602 Stoney — Cause of Double Lines in Spectra. 



follows that the apsidal motion in series P is in the same direction as the revolutions 

 in the partials, whereas in series D and 8 it is in the opposite direction. In most 

 other respects the analysis of this series is much the same as that which has been 

 ajjplied to series S ; and any further separate treatment of series P as a whole is 

 premature till more accurate observations shall have been made. 



But the great yellow sodium line which is the first term of the series, and which 

 corresponds to the Fraueuhofer line D in the solar spectrum, has been more care- 

 fully observed than any other of the sodium lines, and is on this account the best 

 in which to illustrate the extent of the information which can be elicited from 

 observations on a double line in the spectrum. 



For this investigation it is best to use the wavelengths-in-air of Professor Row- 

 land's great map of the Solar Spectrum issued in 1888. Reading from it, the 

 wavelengths-in-air of the two D (sodium) lines are — 



Xj = 5896'15 tenthet- metres, 

 \, = 5890-20 tenthet- metres. 



Taking the reciprocals of these, we find the number of waves in the tenth of a 



millimetre in air to be 



/ci = 169-602, 



/ci = 169-773; 



multiplying the former, and dividing the latter, by 1-000295 (Ketteler's value for /a, 

 the index of refraction of air, for this part of the spectrum, Pkil. Mag.^ Nov., 1866, 

 p. 341), we find for the wavelength in vacuo, which is the same as the periodic 

 time expressed in micro-jots — 



r, = ^\, = 5897-89 micro-jots, 



Ti = ixk2 = 5891-95 micro-jots ; 



and for the number of waves in the tenth of a mm. in vacuo, which is the same as 

 the frequency of the undulation of the aether in each jot of time — 



iVi = Ki/ju, = 169-552 in each jot, 



]V2= K./jx = 169-723 in each jot. 



Now by Problem II., p. 572, iVi = m — n, N^ = m -t- n, where m and n are respectively 

 the frequencies of the revolution of the electron in its ellipse, and of a complete 

 circuit of the apsidal motion. Hence — 



m (the number of elliptic revolutions in each jot) = — ^r — ^ = 169-637, 



n 



(the number of apsidal circuits in each jot) = — ^ — = -0855. 



