Cause of the Structure of Spectra. 267 



numerous experimenters are therefore relative values of 

 (N — l)m/p. For the metals the values given in text-books 

 and tables o£ physical data belong to their atoms in the com- 

 bined state, atomic refraction for the pure metals being a 

 difficult subject, notwithstanding Kundt's measurements of 

 the index for certain metals, because in such very absorbent 

 substances as the pure metals the apparent index of refraction 

 may have a different meaning from the ordinary. But if we 

 take the refraction-equivalents of the metals in compounds, 

 as given, for instance, in Meyer's 'Modern Theories of 

 Chemistry,' and the relative values of m/p the atomic volume 

 given in " Further Studies on Molecular Force " (Phil. Mag. 

 [5] xxxix. pp. 24 & 36), we can calculate N. the index for the 

 atom, and also l/N(m//o)s, which gives relative values of the 

 frequency of free vibration of the aether of the atom, namely, 

 velocity of light through atom divided by linear dimensions 

 of atom. The following table contains the data (N — l)m/p 

 and m/p, and the derived magnitudes N and l/N(»i//o) :! - 











Table IV. 

















Li. 



Na. 



K. 



Eb. Cs. Be. 



Mg. 



Ca. 



Sr. 



Ba. 



Zn. 



Cd. 



(N -])„//> . 



.. 3-8 



4-8 



81 



140 137 57 



70 



104 



136 



15-8 



102 



13& 



in f> 



.. 20 

 2-9 



7'4 

 1-65 



18-6 

 1-44 



34-4 560 10 

 1-41 124 67 



56 



225 



8-6 

 221 



10-6 



2-28 



166 

 1-95 



106 

 1-96 



10-6 



N 



2-28 



1 N(Wp) 3 . 



. '273 



•313 



•262 



•218 211 -149 



•250 



•221 



•200 



•201 



•286 



•200 



Now in the Li family these relative frequencies of the free 

 vibrations of the aether of the atoms run a very similar course 

 to p n and u,n the fundamental frequencies in the principal 

 and usual spectral series of these elements, as the following 

 comparison shows, where in the last line we have given also 

 p n —uti . which Rydberg discovered to be equal to the fre- 

 quencies of the chief lines in the spectra. 



Li. Na. 



•273. -313 



P )i 43506 41453 



«>/ 28600 24490 



p« -tiw — 149 °6 16963 



The first point to arrest attention in this table is the close 

 parallelism between the first and the last rows, both or! which 

 pass through a maximum at Na. But this is probably an 

 accident, because, from the nature of the calculations involved 

 in finding the first row, there is no guarantee that its maximum 

 at Na is a genuine fact ; for example, the small value of m/p 



K. 



Kb. 



Cs. 



•262. 



•218. 



•211. 



35009 



33707 



31372 



21966 



20861 



19721 



13043 



12846 



11651 



