136 THE INTERFEROMETRY OF 



in figure 94, a. As a group the fringes closely resemble the lemniscates of a 

 binaxial crystal in polarized light. The variation of the color-scheme is probably 

 the same, since with sodium homogeneous light the design is in yellow and 

 black. The pattern is not quite dichroic, but appears so, red-green, blue- 

 yellow combinations with an intermediate violet-yellowish succeeding each 

 other. In polarized light the figures are sharpened as a whole, but there is 

 no discrimination. The pattern gradually vanishes with a wide slit, where- 

 upon the achromatic fringes may be seen when the ocular is restored to the 

 principal focal plane. 



If the white slit images pass through each other (in consequence of the 

 vertical rotation specified) the direction of fringes twice changes sign in rapid 

 succession, and this probably occurs when the white slit images are coin- 

 cident. Barring this inversion, the march is regularly proportional to the 

 rotation. 



With the displacement (AJV) of the mirror on the micrometer-screw nor- 

 mal to its face, the fringes pass through a continuous succession of color- 

 schemes, but soon vanish, for they coincide in adjustment with the centered 

 spectrum fringes. Similarly, if a pair of mirrors (MM' or N N', fig. 93) 

 rotates about a vertical axis as a rigid system, the same continuous change 

 of color-scheme and evanescence is apparent. 



These interferences differ, naturally, from the spectrum interferences; 

 they also differ from the achromatic interferences, which are much finer 

 fringes, partaking of the regular fringe pattern seen with biprisms. They are 

 a separate phenomenon, quite sharp and definite, occurring under like con- 

 ditions of adjustment, but under different conditions of observation (ocular 

 out of focus and fine slit). In the principal focus the two sharp, extremely 

 bright slit images are alone present. They are absolutely identical in struct- 

 ure, however, and their spectra when superposed would interfere symmetri- 

 ally throughout their extent. Under these circumstances the rays intersect- 

 ing in the white slit images also interfere before and behind the principal 

 focal plane of the telescopic images specified, and this interference is not 

 destroyed when the slit images are separated (rotation of opaque mirror 

 about vertical axis), or when the slit images are passed through each other. 

 What is not easily seen, however, is the reason of the occurrence of large, 

 sharp, definite hyperbolic forms instead of the usual Young or Fresnel fringes 

 of two slits or slit images. 



On the Michelson interferometer these fringes, like the achromatic fringes, 

 are extremely faint and can hardly be detected except by putting them in 

 slow motion. The spectrum fringes are equally strong in all cases. It appears, 

 therefore, that the two half-silvers are favorable to evolving both the hyper- 

 bolic and the achromatic sets of fringes. The Michelson design is thus not 

 useful here, nor for the measurement of small angles of rotation by the meth- 

 ods described, as the mirrors would have to be rotated in opposite directions. 



Further work with the complementary fringes on different interferometers 

 of the Jamin type showed that to produce the hyperbolics the fine slit images 



