ON COLOUES IN METAL GLASSES, ETC. 255 



In conclusion, we remark that most " colloidal solutions" of gold, even those which 

 are of a ruby colour, contain crystallites in addition to the small spheres to which the 

 colour is primarily due. Thus FAKADAY could detect the green " cone of light," 

 which indicates the absence of large aggregations, only in those liquids which had 

 been cleared by prolonged precipitation and frequent decantation ; and STOEKL and 

 VANINO found that all the gold suspensions which they examined showed a yellowish 

 reflection. A small number of the large aggregations may, however, cause the cone 

 of light to appear yellow or red without appreciably altering the colour of the 

 transmitted light. For, whereas the intensity of the (green) light emitted by a small 

 sphere is proportional to the sixth power of the diameter, the intensity of the (brown) 

 light reflected from a gold crystallite is proportional to the square of its linear 

 dimensions. Gold solutions prepared chemically appear, however, to be freer from 

 aggregated gold than are those prepared by BRE DIG'S method.* 



6. Diffusions of Silver. TJie Nature and Form of the Suspended Particles. 



We proceed to consider the absorption of light produced by diffused particles of 

 silver. The values of nY//nX for v = T6, v = 1 P 5, v = 1'3333, and v = I'O given in 

 Table II. are plotted in fig. 3, the positions of the maximum of each curve being 

 determined as in the case of gold. Since (cf. above, 4) the values of n and HK for 

 silver were all determined from the polished surface of the metal, these curves should 

 represent the absorption produced by diffused spheres of silver in glass, in water, and 

 in vacua, with only a small error, t The dotted curve in fig. 1, which represents the 

 absorptions of diffused molecules of silver in vacuo (and, on different scales, in other 

 non-absorbing and non-dispersive media), shows that the silver molecule has a free 

 period corresponding to X = "3GO, about. The existence of this free period is possibly 

 responsible for the sensitiveness of silver salts to ultra-violet light. 



In fig. 4 the graphs of wV/X for glasses of refractive indices v = 1'GO and v 1'56 

 are shown on such a scale as to have the same ordinate at the D line as the graph 

 of K for the measured glass Ag (B), of which the measured refractive index at the 

 D line was T579. The measured curve resembles those calculated, following them 

 very closely from X = '600 to X = "475, and having a maximum for a value of X 

 intermediate between those values of X which correspond to the maxima of the two 

 calculated curves.^ This close approximation of the observed absorptions to those 



* Cf. ZSIGMONDY, ' Zeitschr. f. Electrochem.,' p. 547. BREDIG'S remark, that his gold solutions were blue 

 red, points to the same conclusion. 



t These curves show that in each case the absorption is less for red than for yellow. This is contrary 

 to statements made in the previous paper (loc. tit., pp. 399 and 420) ; the errors therein made were due to 

 miscalculation for silver (red) (loc. cit., Table I., p. 396). 



J The cause of the depression of the observed maximum below those calculated is doubtless to be found 

 in the fact, to which Lord RAY LEIGH has called attention in a recent lecture at the Royal Institution, that 



