,340 PHILOSOPHICAL TRANSACTIONS. [aNNO I676. 



The reason of this difference I apprehend was, that in the clearest days the 

 light of the white skies, which dilutes and renders invisible the faintest colours 

 at the ends of the image, is a little diminished in a clear day, and so gives leave 

 to the colours to appear to a greater length ; the sun's light at the same time 

 becoming brisker, and so strengthening the colours, and making the faint ones 

 at the two ends more conspicuous. For I have observed that in days something 

 cloudy, whilst the prism has stood unmoved at the window, the image would 

 grow a little longer or a little shorter, accordingly as the sun was more or less 

 obscured by thin clouds which passed over it ; the image being shortest when 

 the cloud was brightest and the sun's light faintest. Whence it is easy to ap- 

 prehend, that if the light of the clouds could be quite taken away, so that the 

 sun might appear surrounded with darkness, or if the sun's light were much 

 stronger than it is, the colours would still appear to a greater length. 



In all these observations the breadth of the image was just 1 inches. But 

 observing that the sides of the two prisms I used were not exactly plain, but 

 a little convex, the convexity being about so much as that of a double convex- 

 glass of a 16 or 18-foot telescope, I took a third prisjn, whose sides were as 

 much concave as those of the other were convex, and this made the breadth of 

 the image to be 1 inches and a third part of an inch; the angles of this 

 prism, and the lengths of the image made by each of those angles, being those 

 expressed in this table. 



Angles of the prism. 



58° 



594- 

 624. 



Lengths of the image. 

 8-1- inches. 

 9 



104- 



In this case, you see the concave figure of the sides of the prism, by making 

 the rays diverge a little, causes the breadth of the image to be greater in pro- 

 portion to its length than it would be otherwise. And this I thought fit to give 

 you notice of, that Mr. Lucas may examine whether his prism have not this 

 fault. If a prism may be had with sides exactly plain, it may do well to try the 

 experiment with that ; but it is better, if the sides be about so much convex as 

 those of mine are, because the image will thereby become much better defined. 

 For this convexity of the sides gives the same effect as if you should use a prism 

 with sides exactly plain, and between it and the hole in the window shutter, 

 place an object-glass of an 18-foot telescope, to make the round image of the 

 sun appear distinctly defined on the wall when the prism is taken away, and 

 consequently the long image made by the prism to be much more distinctly 

 defined, especially at its straight sides, than it would be otherwise. 



