Maxson — Determination of Small Amounts of Gold. 271 



revolving cathode. Solutions of less concentration were then 

 obtained by suitable and accurate dilution. 



A preliminary series of qualitative and quantitative experi- 

 ments gave evidence that the coloration was of a quantitative 

 nature and that the suspensions were sufficiently stable under 

 suitable conditions. 



The Gallenkamp colorimeter was used in the subsequent 

 quantitative experiments immediately following. The delicacy 

 of the readings was increased by placing the instrument in a 

 light-tight box pierced with suitably situated holes ; the influ- 

 ence of external colors was avoided by admitting light to the 

 instrument from a ground glass plate, which was found to be 

 very efficient. 



It is a well known fact that small amounts of electrolyte 

 will rapidly change "red" gold to the "blue" modification. 

 It is necessary therefore to conduct the comparisons in a room 

 reasonably free from fumes, and to have all containing vessels 

 free from soluble material. It was found that flasks which 

 had been treated with steam for a few minutes gave the best 

 results. Red suspensions contained in such flasks gave no trace 

 of blue after an interval of several weeks. 



Using the precautions outlined above, the results shown in 

 the following table were obtained. The error of the personal 

 equation was determined by means of a series of comparisons 

 conducted with different concentrations of the same suspension, 

 and covering a range of amounts of material identical with 

 the amounts handled in the experiments given below. 



The forty-nine analyses given in the table w r ere made 

 under widely varying conditions. The age of the suspensions 

 varied from a few minutes to several weeks. In order to 

 avoid error of measurement, new standard suspensions were 

 prepared from time to time, and their concentrations are shown 

 in the last column of the table. 



The analyses covered a wide range in amounts of gold and 

 were conducted under varying conditions of light. The 

 results are seen to run with fair regularity and the errors are 

 approximately of the same order of magnitude. 



The Gallenkamp colorimeter, used in the foregoing experi- 

 ments, is not only expensive but cumbersome and complicated 

 in construction ; moreover, the instrument is not suitable for 

 the correct estimation of the most minute amounts of gold, the 

 color becoming too faint for accurate comparison because of 

 the shortness of the columns of liquid. A modified form of 

 the type of apparatus proposed by Penfield for the colorimetric 

 estimation of titanium and consisting of comparison tubes set 

 vertically in a dark box and illuminated from below, was there- 

 fore adopted. The tubes used in these comparisons had a 



