10(5 Banis — Use of the Grating in Interferometry . 



the micrometer screw clockwise, tliey move from red to violet 

 and vice versa. 



When the distance GM is longer and G' N shorter, rings 

 also occur ; but only a single set was found. The spectrum 

 lines not being clear, finding them was a matter of discrimi- 

 nation. AVhen the micrometer screw is turned clockwise, how- 

 ever, they marched from violet to red, i. e., in opposite direction 

 to the preceding. 'I'luis the virtual air space M-N, whicli was 

 increased in the preceding case, is liere being decreased. This 

 space is not generally zero. Moreover, violet travels radially 

 more rapidly than red, for the same micrometer displacement. 

 Hence the drift of ellipses agj'ees with the horizontal radial 

 motion of the violet. 



It is an astonishing result that when the grating is reversed 

 (front face put rearward, but leaving the air space unchanged) 

 the ring type is left in the field ; though tlie solitary ring type 

 may change to the multiple type. This is true for case A or 

 case C. It is also true for the eccentric type case B, where 

 line types single or multiple reappear at slightly different mirror 

 angles. Briefly, rotation lias no efl'ect on the march of ellipses 

 through the spectrum ; but in case £ it changes convex lines 

 downward to concave lines, and vice versa. Nor has reversal 

 of grating any effect on the march. The position of the grating 

 with respect to the mirrors (three places being available) alone 

 determines this result, certainly in the opposed cases A and G, 

 and probably in case B. 



The use of a compensator in either of the component rays is 

 always accompanied by the two effects in question ; i. e., there 

 is both relatively large radial motion of the fringes and rela- 

 tively small displacement. Within the field of the telescope 

 one may therefore be evaluated in terms of the other, the two 

 having the stated relation of coarse and fine adjustments. In 

 case JB the fringes will rotate, expand, or contract. 



3. Elementary theory. — The endeavor must now be made 

 to explain these results more in detail. For this purpose the 

 diagrams of fig. 1 may be consulted. 



In the grating used at an incidence of about 45°, the angle 

 of difi'raetion in the first order was about 32° 39' for the 

 D^ sodium line, there being about 2,847 lines to the cm., 

 corresponding to a grating space of 2>="0003512*^™. Thus 

 X/Z> = '1677. The index of refraction was assumed to be 1*53 

 for the same line. 



The diagrams are drawn for an angle of incidence of 45° and 

 for normal reflection from the mirrors M and N. In case of 

 the grating given, the three positions of the grating at whicli 

 interferences occur were about •6'"° apart and are marked 124, 

 100, and 76 scale parts, on the micrometer screw normal to the 



