28 SECTIONAL ADDRESSES. 



empirical so far as they are concerned in the mere analysis of a spectrum, 

 and are subject to such modifications as may be indicated by theoretical 

 considerations. 



Unfortunately, the analysis of a spectrum does not always lead to a 

 knowledge of the actual values of the terms, or energy levels. These can 

 be determined for any of the relatively simple spectra, in which com- 

 paratively extended series can be traced and their limits calculated. In 

 most of the complex spectra, only the relative values of the terms have 

 been deduced, since extended sequences in these spectra are apparently of 

 rare occurrence. Even for these, however, the term of highest numerical 

 value, representing the lowest energy level, can often be identified, and 

 this is of special value in view of its association with the normal state of 

 the atom. 



This completes the story of spectroscopic terms and their possible 

 combinations on what might be called a purely numerical basis ; that is, 

 in so far as the analysis of a spectrum can at present be based merely on a 

 table of wave-lengths and intensities. Especially as regards the more 

 complex spectra, however, advantage has to be taken of every possible 

 experimental aid to the classification of the lines — particularly, in the 

 first instance, as a means of sorting out the lines characteristic of an 

 element at difierent stages of ionisation. I shall return to this subject 

 later. 



Thanks to the industry of numerous workers, many of the complex 

 spectra have now been partially analysed, and two of the principal 

 generalisations foreshadowed some years ago have been greatly 

 strengthened. The first of these is expressed by the so-called 'alternation 

 law,' according to which the arc spectra of the elements are alternately 

 of even and odd multiplicities in passing from the first to the higher groups 

 of the periodic table. No exceptions to the rule have yet been found. 

 Until recently it was thought that the maximum term multiplicity was 

 equal to the chemical group number increased by unity, but recent work 

 has shown that this simple rule is not of general application ; for example, 

 in the arc spectrum of copper quartets occur as well as doublets.™ 



The second generalisation is expressed by the spectroscopic ' dis- 

 placement law,' which states that the first spark (enhanced) spectrum of 

 an element has a structure similar to that of the arc spectrum of the 

 element which precedes it in the periodic table. To make this generally 

 applicable, however, it is necessary to qualify the rule by restricting the 

 meaning of similarity to a common odd or even multiplicity. The 

 spectrum of ionised scandium, for example, though including singlet and 

 triplet terms, differs from the spectrum of neutral calcium in having a 

 ^D term in place of a SS term corresponding to the normal state. The 

 same rule may be extended to higher states of ionisation, the second 

 spark spectrum, for example, resembling the first spark spectrum of the 

 preceding element, or the arc spectrum of the element of atomic number 

 two units smaller. Clearly the alternation law of multiplicities is also 

 applicable when the first or higher orders of spark spectra of the elements 

 are respectively compared. 



20 Shenstono, Phil. Mag., vol. 49 (May 1925) ; Beals, Roy. Soc. Proc, A, 

 vol. Ill, p. 168(1926). 



