On the decomposition of Light. 21 1 



In the second place I drew Fig. 2. This consists of three 

 curves nearly circular and alike, described round the dial itt 

 the follovving niannt^r. I first described three equal circles, 

 hav ng their centres in the radii drawn through thtr divisions 

 of 60, 180, and 300 degrees; and the circumferences of 

 which were targents to the dial at the divisions of 240, 3u0 

 and 120 degrees respectively. I then modified each circum- 

 ference by this law, that, on prolonging the diameters of 

 the dial in every possible direction, the sum of the prolon- 

 gations of every diameter to the new curve should be a con- 

 stant quantity. It is easy to understand this second con- 

 struction, by which it will appear, that the resulting curve 

 differs in fact little from tlie circular circumference. 



Third'v, I conceived, that all the prolongations of the 

 radii of the dial to the red curve represented each a pro- 

 portional quantity of my red matter meii'iioned in the {jam- 

 graph before the precedins: ; so that this dial is surrounded 

 by a red crescent to a certain point, \\ hence it decreases ac- 

 cording to a given law. We must likewise admit a green 

 envelop, analogous to the preceding, and limited by the 

 curve of that colour J and lastly a violet envelop, within the 

 third curve. 



This supposed, if for each pDint of the dial we make a 

 mixture of colours corresponding to that point, we shall 

 have a series of tints in imperceptible gradation from one to 

 the other; which in tone, place, and every other respect, 

 will be extremelv analogous to the colours of the dial, that 

 I had previously traced conformably to the ideas of Newttm, 

 and are such, that the union of two diametrically opposite 

 to each other, will every where form a white identically the 

 same. 



This is a reiult which I offer a^ a further probability 

 greatly in favour of my hypothesis of three colours. 



It is true, the dial constructed by the first method difft-TS 

 a little from that by the last, as in this the purest red is 

 somewhat nearer the plate of the oi'ange, and the violet 

 nearer that of the indiffo. B-it, beside that this difTi'rcnce 

 is little in itself, it is sup|)orted by experience; for the rc- 

 lalion of colours in general, and the progress of their ab- 

 O 2 sorption. 



