ON LIGHT AND COLOURS. 147 



eye. We may thus measure them with a micrometer ; but no 

 such nicety is required, because their increase in breadth is 

 manifest. The only doubt is with respect to their relative 

 breadth when the edges are not very near and just when they 

 begin to form fringes. Sometimes it should seem that these 

 very narrow fringes decrease instead of increasing. How- 

 ever, it is not probable that this should be found true, at 

 least when care is taken to place the two edges exactly 

 opposite each other; because if it were true that at this 

 greater distance of A from B (fig. 17) they decreased, then 

 there must be a minimum value of P M between C and B, and 

 between C and A ; and consequently the law of flexion must 

 vary in the different distances of A and B from the rays P, a 

 supposition at variance it should seem with the law of con- 

 tinuity. 



Exp. 2. The truth of this proposition is rendered more 

 apparent by exposing the two edges to the rays forming the 

 prismatic spectrum. The increase is thus rendered manifest. 

 If the fringes are received on a ground glass plate, you can 

 perceive twelve or thirteen on each side of the image by the 

 direct rays. It is also worth while to make similar observa- 

 tions on artificial lights, and on the moon's light. The pro- 

 position receives additional support from these. But care 

 must always be taken in such observations, which require 

 the eye to be placed near the edges, that we are not misled 

 by the effect of the small aperture in reversing the action of 

 the edges. Thus when viewing the moon or a candle through 

 the interval of two edges, one being in advance of the other, 

 we have the coloured images (or fringes) cast on the wrong 

 side. But if we are only making the experiment required to 

 illustrate this proposition, the edges being to be kept directly 

 opposite, no confusion can arise. 



It is to be noted that the increase of breadth in the fringes 

 is not very rapid in any of these experiments ; nor are we led 

 by the calculus to expect it. Thus suppose m = 1, we find 



( because y = j at the point C, when x - 0, the breadth 



\ a x J 



L 2 



