266 0. N. Rood— Color System. 



In some observations it was found advisable, for the sake of 

 using high coefficients, to employ paper which was slightly 

 fugitive, but by properly arranging the process the evil effects 

 of this were obviated. 



The coefficients of two strictly complementary discs are as 

 their distances from the center or from white, and this distance 

 is of course measured along the two radii that make up the 

 diameter, but in the method used by me for colors not strictly 

 complementary, the cosine of a small angle is necessarily 

 employed and treated as if it were a radius. The angles used 

 however were quite small, from 1° # 5 to 3°, and consequently 

 the error in any individual case was insignificant, but it is 

 cumulative and finally becomes comparable with those of obser- 

 vation. After the uncorrected results have been plotted in a 

 Newton's diagram, the angular distances of the colors will be 

 known, very nearly, and the small errors from this geometrical 

 source can be corrected and a fresh diagram constructed. 



As above stated, it was not found advisable to employ angles 

 larger than about 3°, since with larger angles the results were 

 more uncertain, while with the angle above indicated the 

 manipulation was hardly more difficult than in the case of truly 

 complementary discs. 



Mode of ascertaining the most neutral mixture. 



Let us suppose that we wish to obtain the most neutral mix- 

 ture that can be furnished by yellow and blue discs that are 

 not complementary ; the best method of procedure according 

 to my experience is the following : a pair of large blue and 

 yellow discs are combined as shown in figure 1, and om the 

 same axis comes a black and white disc destined to furnish the 

 comparison grays, and again on this, smaller yellow and blue 

 discs cut from the same papers. Of course these three com- 

 pound discs can be varied independently of each other. We 

 start now with an amount of yellow on the 

 larger disc which is far too small for ap- 

 proximate neutralization, and on the 

 smaller disc with still much less yellow, and 

 call the difference between the sectors the 

 interval. This interval in any given case 

 is always to be kept roughly constant, and 

 the more nearly complementary the discs 

 are, the smaller it may be made. There is 

 however no gain whatever in trying unduly 

 to diminish it, and such attempts tend only to defeat their 

 object, it being far better and safer to have the interval too 

 large than too small. The arrangement in figure 1 indicates 



