172 THEORY OK LIGHT AND COLOURS. 



to the margin ; fo that the blues will appear in the light near- 

 eft the fhadow. The effect will be increafed and modified, 

 when the refleaed light falls within the influence of the oppo- 

 fite edge, fo as to interfere with the light fimply infleaed by 

 that alfo. 

 More particular But, in order to examine the confequences more minutely, 



examination of ., -,, , • , . r r .1 • n ~. r , , 



the confe- d wl " be convenient to fuppofe the mfteaion caufed by an 



quences. ethereal atmofphere, of a denfity varying as a given power of 



the diftance from a centre, as in the eighth propofition of the 

 laft Bakerian Leaure. (Phil. Tranf. for 1801, p. 83.) Put- 

 ting r = 3, and x = f , I have conftruaed a diagram, (Fig. 4,) 

 which fhows, by the two pair of curves, the relative polition 

 of the refleaed and unrefleaed portions of any one undulation 

 at two fucceffive times, and alfo, by fhaded lines drawn acrofs, 

 the parts where the intervals of retardation are in arithmetical 

 progreffion, and where fimilar colours will be exhibited at 

 different diftances from the infleaing fubftance. The refult 

 fully agrees with the obfervations of Newton's third book, and 

 with thofe of later writers. But I do not confider it as quite 

 certain, until further experiments have been made on the in- 

 fleaing power of different fubftances, that Dr. Hooke's ex- 

 planation of infleaion, by the tendency of light to diverge, 

 may not have fome pretentions to truth. I am forry to be 

 obliged to recall here the afiTent which, at firft fight, I was 

 induced to give to a fuppofed improvement of a late author. 

 (Phil. Tranf. for 1800, p. 128.) 



Scholium. In the conftruaion of the diagram, it becomes 



necefTary to find the time fpent by each ray in its paflage. 



i 

 Since the velocity was denoted by x r , on the fuppofition 



i 

 of a projeaile, it will be as x r. on the contrary fuppofition, 

 (Phil. Tranf. for 1801, p. 27. SchoL 2. Prop. I.) and the 



x 

 fluxion of the difiance defcribed being /-■ , that of the 



time will be / * or =- , of which the flu- 



r s 



ent is y/ i— - yn. Therefore, with the radius x 



defcribe 



