Seines System in the Spectrum of Gold. 



11 



(2 8 1 )D ll {l){2S 1 ).e, giving D it (l).« = 2024(H8. Although 

 exact, the double displacement is of small evidential value 

 unless supported by independent considerations. 



In connexion with the third kind of tests I have noted the 

 two combinations d\(l) — d 2 (l), d\(l) — ^i(l), where d' 

 denotes the sequent for the inverse D set already referred to.. 

 Since -;D' 11 (l)= i ?(l)-^ 1 (l) and D n (V y =p(l) -r/^l), it 

 follows that r/' 1 (l)-(/o(l) = D' n (l) + D 19 (l) = 19113-53 + 

 2733'7r=21817-2; this is observed at (2)21846-41. So 

 d\(l) -d 1 ( i) = D'n(l) + B n (l) = 19113*53 + 2980 = 22099,. 

 which is observed at (4w) 22102*18 within possible limits. 

 With the first it would give the satellite separation as 

 255-77. 



The cumulative evidence of these three independent con- 

 siderations is convincing that the D (1) lines as suggested 

 really exist, and that the rule for the extreme satellite 

 mantissa is not broken in the case of Au, as it apparently 

 was when the old observed D (2) line was regarded as be- 

 longing to the first set. The consideration of the F series 

 which follows will give direct proof that the d{l) sequence 

 at least exists, but before taking up that question it is 

 necessary to complete the discussion of the results of Table I. 

 which raise points of importance not only in connexion with 

 the properties of D series, but of those of displacement in 

 general. 



m = 2. No direct lines are found for the weaker D 12 and 

 D 12 , nor for D n . Rydberg had allotted (4) 17050*97 to D ]2 , 

 but K. R. have pointed out that its separation from 

 D 22 = 20858 is too large. The true valne cannot deviate 

 more than '32 from 17043-41. This differs by 7*56±*46 

 from 17050. The oun displacement on d(2) produces 2*93, 

 so that 7*56 is as far as it can be from a value produced by 

 2 8 1 or 3 8 1 and much beyond the possible errors of ±'46. 

 The line in question therefore cannot be a simple displace- 

 ment of D 12 (2). Since the oun on the limit produces 10*76, 

 it may be (»5 1 )D 12 (2)(S 1 ), which is separated from D 12 by 

 10*76 — 2' 93 = 7*83, but is probably unrelated to the D series. 

 The D n and satellite appear each displaced into at least two 

 others. Thus 



(4)41866-75 + -87j» as D(10^) gives D 12 =41896*05+-87p 



(2)41928-72 + -87^ „ D(-ll^) 



(1m)41798*68+1*7jo „ D(5^) 



(5)41859-04+ 1-7 „ D(-16^) 



For m 



D 12 =48896*49+ *87 p 

 Du =41813*18 + l-7jo 

 D u =41812*64 +l-7i» 



3 the satellite separation is due to 2(\ 8 X in line 



