AND THE THEORY OE LIGHT. 85 



colour will be seen^ for the one half fills up the space 

 occupied by the other during rotation; there is no mix- 

 ture of light and shade. The white does not encroach 

 on the dark space surrounding the card. 



129. Now change the centre of revolution of the card 

 from the middle of the end to some point nearer one of 

 the sides^ say to b, at a distance of one fourth of the end 

 from the side A C. A portion of the larger division equal 

 to the difference between it and the lesser, in this case 

 equal to one half of the end of the card, will in every 

 revolution encroach upon the surrounding dark space, and 

 will consequently produce colour ; or in every revolution 

 the opposite side of the portion b f will fill up the space 

 C e, which is equal to it, and the opposite side of C e will 

 fill up its equal space b f, and hence there will be, so to 

 speak, two beats of light on C e ov b f for every beat on 

 a D, or its equal C H when in revolution, 



130. But if we paint C e black, only the middle portion 

 b f will have double beats, and if we paint the reverse or 

 other side of 6/ black, making b still the centre of revo- 

 lution, the whole card will have the same number of beats. 

 On this principle a great many diagrams were formed. 



131. Continuing to operate with the parallelogram, it 

 may now be made | or jV of an inch in breadth. If we 

 place the edge A G ox D B quite perpendicularly on the 

 centre, and in a line with the axis of rotation, we shall 

 obtain a line of greatly concentrated light. I first at- 

 tempted to obtain red in this way, and often did, but 

 have been more successful in getting a delicate pink, 

 approaching to a rich reddish-purple.* 



* When Bristol board is used it is difficvilt to make it remain perpen- 

 dicular for any time, as tlie centrifugal force almost instantaneously gives it 

 an angular motion. Some other material should be used. I have tried 

 silver, but my instrument is not well adapted for holding it firmly, and I 

 have not experimented much with it. 



For perpendicular experiments, the pai't a of fig. 2, jplate IV., is taken 

 off and 5 is put in its place, h is composed of two plates, with a screw to 

 hold fast the cards used. 



