ACCORDING TO THE UNDULATORY THEORY. 493 



imitate with two pieces of mica thephaenomena arising from two retarda- 

 tions, &c. To perform this it is only necessary that the light which 

 is reflected from one plate of mica on the prism must first have been re- 

 flected from the one plate of mica on the other plate. According to what 

 has already been proved, a spectrum is then obtained which contains 

 all the absorptions which have been caused by each single retardation. 



The following is, as I have found, the easiest method of performino- 

 this experiment, which indeed, properly speaking, explains nothing, but 

 which deserves to be mentioned as a beautiful experiment : I generally 

 take a plate of mica, whose surfaces, besides being even and without 

 faults, incline one towards the other, so that the plate is thicker at one 

 border than at the other. Among the plates of mica which I have ex- 

 amined in this respect, I have found one which possesses these properties 

 in a high degree. As I bent this plate into the form of the surface of 

 a vertical cylinder, and placed it so that the light was reflected on the 

 prism from its thick end, I obtained a perfectly regular prism, with 

 about 120 black stripes; but the surface of the cylinder being turned 

 round its axis, so that the reflecting element gradually advanced 

 toward the other end, the distance between the stripes gradually in- 

 creased, while at the same time their number diminished, till at last 

 from the thin end I received but a few more than 20. In order to 

 bring the surface of the cylinder to any desired position, I fastened it 

 on to a small cylindrical pillar, A B (fig. 7), which was fixed by wax, 

 or any other glutinous substance, to an even support. 



In order to produce spectra which contain two series of absorptions, 

 I placed two such cylinder surfaces in the manner shown in A and B 

 (fig. 8). The light from the lamp C, concentrated by means of a 

 great convex lens D, is carried to the first cylinder surface A ; 

 from this it is thrown on to the second, B, and from this further on to 

 the prism E. The light of the flame is hindered from falling on the 

 surface B, by means of a sliding screen F ; and by another similar 

 screen G that light is received, which otherwise might easily be re- 

 flected from the surface A to the prism. By turning both cylinder 

 surfaces round their axes, I can give to the two retardations any de- 

 sired relation to one another ; and in this manner, as just described, I 

 can change ad infinitum the phasnomena of absorption. 



Very small retardations, such for instance as are smaller than a wave- 

 length, cannot, according to this method, be accomplished, because it 

 is almost impossible to give to the mica the necessary degree of thin- 

 ness. But in order to produce also the phaenomena in which small re- 

 tardations are presupposed, I use coloureil fluids whicii are inclosed in 

 a cylindrical tube, between two [)lates of glass whose distance from one 

 another can be altered at pleasure. I have com[)let('iy imitated, with a 

 red absorbing fluicl and a plate of mica, not only tiic s|Kotruni of iodic 



