428 Sir J. J. Thomson on the Origin of 



Pairs and Triplets, 



A common feature in spectra is that the series consists 

 not of single lines, but of a series of pairs, or triplets. 



These are not confined to any particular type of series but 

 occur in the principal as well as the subordinate series, though 

 the law of difference of frequencies is different ; in the 

 principal series the frequency difference between the con- 

 stituents of the pair diminishes as the wave-length dimi- 

 nishes, while in the subordinate series the frequency difference 

 is constant. 



The two constituents of a pair arise from different elec- 

 trons ; for Wood has shown that one of the D lines of sodium 

 can be excited without the other and, again, the Zeeman 

 effect for one line of a pair is quite different from that of 

 the other. 



We should expect to get pairs or triplets when instead of 

 a single electron we had a ring of electrons or a number 

 arranged at the corners of! ;( polyhedron, if the distribution 

 were not perfectly symmetrical about the centre. Thus, for 

 example, if we had a ring of four electrons, arranged not at 

 the corners of a square, but at those of a rectangle whose 

 diagonals were of slightly different lengths ; two of the 

 electrons would be at a greater distance from the centre than 

 the other two, so that the magnetic force, and therefore the 

 frequency, would be smaller for the first than for the second 

 pair of electrons, and since the variable part of the magnetic 

 field, being proportional to r\ would vanish at the centre, 

 the limiting frequency which corresponds to the vibrations 

 of electrons close to the centre would be the same for each 

 pair of electrons. We should thus get a series of pairs 

 behaving like those in the principal series. 



We have seen reason for ascribing; the first subordinate 

 series to the vibrations of the electron in an atom in which 

 the magnetic boundary has crept up to the position occupied 

 by the outer ring of electrons in an atom of the type of that 

 emitting the principal series. If the configuration of the 

 ring is slightly unsymmetrical, that of the magnetic boundary 

 for the atoms giving out the first subordinate series may be 

 expected to be so also. 



If this is so, the term in the magnetic force which does 

 not depend on r would be different in different directions, 

 and therefore the value of the first term in the bracket in 



