300 BELL SYSTEM TECHNICAL JOURNAL 



With a very good spectroscope indeed, each member of each principal- 

 series doublet is in its turn resolved into a pair. This structure is 

 called the "hyperfine structure" of the lines; and this it is which indi- 

 cates that a still further subdivision of the states is necessary, and 

 invites and requires the introduction of the quantum-numbers F and 

 / and the angular momentum of the nucleus. Indeed, the concept of 

 the nucleus as a quantized vector was invented or discovered (which- 

 ever word the reader may prefer) during the interpretation of hyper- 

 fine structure of spectrum lines. 



Hyperfine structure of lines or states (for the name is applied 

 to both) is usually more crowded and compact than fine structure, 

 and yet there are exceptions: the fine structure of hydrogen is much 

 harder to resolve than the hyperfine of (say) the familiar mercury 

 lines 2537 and 5461, which itself was called fine before the theory was 

 developed. These structures, however, are generally near and often, 

 it is to be suspected, beyond the utmost capacities of the most refined 

 of optical instruments; whence, in many cases, extraordinary difti- 

 culties in measuring or even estimating the separations, the relative 

 intensities, actually the mere number of the distinct lines forming a 

 hyperfine pattern ; observers of great skill will often disagree with one 

 another, and judgment will often depend on a photograph taken with a 

 spectroscopic instrument such as an echelon or an etalon, which looks 

 totally different from the pictures obtained with gratings or prisms. 

 Perhaps this last is an advantage after all, as it discourages attempts 

 by the inexpert to interpret published photographs. Often several 

 different isotopes of an element produce different patterns which 

 signify different values of /, and are so nearly superposed on one another 

 as to make analysis superlatively hard. Hyperfine structure is for 

 the present, and quite probably will be forever, the "last frontier" 

 of spectroscopy. 



The task of deriving, from the hyperfine line-pattern connecting 

 two states or (better) state-clusters, the hyperfine subdivision of the 

 state-clusters or "hyperfine multiplets" themselves, is again an ex- 

 ample of the classical function of spectroscopy, which we shall take as 

 having been achieved. Actually it involves, of course, the use of 

 selection-principles, themselves connected with the atom-model, but 

 omitted from this article in order not to complicate it still more. 

 Some confusion may be prevented if I state that in our favorite case of 

 sodium, where the fine-structure splitting of the principal-series lines 

 implies a splitting of the P-states only, the hyperfine splitting implies 

 something more complex: it is due jointly to hyperfine structures of 

 both the P-states and the normal .S"-state, the latter being predominant. 



