44 THE INTERFEROMETRY OF 



earlier work with crossed rays and two reflections (I.e.}, experiments may be 

 made with homogeneous light. Accordingly, the sodium arc with a wide 

 slit was installed and the fringes found without difficulty. Strands of fringes 

 with nodules were obtained as before. These rotated in marked degree ( 180) 

 from vertical hair-lines, through coarse vertical strands with horizontal 

 nodules, back to vertical hair-lines again, as either M or G' was suitably 

 displaced normally to its plane. To shift the fringes of any form into the 

 middle of the wide-slit image, a glass compensator in either b or b f may be 

 resorted to, or both M and G' may be displaced together. Again, whereas 

 the micrometric displacement of M produces a marked displacement of fringes 

 within the strip in accordance with equation (i), the micrometric displace- 

 ment of G' leaves the fringes stationary within the strip. While the strands 

 and nodules were strong, the reticulation of fringes could not be clearly made 

 out, in view of the use of film gratings in place of the ruled gratings used in 

 the earlier report. 



In equation (4), D=i6gXio~ cm., if dX/X = 6Xio- 8 /6Xio- 5 = icr 3 , h = 

 60 cm., 17=169X10-* cm.; whence dh= icr 3 X6o/(i (6o/i6Q) 2 ) = 0.07 cm., 

 nearly. Thus, if with ruled gratings the fringes due to the D\ and D z lines 

 could be separately recognized, it should be possible to distinguish between 

 them here also, as the same phases require a differential displacement of G' 

 of nearly a millimeter. The same result would be recognized at M by a 

 displacement of dh tan 6, where 0=21 nearly, being the mean angle of dif- 

 fraction. The M displacement is thus 0.07X0.36 = 0.025 cm. 



In case of homogeneous light the prism grating G" is not needed and much 

 more light is available if the telescope is used directly. The strands of inter- 

 ferences, being on a yellow ground, are not very strong. Nevertheless a few 

 measurements of ranges of displacement were made by moving both M 

 (displacement e) and G' (displacement h), alternately. The following values 

 of e, h, and h tan 0' were found, the film gratings having nearly the same 

 constants : 



e = o.5cm. /i =1.30 cm. h tan 0' = 0.49 cm. 0=19 37' 0' = 2o4o' 



e and h tan 0' coincide as closely as may be expected, seeing that the fringes 

 in neither case can be quite brought to vanish. 



Experiments were next made with a grating of less dispersive power (D = 

 352X10-* cm.), ruled on glass and a stretched film grating of the same strength. 

 It was found, however, that the long rhombus GMG'N was very difficult to 

 control, owing to the reflection at almost grazing incidence. The spectra 

 also were not quite clear. The method was therefore eventually abandoned, 

 as no fringes could be found. 



The trial was then made with a weak grating at G (D = 352Xio- 6 cm.) 

 and a strong grating at G' (D = 167X10-' cm.). In adjustment the latter 

 naturally overpowers the former and two reversed spectra are seen in the 

 telescope (without prism grating) immediately behind G'. Both spectra were 

 quite strong and sharp. With white light no fringes could be found even after 

 long trial and a variety of adjustments. 



