Apeil 15, 1898.] 



SCIENCE. 



517 



"What physicists need to be reminded of is 

 the fact that Helmholtz's hypothesis is just 

 as uncertain as some of the newer ones. 

 Since the physicist, as such, deals only with 

 the phenomena of color, and not at all with 

 its specific eiiect on the brain, it can really 

 make little or no difference with him what 

 hypothesis, if any, of those now compet- 

 ing for supremacy shall win at last. But 

 physicists cannot be expected wholly to 

 withdraw their interest from subjects essen- 

 tially separate but closely related to physics. 

 It is of definite importance, therefore, that 

 they should have some appreciation of the 

 uncertainties which they may be tempted 

 to treat as long established verities. 



What, then, is a ' piimary color,' or a 

 ' fundamental color sensation ? ' Young 

 seemed to think that a primary color is one 

 of the minimum number whose mixture 

 as lights produces white. This definition 

 can hardly be accepted to-day. Physically 

 there is no reason why any hue of given 

 wave-length should be named primary in 

 preference in some neighboring hue whose 

 wave-length is slightly greater or less. 

 The three primaries assumed by Maxwell 

 were red, green and blue, the selection of 

 wave-length for each standard being not 

 definitely fixed. With the use of appro- 

 priate colored glasses for absorption in 

 front of an electric lantern, the production 

 of white on a screen by mixture of these 

 three hues is easy enough. By the same 

 method it is equally easy to produce white 

 by the mixture of yellow and blue, or with 

 any other pair of complementaries, such as 

 red and the mixture of green and blue, 

 which has been called peacock. If we take 

 X = 0.58 /-i for the yellow, and X = 0.47/^ for 

 the blue, and thus succeed in obtaining 

 white, the components of this, or of any 

 other pair of complementaries, may be thus 

 called primaries. Or, if we mix peacock, 

 purple and yellow, which are the comple- 

 mentaries of red, green and blue, respect- 



ively, the white attained is quite satisfac- 

 tory. If a triplet of colors be deemed 

 necessary, therefore, peacock, purple and 

 yellow may be called the primaries, though 

 it might be harder to designate the purple 

 by any single wave-length. It is thus quite 

 indefinite to speak of a primary as one of a 

 minimum number of hues whose mixture 

 produces white. 



A primary has been otherwise defined as 

 a hue which is incapable of being produced 

 by the mixture of any other two hues. 

 Red is thus called a primary, while yellow 

 is distinctly not such. But the yellow due 

 to a mixture of red and green is alwaj's de- 

 ficient in purity, and a similar comment 

 may be made upon the result of any color 

 mixture. If violet be called a primary, as 

 one of Young's triplet, it may be replied 

 that by suitable mixture of red and blue a 

 violet may be obtained that is quite as good 

 of its kind as the yellow obtained by mix- 

 ing red and green. If blue be called a 

 primary, as one of Maxwell's triplet, it may 

 be replied that by suitable mixture of pea- 

 cock and violet a good blue may be pro- 

 duced. If green be called a primary it 

 may be produced, though with considerable 

 admixture of gray, by mixture of peacock 

 and yellowish green. It thus appears that 

 red is about the only hue to which this de- 

 finition seems to be fully applicable. Ad- 

 mitting it, therefore, as a primary, the selec- 

 tion of its companion primaries is still 

 uncertain. 



Whatever may be the definition finally 

 agreed upon for ' primary color,' the corre- 

 sponding sensation is the 'fundamental' sen- 

 sation. A very large amount of time and 

 labor has been spent in the effort to obtain 

 curves that shall correctly represent these 

 fundamental sensations. The curves as esti- 

 mated by Helmholtz are shown in Fig. 1, and 

 have long been familiar. In the figure they 

 are adjusted, not to the prismatic, but to the 

 normal spectrum. It is seen that the maxi- 



