440 



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



not a D line, but is either a F line, or is attached to a D line (as in the 1864 set in X 

 attached to the D group near 20312). The D-qualification test with D ( co) = S ( oo) 

 can be applied to these suspected D lines, taking the 20438 as a definite D n line. 

 That is the mantissa difference of the lines from that of 20438 must be an oun 

 multiple. A very large number are found to satisfy this test. It must be remem- 

 bered that as the oun is so large as 447 there can be no doubt as to the satisfaction of 

 this test or not, even for the largest allowable observation errors, nor to the actual 

 oun multiple when it is satisfied ; on the contrary the possible observation errors are 

 so great, that the observations do not enable us to increase the accuracy of the oun 

 itself as already found. By combining a large number of cases it would no doubt be 

 possible to diminish the probable error of its value, but the heavy work would not be 

 justified at present, especially as there should be hope of better measures in the 

 immediate future. At the same time the existence of the D lines may be illustrated 

 by the following two sets for m = 1 . 



-(2) 18641-4 (W.) 5387-12 



:(1) 19842-4 (W.) 5371-3 



318 



31J8 



(2) 16062 (W.) 



5650 

 -(6) 21712-09 (R.) 



(3) 16725-1 (W.) 

 "(1) 17493-05 (R.) 



.(0) 18609-94 (R.) 

 "(0) H813 (W.) 

 5625-8 



.(4) 20438-8 (W.) 



5377-68 



5372-9 

 5375-2 



5383 9 



18357 (C.R.) 

 5671 + 4 



(1) 24028 -52 (W.) 2640-71 



19583 (RAM.) 

 5630 



(2) 25213-7 (W.) 



(1) 21456-56 (R.) 



5633-21 



(2) 27089-77 (R.) 



(0) 16445-61 



5652-4 



(4) 22098 (C.R.) 2650 '6 



(1) 22868-07 (W.) 

 (1) 18357 (C.R.) 



5636 

 (10) 23993-82 (R.) 



(1)21036-25 



5633-0 

 (1) 26669-23 



(0) 19086-1 (W.) 

 5662 



(2) 24748-6 (W.) 



In these it should be noted : 



(1) The satellite separations are in the usual ratio 5:3, for in the first 20^x5 

 = 101-25, 34x3 = 102 or for D 23 20^x5 = 102'5, 33|x3 = 101'25. In the second 



18ix 5 = 91'25, 3lx 3 = 9375. Both sets have the same undisplaced limit = 50403, 

 and the two corresponding D n lines are displaced 23<S in their sequences. 



(2) The 21712 which acts as a D n line shows the modified v l separation to a second 

 line but of much smaller intensity, in close analogy with what has been seen in X. 



(3) The mantissa of 16725'1'5 is 804601-26'8+13-4p =25 (32166-16-1'07 

 + < 53p)+J 1 = 25A_, + <5 1 . The second line of the triplet 22098 is one of the copper 



