20{) 



Mr. D. Gill. On the - Presence of Oxygon 



As 

 As 

 As 



or generally 



where 



= 71 + 2$, 



AA. 

 As" 



AX> 



As 



AXx- 2 

 As 



= 7? 



=1 :— = ?i~ 2d 



X = 



X K - njd. 



Such a law can only be applicable to a limited portion of the whole 

 spectrum, for it is obvious that for no value of X can A A. /As be really 

 = 0. 



In order to test over what range of the spectrum this law might be 

 regarded as sufficiently rigorous, four selected iron lines were measured 

 in terms of the millimetre scale which is attached to the instrument, 

 the division errors of which are known for each 5th millimetre. For 

 convenience the measures are converted into screw revolutions of the 

 micrometer microscope which was used in measuring the spectra, viz., 

 1 revolution = 0*5 mm. 







Screw Eev. 



X 1 = Fe 



4583-99 



200-254 



^2 = a 



4442-52 



159 130 



h = „ 



4282-54 



101-802 



A 4 = „ 



4187-99 



59-520 



Hence, 



Xi + Xo . AA-4513.2G 



for — - — we have — -r-— 

 2 As 



3-4400 



cl 



0-0043053 



3714-80 



A.-3 + X s 

 2 



x 3 + x d 



As 



— = 2-7906: 



AA 



■0-0043572 3722-08 



4325-27 



As 



= 2-2361 



Thus for interpolation between adjoining known iron lines the 

 second differences, d, are practically constant over the range of spec- 

 trum with which we have to deal, and consequently the logs of (X - X ) 

 vary proportionately to the measured intervals between the lines. 



If this were strictly true for the whole range of our spectrum one 

 would obtain rigorously accurate interpolation as follows : — 



