Double Refraction of Gold Colloids. 357 



transmitting light from seven narrow ranges of the spectrum. 

 Tables III. and IV. show the values obtained for sol No. 2 

 (2?* = 22 /a//,), and for sol No. 6 (2r = 73 /jl/jl) . The dispersion 

 is very great, the double refraction, which is greatest in the 

 red, increasing with the wave-length. In fig. 4 (PL XIII.) 

 the abscissae are proportional to the wave-lengths and the 



ordinates to the ratio — — . In the case of electric and 



A. 540 



magnetic double refraction in ordinary liquids, the phase- 

 retardation is greatest in the violet part of the spectrum ; 

 gold colloids, on the other hand, show a quite contrary 

 relation. It is the high disperse sols that give the greatest 

 dispersion. 



The influence o, the degree of dispersity. — The effect 

 changes considerably with the size of the particles, as % is clear 

 from the curve in fig. 5 (PL XIII.) (Table V.), where the 



A' 

 values of |j2 X 10 7 as ordinates are plotted against the 



coefficient 2r as abscissae. As soon as the degree of 

 dispersity is about 2r=12/U/x, it is possible to trace and 

 measure an influence of a magnetic field, whose intensity 

 was 21,000 gauss ; the double refraction increasing with r 

 attains a maximum, then diminishes to a minimum, only to 

 rise again with extreme rapidity. As early as a size 

 2r = 100 fi/jL the spontaneous formation of nuclei begins to 

 disturb the preparation of the solutions, in the case of lower 

 degrees of dispersity it is very difficult to obtain homogeneous 

 sols. As I do not consider these values reliable, I do not 

 give them here. 



One must not expect always to record the same values 

 even from preparations that are apparently identical. The 

 magnetic double refraction varies considerably, but within 

 certain limits. Fig. 5 (PL XIII.) (Table V.) represents two 

 Of these series of experiments, the gold concentration being 

 in both cases c = 2'5x 10 -4 normal. It is generally known 

 that a colloid solution of a certain concentration cannot be 

 defined in any single way, its nature being due to the details 

 of the preparation. Even when following carefully the 

 same prescription, one will obtain solutions whose properties 

 are rather different. Of course it is desirable to eliminate 

 these occasional deviations, the causes of which are still 

 unknown. 



Considering the two series of experiments, one may 

 attribute the great difference in the delay of phase of the 

 same coefficient 2r to the fact that the nuclear solution of 



