410 DE. W. M. HICKS : A CRITICAL STUDY OF SPECTEAL SEEIES. 



be 30740'21+2031270=51052-91 which is 27'62 above the normal and has no 

 reference to the oun displacement. If 2031270 corresponds to the -4J, indicated, 

 the limit is 5106779 and the d sequent is the difference, or 30755'09. This is the 

 source of the 1864 separations. It is 2279 greater than the limit 30725'30 of the 

 F series considered in connection with this series and which gave 1864'10 and 829"64. 

 This limit corresponds to a -sequent 6^ less displacement or 29 '82 greater value which 

 is practically exact. This present sequent will therefore increase 186410 by 6 x '45 to 

 186670 and 829'64 by 6 x '22 to 830'96. The new sum is 2697'66, as against the 

 observed 2697 '44 in remarkably close agreement. This then supports the 4c5, 

 indication of the capability test. A SS l displacement would make the sum = 2697 '00 

 which though a worse agreement may yet be within observation errors. The same 

 -3S l would make the modified Vl = 1779'50 + 4'97 = 1784'47, or 72 greater than the 

 observed. One is almost tempted to suspect here an error greater than the ordinary 

 observation error. An error of '36, dX = '09, would make all three tests agree in 

 allotting 20312 to the 3<5, set, and would bring it into the group of lines considered 

 in Case 1. 



It should be noted that this 20312 line is the line with wave-length 4922 referred 

 to by LIVEING and DEWAR for its peculiar behaviour (see p. 350). 



(4) The separations of the triplet 19880 are 1778'42, 815'30. They suggest the 

 separation S t to which belong 1778'43, 815'42. The capability test gives 3^ and 

 the two are incompatible. In the D series given below the calculated limit from 

 the first three lines give the limit as 51045'37 or 20'08 above the normal. With the 

 uncertainty in a limit found in this way this is a displacement of 2<5j which gives 

 21 '2 5. Further there seems some evidence to show that this line with 20021, 20041, 

 20312, 20500, 20559 belong to one D (l) group of lines. The evidence consists in the 

 existence of parallel F sets showing displacements equal to the separations of these 

 lines. In other words these lines are D sets with the same limit. All these tests 

 mutually exclude each other. How can their indications be reconciled ? I suggest 



(1) The limit for the line is ( } ) D ( oo) and the observed v lt v 2 are thus explained. 



(2) The capability test is met by the transfer of six electrons indicated above. 



(3) It is not the first line of the series in question. 



(4) The F separations will be found to offer a natural explanation. 



In what has preceded an attempt has been made to allocate normal D series, but 

 they are clearly not the strongest sets. In my first attack on the D, F problem the 

 procedure adopted was to take the 19880 triplet as a clear satellite set, and attempt 

 by the application of KYDBERG'S table to find a series for this satellite series. The 

 three sets found were : - 



