the Spectrum of Copper, 473 



The above displacements respectively shift the limits by 

 — 61'5, 54*41, 208'96. The results are as follows — where 

 the second alternative refer to an additional two ouns on 

 the limit : — 



32-63 



33-18 



133-42 



la. 

 lb. 



Displ. 

 (12^)8(12**) 



(133)S(13££) 



v. 

 247-74 



247-71 



Shift from 

 -32-40 

 - 32-53 



la. 



2 b. 



(-lH)S(-93) 



(-io§d)S(-8a) 



249-09 

 24906 



33-29 

 33-13 



3 a. 

 3 b. 



(-42<S)S(-32£) 

 (-41£S)S(-3U) 



250-91 



250-88 



133-60 

 133-47 



* 



In each case the first arrangement reproduces the v with 

 practical exactness, but the observed are shifted to the red 

 respectively by *23, *11, *18, or 15, 2a } 3 a by the same 

 amount (say *15) within error limits. These are greater 

 than the limits of accuracy allow. Either therefore the 

 second sets must be taken as correct, or there is a shift 

 produced by say the electric or pole effect in the arc. 

 Relatively to each other the first set completely satisfy the 

 conditions, and if Meissner's and E.V.'s measures for Si (2) 

 are taken the whole set including S (2) will fall into line. 

 But it is safer to deal throughout with the same observer. 

 It may be noted that the two displacements in each case have 

 the same sign. The same rule as is already recognized in 

 the double displacement which D 22 experiences from D n . 



(2). In seeking for the high order lines of P(m), lines 

 were found close to the calculated values of P(7, 8). They 

 were however too strong to be identified w r ith the latter. 

 But it was noticed that the separations of 62289, 61451, 

 60908, 60536 were just those given by Rydberg's tables for 

 denominators 5*91, 6*91, 7*91, 8*91 but in reverse order — 

 that is. they were the higher orders of a summation series. 

 The limit calculated from them was in the neighbourhood of 

 59150. It was thus easy to search for the corresponding 

 difference series. Such was found, but the means gave a 

 limit 59174-8. The data are collected in Table IV., in 

 which the limits of variation are those of equal probability — 

 i. e. due to Handke's measures being only given to *1 A. 

 The quite definite allocations are indicated by * ; the others 



Phil. Mag. Ser. 6. Vol. 39. No. 233. May 1920. 2 I 



