424 Prof. Stokes's Remarks on Prof. Challis's paper, 



chosen, and the intensity of the light properly regulated. Hence 

 any speculations as to the cause of the variations of intensity in 

 the solar spectrum can have no bearing on the subject before us, 

 seeing that the relation between the intensities of the mixed 

 colours necessary for the production of whiteness is a matter of 

 experimental adjustment. 



The reason why the superposition of two coloured bodies does 

 not give the mixture of the colours is known, and is very simple. 

 The composition of the light transmitted through a coloured 

 glass may very conveniently be represented by a curve, in the 

 manner of Sir John Herschel, in which the abscissa x denotes 

 refrangibility, measured, suppose, by the distance from the ex- 

 treme red in some standard spectrum, and the ordinate y denotes 

 the intensity ; so that ydx is the quantity of light between the 

 refrangibilities x and x + dx, the intensity in the incident light 

 being taken equal to unity, for simplicity's sake, whatever be the 

 value of x, as we only care to compare intensities for the same 

 value of X, Let y, y' be the ordinates in the curves belonging 

 to two glasses, y^ the ordinate belonging to the tint obtained by 

 superposing the glasses, y^ the ordinate belonging to the mixed 

 tint, as procured, for instance, by a double-image prism, in which 

 case each of the superposed differently coloured images has half 

 the brightness of the original. Then y^ = yy', but y^=^{y-^ y'), 

 and it is easy to see how different may be the curves whose ordi- 

 nates are y^, y^ respectively. Thus, let the scale of abscissae be 

 such that the spectrum extends from x=0 to x-^ir, and let 

 j/ = ^(l — cosa?)^, y==^(l4-cosa?)'^. In this case ?/^=y\rsin'*i2?, 

 which vanishes at the extremities, and is a maximum in the mid- 

 dle ; whereas y^=\[\ -f cos^ x), which is a maximum at the two 

 extremities, and a minimum in the middle. In the former case, 

 the tint would be a sort of green, a pretty full colour; in the latter, 

 a sort of dilute purple. The colours of two ribbons may very con- 

 veniently be mixed in equal proportion by placing them side by 

 side, and viewing them through a double-image achromatic prism; 

 and it will be seen how different the mixed colour is from that 

 seen on superposing the ribbons and holding them up to the 

 light. 



I cannot agree with Professor Challis, that ^' the coloured light 

 of substances, though derived from sunlight, is in fact new light," 

 except so far as relates to that portion which arises from fluor- 

 escence. But fluorescence is often absent altogether; and even 

 when it exists, the colour thence arising must in most cases be 

 but a small fraction of the whole colour observed when the sub- 

 stance is freely exposed to white light, not viewed under absorb- 

 ing media. I think that any one who has been in the constant 

 habit of analysing by the prism the light transmitted through 



