VOL. XXIX.J PHILOSOPHICAL TRANSACTIONS. 237 



too oblique, least the incident ray be reflected downwards by it, as the ray Rh 

 is by the prism b thrown to e, in fig. 20. Several have confessed to me that 

 they at first used to fail in this experiment, for want of setting the second prism 

 in a due inclination. 



Though the colours by the second refraction on the ceiling appeared un- 

 changed, when seen by the naked eye, yet if viewed through a prism, they 

 aflforded new colours, except some part of the red, and some part of the violet, 

 which was owing to their not being fully separated ; for which reason I made 

 the following experiment, to prove, that if the colours be well separated, they 

 are truly homogeneal and unchangeable. 



N. B. When the prisms are good, and no clouds are near the sun, the 

 extremity of the red or violet will afford unmixed colours in this experiment ; 

 otherwise not. 



Exper. VIII. — Having made a hole in the window-shutter 2 inches wide 

 (fig. 21), I applied to it a tin plate, which sliding up and down hid all this hole 

 in the wood, and only transmitted a small beam through its own hole h, whose 

 diameter was = -^ inch. This beam, by means of the looking-glass l, placed 

 on the board of the window xw, I reflected horizontally to the other end of 

 the room. But to correct the irregularity of the reflection of the looking-glass, 

 I made use of the frame of paste-board pp, which had a hole in it h, of -^ 

 inch also : and placing it at pp, I suffered some of the reflected beams to pass 

 through it, so as to fall on the lens fe (convex on both sides, and ground to a 

 radius of A\ feet) at the distance of Q feet, so that the image of the hole h was 

 projected to f on the other side of the glass, at the distance of Q feet more. 

 Just behind the lens, which by a screw in the stand s might be raised or let 

 down, so as always to receive the beam along its axis, I placed a prism a (up- 

 right on one of its ends, and easily moveable about its axis, by reason of its 

 wire turning freely in a hole in the solid piece of wood t, which stood on an- 

 other stand behind the lens) as near as I could to the lens ef, so that the image 

 of h, instead of being round, white, and projected to f, was cast sidewise on a 

 white paper stretched on a frame, and appeared coloured, and 30 or 40 times 

 its breadth, as at mn. The colours in this case were very vivid and well 

 separated, only the violet had some pale light darting from its end, on account 

 of some veins in the prism a, and the light not coming directly from the sun, 

 but reflected ; which ought not to have been, if the sun had been low enough 

 to have thrown the rays a good way into the room without the help of a 

 looking-glass. 



To show that the colours in this spectrum were simple and homogeneal 

 lights, I made the following experiments. 



