108 DISPLACEMENT INTERFEROMETRY BY 



Figures 104 and 106 show that the interfering rays are non-symmetric; 

 one of them, LMNm'NT is confined to but one side, the other, LMM'N'mN'T 

 takes in the other three sides of the rectangle. Apart from the glass-paths, 

 the latter ray thus has an optic path in excess of the other equal to twice the 

 breadth of the rectangle. Hence the mirror m' is to be moved to m i, behind 

 m' by the equivalent of the breadth in question. This makes it more difficult 

 to locate m'\ but for some purposes (for instance, if a bulky apparatus, placed 

 between m and m', is to be traversed by one ray only) it is a decided advantage. 

 In the case of figure 106, the center of spectrum ellipses may be brought into 

 the field or the achromatic fringes rotated at pleasure, by rotating MN' on a 

 horizontal axis. In most cases L and T may be exchanged at pleasure. The 

 glass reflections from N and N' (non-silvered sides) may usually be blotted 

 out with small screens on m and m'. It is more difficult to get rid of the glass 

 reflection from M, but it is usually sufficiently one-sided not to be much of an 

 annoyance, if ordinary plate glass is used. As a rule, perfect achromatics 

 are more difficult to obtain by the present method. They are liable to be not 

 quite symmetric, greenish on one side and reddish on the other. The com- 

 pensator adjustment, if it can be used, avoids much of this. 



Another method of using this interferometer, frequently convenient, is 

 shown in figure 107. Here the auxiliary mirrors m, m', are coplanar, or nearly 

 so. To elongate the short ray, LNm'NMT, a double V-compensator n, n', 

 is inserted, each consisting of two mirrors at right angles to each other, with 

 corresponding parts parallel. These mirrors are therefore to be parallel or 

 normal to the set M, M', etc. They are conveniently made by silvering a pair 

 of Fizeau bi-plates, inside and outside, and the surfaces must be brilliantly 

 polished. The V-mirror n is stationary, whereas n' is on a micrometer-screw, 

 actuating it in the direction s, preferably parallel to the b- side of the ray 

 rectangle. Each ray-path between n and n' is obviously equal to b. The 

 adjustment for parallelism of n and n' need not be very rigorously made ; but 

 in proportion as it is so the direction us may be slightly inclined on any side 

 without destroying the fringes, for the rays enter and issue from the system 

 n, n', in parallel. The system M, M', N, N', m, m' , may be first adjusted for 

 parallelism and coincidence without n, n' . The latter may then be inserted 

 and the adjustment repeated by its own leveling-screws at n and n'. The 

 supernumerary white images (glass side reflections) may be removed at m 

 and m' and also (partially) at M, by small paper screens covering them. The 

 mirrors M, N, N' (half-silvers) must be of equal thickness, as above, and set 

 with their faces in the directions shown by the dotted lines in figure 107. In 

 this case each component ray traverses the identical glass path four times, 

 and the achromatic fringes are obtained in a degree of perfection depending 

 on the equality of the glass- paths. To find the achromatics, it is as usual 

 necessary to find the spectrum fringes first, and to move the micrometer n's 

 until these are horizontal. Fringes are enlarged or rotated by turning M and 

 M', slightly, on horizontal or vertical axes. Again, by rotating n n' (as a 

 whole) on a vertical or a horizontal axis, fringes may be rapidly rotated or 



