REVERSED AND NON-REVERSED SPECTRA. 



15 



grating space, and thereafter interfere along the line T, entering the telescope. 

 To treat the case the mirrors M, etc., may be rotated on the axis T normal to 

 G' in the position M\. G' n and G' m show the reflections of G' in the mirrors 



\ 



//^s? 



/kjfa 



\ 



AT and MI. We thus have a case resembling the interferences of thin plates, 

 and if e m is the normal distance apart of the mirrors M\ and N, the displace- 

 ment Ae m per fringe is given by 



X = 2Atf m COS 6/2 



where 5 is the angle between the rays incident and reflected at the mirrors. 

 This is the equation used above. If the mirrors and the reflections of the 

 gratings G' make angles a/ 2 and a with G', the actual lengths of the rays (pro- 

 longed) before meeting to interfere terminate in e and f respectively. Let the 

 image of G' be at a normal distance e apart. Then e ze m cos 0/2 , for the figure 

 fdbs is a parallelogram. If the distance eg is called C we may also write 



X = e cos 8t-\-C sin 62 

 since C = 2e m sin <r/2 and the angle of diffraction fl=( 



7. Compensator measurements. Sharp wedge. With the object of testing 

 the interferometer under a variety of conditions, measurements were made 

 with a number of different compensators and the experience obtained may 

 be briefly given here. The first of these was a very sharp wedge, such as 

 may be obtained from ordinary plate glass. The piece selected, cut from an 

 old mirror, on being calipered, showed the following dimensions: Length, 

 5 cm. ; thickness at ends, 0.375 and 0.367 cm. Hence the angle of the wedge 

 is a = 0.0016 radian, or about o. i. No difficulty is experienced from the devia- 

 tion of the rays for so small an angle, though sometimes the fringes are unequal 

 and the lines presumably curved. This wedge was attached to a Fraunhofer 

 micrometer moving horizontally, and the normality of the rays passing through 



