34? 



dispersion curve. There is still another reason why strong lines 

 — many of wliich lose their "wings" near the limb — require 

 separate treatment, nairiely becanse for such lines, according as the 

 limb is approached, it is indispensable to make due allowance for 

 the spherical shape of the source of light when the consequences 

 of diffusion, and particularly of irregular ray-curving, are inquired 

 into. Indeed, looking almost tangenlially towards the source, we are 

 no longer allowed to assume that the darkness of the line increases 



with -y. and with h, particularly not it n — 1 has great values. Such 



considerations suggest that in a further development of the theory 

 it will be necessary to reckon with a different set of conditions and 

 circumstances for different lines, especially very near the linil), where 

 the Fraunhofer spectrum passes gradually into the chromospheric 

 spectrum. 



The sharply differentiated structure visible in the chromosphere 

 at times of excellent seeing indicates that, at least at a level only 

 slightly outside the apparent edge of the disk, the gaseous medium 

 must be highly transparent along the path of the nearly tangential 

 rays, even for waves belonging to the very Fraunhofer lines. This 

 proves that in those layers molecular scattering is unable to make 

 the medium appear "foggy", in other words : that anomalous irregular 

 refraction plays a greater part there in determining the distribution 

 of the light, than anomalous molecular scattering. 



We infer that probably with most Fraunhofer lines, also with the 

 weaker ones, the darkness will depend to a greater extent on 

 refraction than on molecular scattering — though it appears possible 

 that the proportion between the respective influences differs from 

 line to line. 



AH this has to be taken in consideration when comparing our 

 theoretical results with observational data. Fig. 7 shows the upper 

 limits of the effects of mutual influence to be expected in the cases 

 we discussed, if the lines were [Hire refraction- or pure diffusion- 

 lines. In Fraunhofer lines the two processes are probably intermingled 

 and the respective displacements opposed; but refraction is likely 

 to have the advantage. 



We therefore may expect, e.g., if the distance between certain 

 Fraunhofer lines lies between 1,5 and 3 times theii- width, that 

 their mutual repulsion at the limb will exceed their repulsion in 



o 



the centre by an amount certainly not greater than 0,ÜÜ2 A. 



Now, according to the above-mentioned observations of Mount 



Wilson and Kokaikanal, the examined effect has the average value 



23 

 Proceedings Royal Acad. Amsterdam. Vol. XXVI. 



