M. H. Helmboltz on the Theory of Compowid Colours. 527 



from these colours a clear pure white. If the investigation were 

 conducted with instruments more complete than those applied 

 by me, and which would permit of the formation of a larger field 

 of compound colours, the limits of those raj^s which produce 

 white would probably be estimated with greater accuracy, inas- 

 much as the comparison of the hues of large surfaces is capable 

 of being effected with much more ease and sharpness. 



By the rays which produce white, the whole width of the spec- 

 trum is divided into three sections. The first of these corre- 

 sponds to the red, and, if we compare the ratio of the luminous 

 vibrations with those of sonorous waves, answers to about the in- 

 terval of a small third, the middle green section to a great third, the 

 third and violet section being somewhat smaller than a small 

 third. Colours of the fii'st and second sections combine to tones 

 of yellow, with transitions into red, tlesh-colour, white, and green; 

 those of the second and third combine to blue, with transitions 

 into green, white, and violet ; those of the first and third com- 

 bine to purple-red, with transitions into flesh-colours, rose, and 

 violet. 



With respect to the combination of three simple colours, we 

 may conclude that white can only be the result when rays from 

 the three different sections of the spectrum are suitably united. 

 At least it cannot be supposed, although all possible combi- 

 nations cannot, of course, be exhausted by experiment, that 

 the yellow or yellowish colours, for example, which are derived 

 from the red and green sections, can, by the addition of one or 

 more colours contained in these sections, red, yellow, or green, 

 pass o\er into white. This remark is also applicable to the mix- 

 tures of the green and violet, as also to those of the red and 

 violet sections. We may, on the contrary, succeed in obtaining 

 white from various combinations of three colours taken simulta- 

 neously from the thi-ee sections. Tor this purpose I have made 

 use of a black screen with three slits. Two of these were parallel 

 and inclined at an angle of 45° to the horizon ; they stood at 

 such a distance from each other, that, when observed through 

 the prism at the ordinary distance, the violet of the one fell upon 

 the red of the other. The slit from which the violet is obtained 

 must be made about twice as w ide as the other, for otherwise the 

 violet is too feeble in comjjarisou with the red. A third slit, 

 which is to yield green for the mixture, was cut between the two 

 others and at right angles to them, so that the three slits 

 together presented a figure similar to a Z. The spectrum of the 

 third slit intersects ut right angles the jjvu'ijIc stri])e given by the 

 two others, and generates a series of niixeil colours in which tlie 

 whitest jjortion is easily sought out. Ey turning the prism 

 round the axis of the telescope, the mixed colours can be so 



