Doable Refraction of Gold Colloids. 355 



The intensity of the field was determined by the induction- 

 method, i. e. with the aid of an exploring coil and a ballistic 

 galvanometer. 



The tube G was made of Jena-glass, whose smoothly cut 

 end-surfaces were covered by glass plates of 0*15 mm., 

 cemented with Canada balsam. The glass plates were 

 previously examined and were free from double refraction 

 due to mechanical stress. The length of the tube was 

 26-0 mm. 



The preparation of the Colloids. — In order to prepare the 

 gold sols, specially pure water was used. Ordinary distilled 

 water was redistilled, first with the addition of permanganate 

 of potassium and sulphuric acid, then with the addition of 

 barium hydroxide, and finally once more in an apparatus of 

 Jena-glass. The conductivity of the water obtained in this 

 way was about 10 -6 mho. All the sols were prepared in 

 flasks of Jena-glass, which, after being carefully cleaned 

 with boiling aqua regia and chromic acid, were finally boiled 

 with distilled water. 



To obtain the sols 1 used Zsigmondy's nuclear method. 

 Potassium carbonate is added to a solution of gold chloride, 

 and the mixture is reduced by an ethereal solution of 

 phosphorus. After the lapse ol about 12 hours, the solution 

 is boiled so as to remove the ether, and a current of air is 

 passed through it in order to oxidize the phosphorus. The 

 product is a sol of a characteristically light red colour. 

 Starting from this nuclear liquid a series of solutions are 

 made with increasing size of the part'cles. As a reducing 

 agent hydrogen peroxide was used. This reaction has been 

 studied by F. Doerinchel * and A. Westgren f . 



The size of the particles was determined partly by the 

 rate of settling, partly by counting in the ultramicroscope. 

 In the former method one measures the velocity of fall of 

 the particles and obtains a value of the radius of the particle 

 in accordance with Stokes's law, under the assumption that 

 the particle has a spherical shape. Since the sh -ipe of the 

 particles differs considerably from a spherical one, we cannot 

 apply Stokes's law ; the two methods, however, show a 

 certain degree of agreement, and we are therefore entitled 

 to define r as a coefficient proportional to V m, where m is 

 the mass of the particle. 



* Zeitschr.f. An, Chemie, lxiii. p. 344 (1908). 



t Zeitschr.f. Phys. Chemie, xcix. p. 03 (1914). 



2 B 2 



