REVERSED AND NON-REVERSED SPECTRA. 45 



The sodium arc was now used with a very wide slit and the fringes were 

 found without difficulty. They consisted as usual of vertical hair-lines 

 rotating through a usually horizontal maximum back to vertical hair-lines 

 again, as either M, N, G' were displaced normal to their faces on their respec- 

 tive micrometers. These fringes are simply due to either one or the other 

 sodium line separately and therefore seen on a yellow ground free from inter- 

 ference. Even after this, with the apparatus in adjustment for homogeneous 

 light, white light was tried again in alternation, but no fringes appeared. 



With sodium light a few measurements of the ranges of displacement were 

 made. If 6 and 6' are the angles of diffraction, 5= 180 (0+0') and x = 2<? 

 cos 5/2, when x is the path-difference cut off at one end by the displacement e 

 of the mirror M. The displacement of G' being y, it appears that, apart 

 from sliding, e and y tan 6' should be nearly equal. The results were 



^ = 0.42 0.42 0.45 cm. = 9 39' 



x = 0.8 1 0.8 1 0.87 cm. 6'= 20 40' 



y= 1.74 i. 80 cm. 



l = o.66 0.68 cm. 5 = 



It was not practicable to make the hair-lines quite disappear without a 

 large excess of displacement in both cases. Even so the difference of e and 

 y tan 6' is too large to be explained by such an error. But the work with 

 the present apparatus (screws not long enough) is not sufficiently accurate 

 to make further discussion fruitful. The error will probably be associated 

 with the oblique incidence of rays in case of a wide slit. 



Very remarkable results were finally obtained with compensators of glass 

 plate. Placed in one or both beams and rotated around a vertical axis, they 

 rotate the fringes. This would be referable to the sliding of the ends of the 

 two pencils on the grating G'. If, however, they are placed nearly normally 

 in one beam, they produce no effect either of rotation or on the size of the 

 fringes; but the pattern is displaced bodily across the wide yellow slit image. 

 Glass plates 0.2 and 0.5 cm. were used. It is not until the thickness of plate 

 reaches 2 cm. that appreciable thinning of the interference fringes occurs 

 when the plate is placed in one beam. With optic plate this would be an 

 excellent method for testing the lengths of uniform wave-trains. Finally, 

 with a fine slit and coincident sodium lines, the fringes could be seen in the 

 presence of a continuous spectrum as marked dots on the enhanced sodium 

 lines. But nothing could be detected with non-coincident lines.* 



21. The same, continued. Duplicate fringes. The occurrence of strands 

 and apparently duplicated fringes has already been suggested in the preceding 

 paragraph. In further experiments definite results were eventually obtained 

 with sunlight. These occur in very great variety, but two typical phases 

 may be accentuated, given in figure 27. In the case a the two sets are more 



* Such fringes were since found, incidentally, in the white flash of a sodium arc. They 

 were very clear, but could not be controlled. 



