298 THE AMERICAN MONTHLY [Oct 



green, followed by the yellow, whereas the red is the least 

 bent of all. This difference in bending of the rays, it is 

 well-known, is entirely dne to the difference in the re- 

 spective wave-lengths of the individual colors. If we 

 pursue our experiments a little further, we shall find that 

 the more dense the glass becomes of which the prism is 

 made, so much the more usually are the colors bent as well 

 as separated out, such separation being technically known 

 as ( *'dispersion." But this is not all. When comparing 

 the dispersion produced--by different kinds of glass, that 

 is, glass in the manufacture of which different minerals 

 are used, we notice another peculiarity and it is this, that 

 although two glasses may not perhaps differ so very 

 largely in density, yet prisms made with them do not pro- 

 duce the same kind of spectra. One glass, for example, 

 will spread out the violet end and yet bunch together the 

 red and yellow, whereas another will largely spread out 

 the red and the yellow and afford but little dispersion of 

 the colors in the violet end of the spectrum. Such pecu- 

 liarity is rightly called the "irrationality of the spect^ 

 rum," and the colors introduced from this cause in an im- 

 age are said to result from the non-elimination of the 

 "secondary spectrum." 



Let us now consider for a moment what all this has to 

 do with our subject ? To fix our ideas we must first bear 

 in mind that an image of the illuminant is formed by each 

 color, so that if the sun had illuminated the prism in our 

 previous experiment, with suitable arrangements we 

 should have seen an image of that luminary in each color. 

 Then, secondly, as a simple lens is nothing but a congeries 

 of approximated prisms varying only in size and shape, 

 so it is not difficult to understand that along its axis there 

 would have been represented a consecutive series of sun- 

 images in all the colors of the rainbow, not one of which 

 would have lain in the same plane as any other. 



In theoretical achromatism then for color, whatthp op- 



