448 G. K. Gilbert — Congress of Geologists. 



else has undertaken to use the spectrum colors more than once, 

 under the impression that they do not afford the necessary 

 range or variety. This impression is based largely upon the 

 popular meaning of the word color. It is indeed true that if 

 we limit ourselves to those parts of the spectral series which 

 have univocal names, we have only six or seven distinctions; 

 and it is further true that if we have recourse to binomial 

 designation?, such as yellowish green and greenish yellow, we 

 obtain rather indefinite conceptions; but to men of science 

 there are better resources than those afforded by the language 

 of every day life. The spectrum has been elaborately studied, 

 and the relations of its dark lines to its colors have been de- 

 termined. Its wave lengths have, moreover, been measured, 

 and by such means as these we are furnished with three differ- 

 ent scales, any one of which is adequate to the precise defini- 

 tion of any hue of the continuous series. What needs to be 

 done is this. When the divisions of the time scale have been 

 decided on, the spectrum must be studied to ascertain the best 

 selection of hues. Their number must, of course, be that of 

 the number of divisions of the time scale, and they must be so 

 chosen that the degree of separateness of adjacent colors shall 

 be everywhere the same, as judged by the normal human eye. 

 Then define each hue by its wave length, or its position in the 

 Kirchhoff scale, and define it also in terms of the best combi- 

 nation of pigments with which it can be approximately repro- 

 duced for practical use. It is of course impossible to copy the 

 prismatic colors with accuracy, because the colors of pigments 

 are impure, but this difficulty will not seriously interfere with 

 the employment of the prismatic colors as a standard. 



The practical question whether the spectrum will give a suffi- 

 cient number of hues so far separated from each other as to be 

 distinguishable in all the arrangements occurring on maps has 

 received such consideration as I have been able to give it, and 

 it is my judgment that the maximum number of hues that can 

 safely be used .falls somewhere between fifteen and twenty. 

 There will certainly be no difficulty in thus constructing a 

 standard color scale with about a dozen terms. 



The employment of the spectral colors in this manner leaves 

 three groups of colors unassigned, the purples, the browns and 

 the grays. If the spectral colors be arranged on the circumfer- 

 ence of a circle so that each diameter of the circle connects 

 hues that are complementary, it is found that they occupy the 

 greater part, but not quite all, of the circumference, and the 

 color needed to fill the vacant arc is purple. The hues of pur- 

 ple might then, if deemed necessary, be added to one end or the 

 other of the spectrum, thus increasing the range from which to 

 select colors for the time scale. 



