128 Dr BREWSTEII on a New Analysis of Solar Light. 



or what is the same thing, the light at x will be orange, rendered 

 brighter by a mixture of white light. The two blue rays, there- 

 fore, which enter into the composition of the light at a?, will not 

 communicate any blue tinge to the prevailing colour. 



If the point x is taken nearer M, and if, at that point, the 

 blue rays are more numerous in proportion to the yellow than 

 2 to 5, that is, if they are as 3 to 5, then there will be 1 blue 

 ray more than what is necessary to make white light with the 

 2 yellow and the 3 red rays, and this blue ray will give a blue 

 tinge to that part of the spectrum, or will modify the peculiar 

 colour of pure red light. In like manner, the blue extremity of 

 the spectrum may have its peculiar colour modified by an excess of 

 red rays so as to convert it into violet light. In this manner the 

 tinge of red light at the blue extremity of the spectrum, and of 

 blue light at the red extremity, may be explained, even if the 

 least refrangible branch BM of the blue curve is every where 

 within the least refrangible branch YM of the yellow curve, and 

 the most refrangible branch RN of the red curve every where 

 within the most refrangible branch YN of the yellow curve. On 

 this supposition the excess of blue light over the yellow will be- 

 gin to modify the red space at that point where the ordinates ex, 

 bx, are in the ratio of 2 to 5 ; the ratio in which they exist in 

 white light, and the excess of red light over the yellow will be- 

 gin to modify the blue space at that point where the ordinates of 

 the most refrangible red and most refrangible yellow branch are 

 as 3 to 5, the ratio in which the corresponding rays exist in white 

 light. But it is not improbable that the blue branch BM may 

 actually cross the yellow branch YM at some point m, as shewn 

 in Fig 5 ; and the red branch RN the yellow branch YN, so 

 that the blue ordinates in the one case, and the red ordinates 

 in the other, will exceed the yellow ordinates at every point 

 beyond the points of intersection m and n. If this should prove 

 to be true, it follows, that, at and beyond n, the red should, as it 



