( 1271 9 
of the original brightness, and this effect must be strongest for the 
most refrangible waves corresponding to d, and d,; A,, which is 
situated there between those wave-lengths, is relatively narrow. But 
at some distance from the centre, where for every wave-length the 
average brightness is less than at the centre, density gradients may 
bring about, that light, emitted normally by the photosphere, curves 
toward the observer, thus producing a local increase of brightness. 
Now again we fall back upon the most refrangible kinds of light, 
those at d, and d,, as the ones that will be able to produce the 
latter effect in the strongest degree. Putting it otherwise: on the 
non-central places of the disk refraction will here and there contribute 
bright patches to the formation of the K-line, whose maxima of 
intensity have the greatest chance to correspond to the wave-lengths 
at d, andd,. This effect combines with the before-mentioned scattering- 
effect. The components of A, must therefore, on an average, be 
farther apart than on the middle of the disk. 
Proceeding toward the limb, the first part of the effect, that which 
is due to the dotted part of the scattering curve, diminishes, as already 
stated; while the second part, caused by refraction, gains in importance. 
Consequently the average distance between the components of K, 
increases, and, at the limb, becomes equal to that between d, and 
d,. The bright components are not interrupted there by dark patches, 
as elsewhere on the disk, because at the limb the strongly scattered 
waves always enhance the brightness. 
The influence of the two little tops of the R,-curve, which on the 
middle of the disk determined the K,-line, has disappeared near the 
limb; hence A, is wider than at the centre. Moreover, the displace- 
ments of A, and , to the red and the violet respectively have 
gradually decreased on approaching the limb; for they depended on 
the asymmetry of the little tops. The points d, and d,, determining 
the average places of the components of A, at the limb, are situated 
at equal distances from the real absorption line. 
§ 6. Weak points of the new and of the old explanation. 
While, all in all, these conclusions drawn from the dispersion 
theory show a very close agreement with the results of the obser- 
vations, we should by no means neglect to pay due attention to the 
points where discordance exists or seems to exist. Only if searching 
for defects, one has a chance to improve one’s views. 
So we must notice that, if scattering and irregular refraction 
determine the phenomenon, we have some reason to expect asym- 
metry of A, near the limb; and if we were right in assuming the 
