340 



theory, the widening of tlie lines near the limb and tiie increase 

 of their mutnal inflnence uinst be connected. Indeed, if tlie distribu- 

 tion of the relative intensity were established both for an isolated 

 refraction line of the centre type and for the corresponding limb 

 line, it would now be possible to plot the curve giving the distribu- 

 tion of the light in a set of two neighbouring equal lines, and to 

 examine how the asymmetry in it increases when |)assing from the 

 centre to the limb of the disk. It wonld only be necessary to sub- 

 stitnte for B and B' the values corresponding to relative intensities 

 0,9, 0,8, 0,7 etc., and then to calculate from (7) and (9) where the 

 places of equal intensity ought to be found in the pair. ') 



For the present, however, the intensity curves are not sufficiently 

 known; the observers of the solnr spectrum provide us with the 

 "visual widths" and the wave-lengths of the "centres of gravity" 

 of the lines, quantities by no means free from subjectivity. 



Nevertheless we can draw from such observations some useful 

 inferences concerning our problem. 



§ 4. Limitation to ivhat may be derived from already existing 

 data. 



The average widening in the limb-spectrum of lines whose widths 



o 



lie between 0,07 and 0,16 A was found by Fabry and BuissoN ') 



o 



to be 0,01 A. Although we do not know the exact value of the 

 relative intensity at the places where their interference method made 

 them estimate the "boundaries" of the line, there was yet in this 

 way assigned a definite width to each line. (We have some reason 

 to think that in the ordinary visual estimates of the width of a 

 dark line the relative intensity at the borders is about 0,8. This 

 statement reposes on extrapolation of an empirical formula by which, 

 in an earlier investigation, we were able to lepreseiit the visual 

 boundaries of bright lines on the photographic plate. Cf. Ann. d. 

 Phys., 71, 59, 1923). 



Whilst under the influence of a neighbouring line these boundaries 

 shift asymmetrically, the central parts of the dispersion line, with 



1) If limb- and centre-lines have been photographed on one and the same plate 

 (the centre spectra with shorter exposition so as to make the intensity of clear 

 spaces equal in both centre and limbspectra) it is even possible to use the 

 transparency values of the single lines directly for computing, 'by means of our 

 formulae, the course of the transparency in pairs of lines occuring on the same 

 plate. It is unnecessary then, first to translate the degrees of blackening into 

 original intensities. 



2) Fabry and Buisson, G. R. 148, 1741, (1909); Astroph. Journ. 31, 97, (191U). 



