AS MEASURED BY MEANS OF THE RADIOMICROMETER. 63 



DESCRIPTION OF CELLS USED. 



In all cases where we were dealing with different depths of layer, it was 

 necessary to use cells adjustable in length. A very satisfactory form of cell 

 was devised and used throughout the latter part of this work. It consisted 

 essentially of two brass cylinders telescoping neatly into each other. The 

 external diameter of the outside cylinder was about 2| inches, and the thick- 

 ness of the walls was in every case about 2 mm., which was sufficient to with- 

 stand handling without danger of changing the shape of the cell. Into the 

 ends of each cylinder there was sealed, by means of Wood's metal, a glass 

 plate about 1 mm. thick, made of the very best optical glass. In all cases 

 the glass plates were so nearly parallel as to show interference fringes ; and 

 both cells gave the same deflections, either when empty or filled with the 

 same solution and placed in the path of the light before the radiomicrometer. 



After adjusting the glass ends and fixing them securely by means of Wood's 

 metal, the entire cell was first plated with silver, being taken out of the 

 plating-bath from time to time and polished to a bright surface with the 

 finest crocus paper. On top of this silver coating a heavy plating of gold 

 was deposited. The distance between the glass plates fastened to the ends 

 of the telescoping cylinders, which determined the depth of layer of solution 

 used, was in all cases fixed by gold-plated washers, whose thickness had been 

 accurately measured to 0.001 inch by means of a vernier caliper. 



Before any series of readings was made, the positions of the two cells was 

 so adjusted in the sliding carriage as to give equal deflections, when alter- 

 nately placed in the same position before the radiomicrometer, in that part 

 of the spectrum where neither the solute nor solvent had any absorption; 

 and from time to time throughout the experiment duplicate readings were 

 made on this point to see that the cells had not changed their relative 

 positions. In case any change was noted, a duplicate series of readings was 

 always made. Such readings upon the same cell usually agreed to about 

 one division of the scale, which corresponded to about 1 to 2 per cent, 

 depending upon the throw of the instrument. In the midst of the very 

 intense absorption bands,where the deflections of the instrument were small, 

 reaching zero at many points, the error resulting from any drift in the instru- 

 ment or reading of the scale was greater than the mean error given above. 



In nearly all cases new solutions were made up and the results duplicated, 

 so that the tables and curves below represent a mean of several series of 

 readings. In most cases the agreement was very satisfactory, usually the 

 difference not being over 3 per cent. 



Since any change in the position of the prism was a determining factor in 

 the portion of the spectrum which fell upon the thermo-j unction, and since 

 in the very intense, sharp bands of the neodymium salts any slight change in 

 the position of the prism would make a great difference in the final results, 

 great care had to be exerted in setting the head reading of the spectroscope. 

 Such difficulties were not met with in solutions where the absorption bands 

 were broad and diffuse, as in salts of cobalt, nickel, etc. 



