408 DR. W. M. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES. 



Consequently in the above list the value of has no importance in settling the order 

 of the displacements. 



The whole of the foregoing argument is based on the constancy of A 2 for all series. 

 This matter has been referred to in the introduction in which the question of what is 

 to be understood by the atomic weight was brought up. The accuracy in the 

 determination of the oun is rendered so great by the constitution of the d and/ 

 sequences, that the mass of the electrons connected with the nucleus affect it. It may 

 be interesting to illustrate the considerations there adduced by a concrete example. 

 The example we will take is (l), as the result may possibly throw some light on a 

 difficulty which will appear later. The line 19880 is seen to require a displacement 

 of _ Let us determine the transfer of electrons in order to produce a change by 

 one oun. Suppose this transfer changes x to x + x'. This means that the mantissa 

 as represented in the list must be diminished by 8 Ox', by putting y = -1. 



Hence 



x' = - 2 = '65. db = '036. 



80 



The change in the number of electrons (see p. 342) = 925 x '036/611 x 130 = 1'97 = 2. 

 The addition of eight electrons to the mass acting in our standard case, would render 

 19980 a possible D line with limit ( <S,)D(o>) instead of a possible one with 



(-J)D(oo). 



The preceding treatment of the material is only a first step towards unravelling the 

 intricacies of these D systems. An exhaustive treatment is here impossible, and 

 would involve the consideration of other data the triplet separations, linkages, 

 similarities of arrangement, dependence on F series and so on. All that can be done 

 is to give a few illustrative instances and to bring into prominence certain problems 

 whose solution in the future may be of extreme importance. 



(l) The line 20021 is given as requiring the displacement 3^ in the limit, and in 

 this case the mantissa is 80A 2 . This displacement increases the limit 51025 by 31 '8 8, 

 and the resulting Vl should be greater by 3 x '535 = 1'60 and = 1779'50. As a fact the 

 line (see p. 382) forms a doublet with separation 17 79 '51, and with intensities 1, 10, in 

 the proper order for a D 12 set. All the tests support each other. Again in the 



doublets 



(7) 20500-13 1784-68 (2) 22284'82 



(8)20559-08 1785'64 (< In) 2234472 



the same displacements (or of +3^) are indicated. If we suppose these modified 

 values of i/j produced in the same way as in Kr by a relative displacement in the 

 sequent, the oun in this case alters the separation by 4'91. The modified v l therefore 

 becomes 1779'50 + 4'91 = 1784'41, or within error limits of the observed value for 

 20500, but too small for 20559. That for 20559 corresponds to an extra displacement 

 of 2<$, on the limit, or 5<5, in all, which is quite inadmissible on the quak'fying test. 



