MEMOIR III.] INVISIBLE LINES IN THE SUN'S SPECTRUM. f5 



great change in the relative visibility of Fraunhofer's 

 lines when seen on different occasions. There are times 

 at which the strong lines seen in the red ray are so fee- 

 ble that the eye can barely catch them, and then again 

 they come out as dark as though marked with India-ink 

 on the paper. During these changes the other lines may 

 or may not undergo corresponding variations. The same 

 remark applies equally to the blue and yellow rays. It 

 has seemed to me that the lines in the red are more visi- 

 ble as the sun approaches the horizon, and those at the 

 more refrangible end of the spectrum are plainer in the 

 middle of the day. 



(I subsequently substantiated this remark, and satis- 

 fied myself that many of the lines in the red are due to 

 absorption by the earth's atmosphere, and therefore more 

 distinct with a rising or setting sun. Those in the more 

 refrangible regions, the blue, the indigo, and the violet, 

 are due to absorption by the atmosphere of the sun.) 



A sunbeam, passing horizontally from a heliostat mir- 

 ror into a dark room, was received on a metal plate with 

 a slit in its centre, the slit being formed by a pair of 

 parallel knife edges, one of which was movable by a 

 micrometer screw, the instrument being, in fact, the com- 

 mon one used for showing diffracted fringes. The screw 

 was adjusted so as to give an aperture -% inch wide, and 

 the light passing through fell upon an equiangular flint- 

 glass prism placed at a distance of eleven feet. Imme- 

 diately on the posterior face of the prism the ray was 

 received on an achromatic lens, the object-glass of a tele- 

 scope, and brought to a focus at the distance of six feet 

 six inches, at which an arrangement was adjusted for ex- 

 posing white paper screens, on which the greater fixed 

 lines might be seen, or sensitive plates substituted for 

 the screens, occupying precisely the same position. The 

 lines on the screens could, therefore, be compared with 



