DISPLACEMENT INTERFEROMETRY. 89 



operating 5'. This is the case when other screws, except those normal to M 

 and M' are used. V s f is conveniently near the observer at the telescope T. 



If the offset or V-compensator is to be avoided, a design (fig. 87) must be 

 used in which all mirrors are parallel. If the half -silvers are equally thick and 

 placed with their reflecting faces as shown in the figure, each ray traverses 

 the glass-path twice and the compensation is complete, seeing that twice the 

 glass-path in N is identical with the glass-path in M and A r/ . The air-paths 

 are the same. Moreover the mirror M (for instance) may be displaced by a 

 micrometer-screw in the direction s (within limits) without changing the 

 adjustment at the cylinder GG' . 



If micrometer facilities are to be dispensed with (and that is permissible 

 in the present experiment), the design shown in figure 88, which is now a 

 modification of Michelson's interferometer, suffices. The white light L from 

 the collimator takes the respective paths dCC'd'b and bd'C'Cd, the plate N 

 being half-silvered and N' an opaque mirror. The telescope or spectro-tele- 

 scope is at T. The glass face at N may be turned either way. 



Such an interferometer is self-adjusting. In the design, figure 88, two 

 reversed spectra will be visible in the telescope, which if superimposed by 

 rotating N or N' on a vertical axis will show the linear phenomenon at once, 

 in any color at pleasure. The fringes may be enlarged by rotating N or N' on 

 a horizontal axis, and they are symmetrically equal in size on the two sides of 

 the adjustment for infinitely large fringes. 



If the achromatics are wanted, a prism must be inserted into the rays b 

 (preferably between N and N', figure 88, with a prism angle selected to counter- 

 act the refraction of the cylinder GG' in the manner indicated in figure 93 of 

 paragraph 68. 



65. Apparatus. Two internal reflections. As the fringes were found with- 

 out much difficulty (68) in case of one internal reflection, it seemed desir- 

 able to determine whether this would still be feasible in the apparently more 

 favorable but also more difficult case of two internal reflections. In figure 89, 

 white light arrives from a collimator at L and strikes an auxiliary mirror in, 

 before reaching the half-silver N. If m is capable of rotating both on a hori- 

 zontal and vertical axis, as well as sliding right and left in the diagram, it 

 greatly facilitates adjustments of angle and location of rays. The two beams 

 bcdef and bfedc reunite at b after passing the glass cylinder G (rotating around 

 the axis a) and are observed by the telescope at T. As the spectra (after refrac- 

 tion at c and /) are reflected three and two times respectively, the fringes of 

 non-reversed spectra will be obtained covering the whole length of spectrum. 

 A glass G of low index of refraction will here usually be preferable. 



In case of a half-silver mirror at a small glancing angle, there are usually 

 two pairs of bright spectrum images, and one fainter pair, apart from very 

 faint ones. One bright and one faint pair carry identical fringes and the spec- 



* Cf. Michelson and Morley: Amer. Journal, XXXI, p. 377, 1886. Also Zeeman, below. 



