give rise to a separate crystal : 

 the small spheres tend to coagulate; forming first rows and then groups of other and 

 more complicated shapes, until the crystal is ultimately formed. To the intermediate 

 bodies he gives the name of crystallites. 



That the spherical form of the nascent crystal is governed by surface tension, was 

 suggested in the former paper.;}; If this suggestion is correct, we should expect that 

 when the conditions are not the same in all directions, the spherical form of the 

 nascent crystal will be replaced by an ellipsoidal form. In particular, when a very 

 thin film of amorphous metal is heated until the molecules are sufficiently free to 

 allow crystallisation to commence, the nascent crystals may be expected to be 

 spheroids of the planetary type, having their axes normal to the film. Mr. G. T. 

 BKILBY has observed such spheroids in thin films of gold and silver. 



Now it will appear below that metals are not only dichroic, exhibiting one colour 

 by reflected light and, in thin films, another by transmitted light ; but that one and the 

 same metal may, as its physical condition is altered, show a great variety of colours by 

 reflected light, and a corresponding other series of tints by transmitted light. The 

 ultimate cause of all these colours is to be found in the structure of the molecule itself. 

 Juxtaposition, however, causes one molecule to afl'ect the vibrations of another. 

 Thus consider a substance composed of molecules of a given metal separated from 

 each other by the oether or by any other non-absorbing medium :|| the " effective free 

 period " of the molecule of such a substance is dependent on the geometrical arrange- 

 ment and density of distribution of the molecules in question. The optical properties 

 of the substance will therefore depend on its microstructure. The object of this 

 paper is to obtain information concerning the ultramicroscopic structure of various 

 metal glasses, colloidal solutions, and metallic films, by calculating optical properties 

 corresponding to certain assumed microstructures, and by comparing the calculated 

 properties with those observed. 



* Lot:, cit. (pp. 388-392). When writing the former paper here cited I was unaware of VOGELSANG'S 

 work. 



t H. VOGELSANG, "Sur les Cristallites," 'Archives Neerlandaises,' V. (1870), p, 156; VI. (1871), p. 223; 

 VII. (1872), pp. 38-385. 



J Loc. cit., p. 392. The further suggestion there made that iu the colourless gold glass there are the 

 molecules of gold present is, as will appear below, p. 251, erroneous. It is almost certain that in the 

 colourless glass a gold salt is in solution, so that the heating has first to reduce the gold and then to allow 

 the isolated molecules to run together into spheres. 



'Hurter Memorial Lecture,' Glasgow, 1903, p. 46. 



il As, for example, glass in a metal glass, or water in a colloidal solution. 



