the Spectrum of Copper. 471 



Appendix. 



I collect here a few examples whose importance lies in the 

 light they are capable of throwing on certain fundamental 

 points in spectral theory. The first illustrates the phe- 

 nomenon of double displacement. The laws governing 

 this effect are at present quite unknown and their discovery 

 is a matter of the first importance. The discussion is 

 however difficult because mere numerical coincidence 

 between a suggested double displacement and observation 

 has little weight as evidence that the suggestion corresponds 

 to a real connexion. In fact it only has weight if the 

 observed values are exact to within dn — ±'02, or where 

 the oun itself is very large or where there is additional 

 evidence in its favour. In the case here illustrated the 

 last consideration is present. The second is of interest 

 as affording the first example of the discovery of a new 

 series through the summation lines, and also further 

 evidence of the fraying out of the normal P series in 

 copper. The third contains further evidence for multiple 

 D CI) systems or for separate such displaced ones. The 

 fourth will illustrate the existence of the complicated 

 systems in one order of the F series due to linkage and 

 successive displacements in sequents and limits. 



(1). In the neighbourhood of S (2) the lines between 

 A, = 81 14 and 7848 may be arranged as follows — in I. A. 

 wave numbers : 



8(2), 



2)1232073(-03)32-63 (10)12353-36(-03) 3318 (2) 12386-54(05) 133*42 (2) 1248678(-03) 

 24774 24843 24910 25092 



2) 12568-47 ("03) (10) 12601-79 (-02) (1) 12635-64 (-05) (1) 12737-70('05) 



together with (2) 124G8'84(-03) which is 115-48 ahead 

 of Si (2). The separations in the top line are given as 

 from Si (2) . They clearly denote systems of doublets which 

 are displaced in some definite way from the normal S(2) 

 lines. They show, as they ought on this supposition, 

 increasing doublet separations with shift towards the violet. 

 Before considering the amounts however it is essential 

 to know the precise limits of error in the measures. They 

 are all by the same observer (Meggers*) in I. A. The 

 number in brackets on the right of each wave number 

 gives the variation dn as calculated from his estimates 

 of probable error d\. We have to do, however, with 



* Bureau of Standards, No. 312 (1918). 



