Stone Y — Analysis of the Spectrum of Sodium. 215 



and computing the values of x, i. e. n, for the other terms, we get — 



Inferences. 



From the whole investigation we may draw the following 

 inferences : — 



1. That the outstanding differences are sufficiently large to 

 warrant the conclusion that the primary curve is not an exact 

 curve of the second degree, although in the case of Series P it 

 approximates to a hyperbola ; in the case of Series S to a hyperbola 

 or ellipse, probably to a hyperbola ; and to a parabola in the case 

 of Series D. 



2. That the approach to the parabola is very close in the case 

 of Series D, but that, in the case of Series P and in the case of 

 Series S, the actual curve, as indicated by the observations, is 

 somewhat more curved in the vicinity of its vertex, than is the 

 hyperbola which approximates to it. 



3. That the double line which has been regarded by Eydberg 

 as a satellite of one of the 'terms of Series D, and by Kayser and 

 Runge as belonging to a fourth series, is in reality the first, or the 

 second, term of Series S — the first, if the primary curve of 

 Series S is of a hyperbolic form ; the second if it is elliptic. 



4. That negative values of n furnish real lines in spectra, of 

 which the double line spoken of above is an instance. 



5. That in Series P the term corresponding to m = 1 has a 

 negative value for its w, viz. 469*4, approximately corresponding 

 to wave-length A = 2130. This is perhaps not at too great a 



