Value of the Silver Gun. 309 



iind the following material, which supports the evidence for 

 S series. 



s. S. 



I 12082-G3(-7) 30626-56+1 (2»)4yi7u'49(2"4) (-481,414) 

 92079 



13003-42(-8) 31575-30+1 ( 1)50147-19 (2"5) (-153,464) 



30642-63+15 (49202-56) 



31562-47 ±15 (50121-53) 



I (8) 21413-37 (•< 

 m=3l 92042 

 t (6)22333-79 (•{ 



(8) 21413-37 (-46; 30641-85 ±'7 (1)39870'34 (S) 

 •5) 31564-90±-7 (l;40796l)l (*8) 



,(1)25106-89(1) 30643-65 ±1-3 (361 80-42) (1-6) 



m=U 91871 



'(2)26025-60(2) 31565-38+1-3 (37105-17) 



That there is something abnormal in the S series of the 

 Ou subgroup is already known. This is shown by the great 

 difference in the limits found by formulae from the S and D 

 series in all this group. Directly observed S lines are found 

 for m==3. tFor m = 2 the only lines observed near suitable 

 positions are those <.<iven (the differences from the next lines 

 on either side are inserted on the right). The mean for 

 S 1 (oo)is nearer that found by direct calculation from the 

 series (30616), but the mean S 2 (sc) deduced shows a 

 separation of 950 and quite abnormal. The observed lines 

 cannot therefore both belong to S. For m = 2 the oun dis- 

 placement on the sequent gives a change of 1*604. If they 

 be regarded as displaced lines S,(2)(5S) and S 2 (2)(4S) we 

 get the set shown in the table below the others. The mean 

 limits as there shown will not, however, be treated as of 

 evidential value except as showing that if the allocated 

 displacements are correct, f is about — 1 or — 2, and so falls 

 in with the results obtained from the D series. It is probable, 

 however, that the displacements are not wholly on the S lines 

 but that some at least occur in the S, and so account for the 

 •difficulty as to the calculated limits referred to above. In 

 fact this consideration might give data to settle the actual 

 displacements taking place in S. Such can only take place 

 in the sequents for otherwise the normal separation, 920*43 

 within error limits, would not exist. 



For ??i = 4 there are no directly observed lines, but clear 

 linked ones, viz. (In) 39951/42(1 "6) as Si(4).^ gives 

 S = 36180-42(l-G) whilst (1) 40876-89 (-8) as S 2 (4).« and 

 (1)33333*44(*5) as <?.S 2 (4) give respectively S 2 = 37105*89 ('8) 

 and ...04*44 (*5) of which the mean is taken in the table. 



