126 BARUS— ELLIPTIC INTERFERENCE [April 21, 



gratings Gm and G„, placed symmetrically with respect to M at an 

 angle / to the E and L directions. The undeviated rays pass off 

 eccentrically at R and are not seen in the telescope at E. They may, 

 however, be seen in an auxiliary telescope pointed in the line R and 

 they then facilitate the adjustments. Rays diffracted at the angle 

 2i, however, are respectively transmitted and reflected by ^1/ and 

 interfere in the telescope in the line E. Similarly rays diffracted at 

 an angle 6' > / interfere in the line D. 



To make the adjustment it is sufficient to bring the Fraunhofer 

 lines in the two spectra seen at E into complete coincidence, hori- 

 zontally and vertically. Coincidence of slit images at R (at least 

 vertically) aids in the same result. It is also necessary that the 

 rulings on Gm and G„ and the slit should be parallel, or that the 

 images of slit and spectra shall lie between the same horizontals. 

 One of the gratings, C,,, may now be moved parallel to itself by the 

 micrometer screw until the elliptic interferences appear. If the 

 plate M is not half silvered there are three groups of these as 

 described in the preceding paper. Each group passes from the initial 

 degree of extreme fineness, through maximum size, to a final degree, 

 for a play of the screw of about i mm. There is the usual radial 

 motion of the fringes, together with the drift through the spectrum 

 as a whole. To bring out the set of solitary ellipses, the silvered 

 surface of M should be towards the light and remote from the eye. 

 As a rule the adjustment is difficult, as an extra condition is imposed 

 in the parallelism of the slit and the rulings of the gratings. The 

 ellipses are liable to be coarse with their axes oljlique. clearer in some 

 parts of the spectrum than in others, unless means are provided for 

 placing the rulings accurately parallel. Even when well adjusted 

 they are rather polygonal than rounded in their contours. They are 

 about as strong with non-silvered glass M as with half-silvered glass ; 

 but in view of the multij^le spectra the adjustment is much more 

 difficult in the former case. 



It has been suggested that the white slit images must appear 

 eccentrically in the direction R. Hence if a special telescope is 

 directed in this line, the final adjustment is reached on coincidence 

 of the proper slit images, provided the rulings of the gratings and the 

 slit are parallel. 



