546 



FOURTH RKPORT— 1834. 



favours the idea of the possibility of an analysis of white light 

 distinct from that afforded by the prism. {Treatise on Light, 

 Art. 516.) The tendency of the preceding theory is to show, 

 that the possibility of decomposing one ray is no ground for 

 thinking that another exactly the same in appearance is also 

 decomposable. 



If ?n were not the same in the two component rays, the com- 

 pound ray would not be so like a homogeneous ray, since the 

 intervals between the points of no velocity would not all be 

 equal. In mixing two simple rays there must consequently be 

 an adjustment of the quantity of light in each to bring out the 

 purest compound. 



Composition of all the Colours of the Spectrum. — Let RQBV 

 (fig. 4.) be the curve line (as determined by the experiments 

 of Fraunhofer,) whose ordinates represent the intensity of 

 light from R, the red end, to V, the violet end of the spectrum. 



Fig. 4. 



Draw AB, dividing the area into two equal parts, R AB, V AB. 

 Divide each of the parts into the same number of small equal 

 portions (w). Let RH and VK be taken in the proportion of 

 the values of X for the extreme red and violet rays, and let the 

 curve line HPK be such that an ordinate PM is as the value of 

 A corresponding to the intensity Q M. Also let D E be an har- 

 monic mean between RH and VK. Then, compounding the 

 small portions two and two, taking one in each of the areas 

 RAB, VAB, beginning with the extreme portions at R and 

 V, and proceeding with the others in succession to B, there will 

 be as many compound rays as there are portions, and each will 

 be expressed by such a function as 



2 m cos I — — V cA sm I ^j^ — V c^U 



in which I is greater in proportion as the combined portions 

 are nearer each other, and L is always intermediate to the 

 values of X corresponding to A C and D E. The sum of all 

 these functions is the function expressing the result of com- 

 pounding the whole spectrum. Nothing can be anticipated 

 respecting the nature of the final expression, except that it in- 



