INTERFEROMETRY OF SPECTRA. 131 



horizontally and vertically. This is usually more convenient and as a rule 

 adequate. Other methods are given below. 



If the distances ac and bd, ab and cd have previously been made nearly 

 equal and the angles approximately 90, the fringes will usually be found on 

 moving the micrometer-screw normal to Hz. 



As the mirrors are thick glass plates, it is preferable that the half-silvered 

 sides of HI and H 2 be toward L and the half -silvered side of H 3 toward T. 

 In this case each ray passes the plates twice, as indicated in figure 88. With 

 ordinary plate-glass the fringes when found are still apt to be small. They 

 are then to be enlarged and centered, by compensator of clear glass C and C', 

 in the two rays respectively, rotated in opposite directions around a hori- 

 zontal axis until the center of ellipses is in the field of the spectroscope. It 

 may be necessary to actuate the micrometer-screw at d to complete the ad- 

 justment. If m is adjustable on two axes, the compensators C, C' are super- 

 fluous, as will presently appear. 



When the ellipses are centered, the direct-vision spectroscope g removed, 

 and the slit widened or removed, the residual or achromatic fringes appear in 

 sight and are ready for use. These are always strong. The spectrum fringes 

 are apt to be less so, since the parts of the ray L pass through two half- 

 silvered surfaces, H\ H? or HI H 3 , in succession. The spectrum fringes are only 

 sharp when the slit is fine. If the white residual fringes are too dazzling, a 

 single or two half -silvers may be placed before the objective of the telescope 

 with advantage. Two plates with their half -silvered sides in contact and 

 held so by a steel clip are excellent for this purpose, while they are at the same 

 time protected from sulphur corrosion. This, in fact, is the best method of 

 preserving silver mirrors (in pairs) when not in use. 



If a is the fraction of light transmitted and i a reflected, the fraction of 

 the original light L reaching the telescope T will be 2 2 (i a) 2 . This is a 

 maximum if ct = y 2 . Thus the illumination is reduced to }i. 



When the spectra are in coincidence and the fringes sharp, the mirror m 

 may be rotated around a vertical axis at A into some position, m'. In such 

 a case the two spectra will move through the field of the telescope at T, but 

 their coincidence will not be destroyed. The D lines, for instance, will con- 

 tinue to be superposed throughout. Considerable path-difference is, however, 

 introduced in this way, and hence the fringes will march through the spec- 

 trum at an enormously more rapid rate. The following data may be given, 

 where a is the angle of rotation of the mirror m and N the reading of the 

 micrometer at H 3 (screw in the normal dn) necessary to bring the center of 

 ellipses back to the sodium lines. In both cases the centers were out of the field 

 above or below, so that horizontal fringes were made the criterion for adjustment. 

 This method is somewhat rough, but adequate for the present purposes. 



(i) Fine thin fringes. Relatively large glass-path. Distance ab, figure 88, 2R = 2i cm. 

 Thickness of glass plates (half-silvers), e =0.70 cm. 



a =0 0.05 0.20 0.30 0.40 0.50 



o 3 = 23 30 128 162 215 258 



