of Light hy Potassium Vapour. 295 



In practice it is impossible to take measurements close up to 

 the absorption lines. 



It n be the observed fringe number, taking ?i=l for the 

 first visible fringe on the photograph, 



%n = a{Ze + a 2 2/+ JcZ r • 

 A, 



Applying the method of Least Squares, the probable error 

 for all observations being approximately the same, we have : — 



a^ + ajS^+flS^Sen. 



A, 



A, 



6 f n 



1 e f 



Values of e 2 , / 3 , ef, ~ 2 , -, ^ were calculated for a 



Ay Ay A 



wave-length table which covers the distance between the 

 two absorption lines. These values serve for all photographs. 

 For each photograph readings were taken of the positions at 

 which the fringes cross a horizontal line. The positions of 

 each line of the iron spectrum were also noted, and the lines 

 having been identified on a map a wave-length curve was 

 plotted. One curve was found to serve for all photographs 

 to a close approximation. The wave-lengths corresponding 

 to the microscope readings for points where the fringes cross 

 the line were found from the first graph. These were plotted 

 against the fringe numbers. The fringe numbers corre- 

 sponding to the wave-lengths in the wave-length table were 

 then found from this curve, and the values of n, en,fn, ?i/\ 

 calculated for each photograph. 



Summations were then taken over the widest possible 

 wave-length range and values of a x and a 2 calculated. 



Many series of photographs were taken, but it was not 

 easy with the apparatus available to obtain consecutive 

 photographs over a wide temperature range with good 

 enough definition for the measurements required. At high 

 temperatures it was difficult to obtain clear photographs', and 

 at low temperatures the slope of the fringes was small over 

 the greater part of the spectrum ; so that several series 



