122 



Prof. C. Barus on Elliptic and other 



parallel to the rulings. The half g' is provided with a micro- 

 meter screw, so that it may be successively moved from the 

 position g l in fig. 2 to the position g' in fig. 3, through all 



3> 3 



intermediate positions, while the half: g remains stationary. 

 Each of the halves g and g' is controlled by three adjustment 

 screws (horizontal and vertical axes of rotation), to secure 

 complete parallelism of the faces of the grating. Each, 

 moreover, may be rotated around a horizontal axis to place 

 the lines parallel to the slit of the collimator. The duplex 

 grating is mounted on a spectrometer as is usual for reflexion. 

 Finally, each half may be raised and lowered and moved 

 horizontally to and fro, parallel to itself, so that the half 

 gratings when coplanar may approximately reproduce the 

 original grating. 



After each of the spectra is clear as to Fraunhofer lines, 

 the interferences here in question are produced by bringing 

 these lines (the D lines for instance) into perfect coincidence, 

 horizontally and vertically. Under these circumstances i£ 

 the distance apart, e, is suitably chosen, the interference 

 fringes will appear throughout the spectrum. These consist 

 of an approximately equidistant series of lines parallel to the 

 slit, i, e. vertical lines, which are finer, cast, par., as the breadth 

 of the crack at S between the gratings is larger. They may 

 be increased from the extreme fineness as they enter the 

 range of visibility, to a maximum coarseness (in the above 

 experiments) of about 3 to 5 minutes of arc per fringe, after 

 which they vanish. They cannot in practice be passed through 

 infinite size ; neither can they be produced symmetrically 

 on the two sides of the adjustment for infinite size. They 

 cannot, in other words, be changed from the positive to the 

 negative condition of appearance. 



