93 



scale is in Aongstrom unites. It will be noted in the last of 

 the three figures that the width of the fainter companion is about 

 one one-hundredth of an A. U. The limit of resolution of the 

 instrument is about half as much, or its resolving power is over 

 a million. (Figs. 7, 8, 9, Plates I and II. )t 



It will be observed that the echelon spectra are repeated, thus, 

 ai and aj are two successive spectra of the same line. This 

 is true of any grating spectrum, and the difficulties which arise 

 from the overlapping of the successive orders of spectrum may 

 be overcome by separating these by a prism whose refracting 

 edges are perpendicular to the lines of the grating. The same is 

 true of the echelon spectrum — save that the order of the over- 

 lapping spectra is so high that a prism is hardly adequate, and 

 recourse must be had to a grating with its plane of diffraction 

 perpendicular to that of the echelon. 



With this arrangement it is possible to photograph a large 

 part of the spectrum at once.* 



A photograph of the iron spectrum may be taken so that it 

 may be noted that this combination of grating and echelon 

 makes it possible to observe absorption spectra as well as bright 

 line spectra. 



A photograph of the solar spectrum may be so taken as to 

 show that the spectral "lines" are generally too broad to justify 

 the use of so great a resolving power. 



Finally it may be pointed out that this combination gives us 

 the means of comparing the wave lengths of spectral lines with 

 a degree of accuracy far superior to that of the grating. 



+For these illustrations see Michelson's paper as originally published. 

 *If the preliminary analysis has been made before the light entered the slit of echelon 

 spectroscope, it would be possible to examine but one— at most a few— lines at a time. 



