252 ME. J. C. MAXWELL GAENETT 



precipitated the gold from its chloride by means of various reducing agents ; and by 

 BREUIG* and later by EHRENHAFT, t who used a gold terminal for an electric arc 

 which was caused to spark under water. 



All these preparations exhibited a gradual change in colour from red through purple 

 to blue ; this change was greatly accelerated by the introduction of a trace of salt 

 into the water. ZSIGMONDY^ gives the absorption curves of a number of "solutions" 

 of gold. STOEKL and VANINO measured the absorptions of a red suspension con- 

 taining a known volume proportion of gold. Lastly, EHUENHAFT|| has made careful 

 measurements of the absorptions of "colloidal" gold. The curves plotted from his 

 measurements of the red "solutions" resemble the continuous curve shown in fig. 1. 

 Again, EHRENHAFT statesll that the absorption band of a gold suspension which 

 possessed a beautiful red colour began at X = "560 and attained a maximum at 

 X = -520, while the solution was almost transparent in the ultra-violet. Now the 

 maximum of the calculated absorption curve for spheres of gold in water (v = 1/3333) 

 occurs at X = '533.** Again, the dotted curve in fig. 1, which will represent the 

 absorption produced by a true solution of gold, does not sufficiently agree with the 

 measured absorptions to admit of the gold being in true solution in the water. These 

 results suggest that the coloration is due to diffused spheres ft of gold, although some 

 discrete gold molecules may also be present. 



* BHKDIG, ' Zeitschr. f. Phys. Ghcm.,' XXXIL, p. 127. 



t F. EHREXHAFT, 'Ann. der Phys.,' XI. (1903), p. 489. 



{ ZSIGMONDY, 'LiKP,. Ann.,' vol. 301 (1898), pp. 46-48. 



Loc,. cit., p. 108. For a discussion of their rusults see below (footnote, p. 253). 



|[ Loc. fit., pp. 505, 506. 



H Of., table given, lor. fit., p. 507. 



** Thus the differences in wave-length between the observed maximum absorption of gold ruby glass and 

 of the calculated maximum for gold spheres in glass (;' = 1'5), and between the observed maximum for 

 colloidal gold and the calculated maximum for gold spheres in water, are respectively '017 and "013, and 

 these differences are of the same si/.e. 



ft EHREXHAFT also supposed that the gold was present in the form of small spheres ; but he proceeded 

 to define the average size of these spheres (and also of those of Ag, Pt, &c., in the " colloidal" solutions of 

 these metals) by means of J. J. THOMSON'S equation connecting the radius of a conducting sphere with 

 the wave-length corresponding to the free periods of its vibration, this wave-length being assumed to be 

 that of the absorption maximum. KIRCHXER and ZSIGMOXDY (' Ann. der Phys.,' 1904, p. 575), however, 

 point out that there is no connection between size of particles and the absorption of light produced by 

 them, and this we have seen to be the case, provided there are many particles to a wave-length ; also the 

 very small size (if spherical, their average diameters would be 7///J.) of the particles of gold, the gold content 

 of which ZsiGMONDY measured would require the absorption maximum to be in the ultra-violet. 

 KlRCHXER and ZsiGMOXDY add that it would only be possible to get a large enough linear dimension to 

 give a free period if the particles were not iso-dimensional, and they conclude therefore that the gold 

 particles must be in the form of leaves or of rods ; but they do not reconcile such a form with the 

 polarisation and green colour of the cone of light emitted by the smaller particles. Since, however, we 

 find that the small-sphere hypothesis accounts for the observed phenomena, we must agree with 

 EHREXHAFT that the particles are spherical, although we cannot admit that the average diameter of the 

 spheres is correlated to the wave-length of the light most absorbed. 



