REVERSED AND NON-REVERSED SPECTRA. 79 



be moved, whereas the others, m' and n', are ends of the same rigid plate. 

 Gratings of different constants may advantageously contribute to this end. 

 Beyond this, the paths mn' and nm' and mm' and nn' may be increased to 

 any length, either directly or by multiple reflections from a special system. 

 Many other modifications are suggested. If white light is used, the phe- 

 nomenon is confined to a narrow strip of spectrum and the fringes must be 

 horizontal. 



As I did not have two ruled transmitting gratings and as film gratings 

 seemed unpromising for work of this kind, the method of figure 57 represents 

 a simple disposition of reflecting gratings, of which several were available. 

 The ruled faces of the gratings G' and G face away from each other. 



58 



The former receives the collimated pencil of homogeneous light, L, and 

 after diffraction the partial beams pass to the pair of opaque mirrors m and 

 n (symmetrically placed), and thence by reflection to a similar pair of mirrors, 

 M and N. From here the pencils reach the second grating, G', where each 

 is again diffracted into the common ray G'T, entering the telescope T. The 

 grating G' may be concave with the lens at T beyond the principal focus. 

 If the mirrors M and N are symmetrically rotated, the parallel component 

 pencils Nn and Mm may be replaced by the pencils Mn and Nm, crossed at 

 any angle. Homogeneous light is preferable. Simultaneously the rays are 

 exchanged. The pencils, Mm, etc., may be of any length, and in general the 

 remarks in the preceding paragraph apply. 



A more flexible design also suggests itself, with four fixed mirrors, m, n, 

 m', n', four movable mirrors, M, N, M', N', rotating symmetrically around 

 vertical axes parallel to the faces of the gratings G and G', these being parallel 

 to each other, as in figure 57. On rotating M, N, M', N', the rays may be 

 exchanged. Here M . . . . N' should be a near system, m . . . . n' a fixed 

 and far system of mirrors. Other methods will presently be described. 



35. Experiments. Reflecting gratings. Parallel rays. The experiments 

 were begun with the apparatus as in figure 57, G being a Michelson grating 

 and G' a Rowland grating, each with somewhat less than 15,000 lines to the 

 inch. The distance of G from the mirrors m and n was about 22 cm., of G 

 from G' about 60 cm., and of G' to the focal point just ahead of the lens (or 

 the line of mirrors M and N) about 90 cm. The latter were about 50 cm. 

 apart. In the absence of sunlight, the arc lamp was used, and the fringes for 

 reversed spectra were found without great difficulty. It was also easy to 



