530 
lines may have on the estimation of their distance must be practically 
identical in the two spectra: the exaggeration of the distance in the 
limb spectrum would also be 0,003 A. A mutual influence depending 
on errors of estimation will not appear in the limb-centre differences. 
The well-established fact that neighbouring lines are nevertheless 
more distant in the spectrum of the limb than in the spectrum of 
the centre must therefore be a real phenomenon. 
The only way to explain this mutual repulsion of neighbouring 
lines seems to be the one indicated by the dispersion theory owing 
to which, moreover, the phenomenon was discovered. A powerful 
support is thus given to our contention that Fraunhofer lines have 
indeed to be considered as dispersion lines, enveloping the (much 
weaker and narrower) real absorption lines in a generally asym- 
metrical way. 
The observational data as yet available contradict the inference that 
the Fraunhofer lines should be displaced by gravitation. 
We shall now refer to the displacements of Fraunhofer lines in 
the spectrum of the central parts of the solar disk with respect to 
the corresponding lines in terrestrial are spectra. 
The laws of these displacements are less easily derived from direct 
observations than those of the limb-centre displacements, because in 
many cases the wave-lengths of the are lines have not yet been 
determined with sufficient precision. It has however been established 
beyond doubt that here again the same peculiarities appear: great 
variety in magnitude of the displacements; dependence on line-inten- 
sity; mutual influencing of the neighbouring lines). This agrees 
with the conception of Fraunhofer lines as dispersion lines; indeed, 
also the radiation coming from the centre of the disk has been 
weakened by anomalous dispersion on its way through thick layers 
of gas, and therefore shows dispersion lines in its spectrum. As a 
rule these dispersion lines will cover the cores of the lines asym- 
metrically. (Ihe positions of the cores are determined by the solar 
values of their proper frequencies). 
As the radiation from the centre of the disk has in general travelled 
along shorter paths through the refracting and scattering gaseous 
1) ArBreEcHT, Astroph. Journ. 41, 333 (95); 44, 1 (1916). Royps, who 
challenges ALBRECHT’s conclusions, yet finds himself that 17 lines with companion 
on the red side give a mean sun-are displacement towards the red of only 0,0032 A, 
whereas 30 lines with a violet companion give a displacement towards the red 
of 0,0079 A. (Kodaikanal Bull. 48). 
