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



The D, satellite gives 



928A,, = 980956 +2> 3 -35'44 obs. error = '02p 3 

 A a = 1057-064 + O4.ps-0882 



Also the/(l) gives 



A 2 = 1057-047 + '019Qp 1 --0346 



The value of in /(l) is not precisely the same as in the others as it involves the 

 observed error in D n . If is the value for d n (l) that for/(l) is f-'12jt> 2 (dA = - 02p 2 ) 



and 



A, = 1057-047 + -0190P, + '004^- '0346 



Thus p, is of the order '9 and the best value is 



A, = 1057'064-'0382f'004. 



The value of is probably a very small positive quantity. The O C for D,, (4) is 

 dn = 1'25 pointing to a positive The data for the accurate determination from 

 ( F + F) are defective. Of directly observed lines those for in = 6 appear reliable as 

 the separations are good. This gives the limit as 27832'12 or = '17. Failing any 

 better determination we put = '17 + 



A, = 1057'057-'0382'004. 8 = 57'9209-'0021 '0002. 



Probal >ly 



> I and 8= 57 '921 '002. 



The manner in which so many independent conditions fit in to give this value is 

 very remarkable, and should give great weight to the accuracy of the above value. It 

 must not, however, by itself be regarded as a definitive value free from all possible 

 doubt without further support from independent groups of D or F series. This is not 

 easy to obtain. In the first place the spectrum of A seems to approximate more to 

 the doublet sets of He, and we lose the advantage of dealing with triplets. Again, it is 

 clear from the way in which the lines are crowded in the region below n = 2300 and 

 in which modified separations in the neighbourhood of 180 preponderate that a large 

 number of displaced groups must exist, and displaced groups have generally shown a 

 fragmentary quality. It is difficult, however, to deal with these in the same way as 

 in Kr, X, or RaEm, because the oun is so small that the separations produced by one 

 oun displacement are comparable with observation errors or uncertainties in the 

 limits.* 



* I had prepared the details of an additional D group with corresponding F and F series depending on 

 a displaced D ( oo). As however, while supporting the above value of the oun, it does not decrease its 

 possible error, no present object is to be gained by printing it. 



