TRANSACTIONS OP THE SECTIONS. 547 



dicates that the points of no velocity are at irregular intervals 

 from ea,ch other, as exhibited in fig. 5. This would appear by 

 combining two or three of the functions by mechanical construc- 

 tion. It seems probable, then, that this condition is necessary 

 to produce white light, and that the whiteness is more perfect 

 in proportion as the intervals are more irregular. The colours 

 which are neither white nor those of the spectrum, may be con- 

 ceived to correspond to undulations in which there is an ap- 

 proach to regularity by the preponderance of two or more sets 

 of equal intervals. 



Newton asserts that the sun's light is not perfectly white, but 

 has a tincture of yellow. If there be a preponderance of any 

 colour, the preceding theory would lead us to expect it would 

 be of that which corresponds to the ordinate AC, which, as 

 may be judged from Fraunhofer's curve, is situated in the yellow 

 part of the spectrum. (See the figures to Articles 419 and 496 

 of Herschel's Treatise.) For the greater number of the com- 

 ponent portions have I very large, and L very nearly equal to 

 the value of A, corresponding to the ordinate AC*. 



If a part of the spectrum towards the violet end be inter- 

 cepted, and the rest compounded as before, AC will be shifted 

 a httle towards the red end, but D E considerably more so. 

 Thus DE and AC will be brought nearer each other, and the 

 compound, if yellow before, will now be more decidedly yellow. 

 By stopping a still greater portion, these ordinates will approxi- 

 mate still more, till they coincide, and at length DE passes to 

 the other side of AC. In the mean while the resulting colour 

 will pass through orange till it becomes red. 



If the spectrum he progressively stopped, beginning at the 

 other end, the resulting colours will be approximations to 

 those that he towards the violet end. The ordinates AC, D E 

 will never in this case coincide, since the greater portion of 

 the light of the spectrum lies towards the red end. 



If the middle part of the spectrum were stopped, the colout 

 which results by compounding the remainder may not be any 

 m the spectrum, though the two parts of which it is composed, 

 taken separately, give nearly spectrum colours; for by the union 

 of these two parts, the intervals between the points of no ve- 

 locity become more irregular than in either of them, the efl^ect 

 of combination being in general to increase the irregularity. 



AH this agrees very well with what is said in Art. 409 of 

 Herschel's Treatise. " If the violet light be intercepted, the 



* The residuum colour would be different for a different form of the curve. 

 May not the colours of the fixed stars be owing to a difference of this kind' 



2n 2 



