DUNSTAN, RICE, KRAUS: SPECTRAL LINES OF SODIUM. 79 
as a compensator, so that the two pencils shall have traversed equal 
thicknesses of glass before reaching the eye. 
Under these conditions, after adjusting slightly the mirror G, the 
observer will see a series of brilliant interference bands. 
By turning the screw M so as to increase slightly the distance of 
the mirror H from the surface E, thereby increasing the differ- 
ence of path of the two interfering pencils, and suitably adjusting 
the mirror G, the interference bands, or fringes, can be made cir- 
cular and concentric. It is upon the ‘‘visibility” or distinctness of 
these concentric circular interference fringes, as the difference of 
path is further increased and the magnet excited or not excited, 
that the results stated in this paper are based. 
The fringes were observed by means of the telescope K, focused 
for parallel rays. 
The observations consist in estimating the ‘‘visibility” or distinct- 
ness of these fringes for successive differences of path of the inter- 
fering pencils. Curves are then plotted, using differences of path 
in mms. as abscissas and visibilities expressed as a proper fraction, 
as ordinates. From these curves the distribution of light in the 
source may be deduced, which distribution may also be expressed 
by a curve with wave lengths in Angstrém units as abscissas and 
intensities expressed as a proper fraction as ordinates. 
Michelson (Phil. Mag. Vol. 31, p. 338, and Vol. 34, p. 280,) has 
discussed fully the theory of the interferometer, which he shows 
will give the same interference phenomena as would be given by 
two plane luminous surfaces vibrating in the same phase, making a 
small angle with each other. When the mirror G is adjusted so as 
to show circular fringes, the angle between the two luminous sur- 
faces is zero. It is under these conditions that we have used the 
instrument, consequently we give the theory only for this case. 
If the distribution of light in the source is given by y=$(x) 
where y is intensity and x is expressed in differences of wave num- 
bers, measured from the mean wave number of the source, Michel 
C#+$? 
son has shown that the ‘‘visibility”, V, is given by ve i, 
Pp? 
Where; 
na 
(9° x)cos2mrDxdx 
« 
Not ( #(x)sinawDxdx 
HI { #(xpdx 
