531 
layers than has the light from the limb, the asymmetry of the centre 
lines is smaller, as a rule, than the asymmetry of the limb lines; 
the difference manifests itself in the limb-centre displacements. 
Which are the greater displacements, those of the limb lines with 
respect to the centre lines, or those of the centre lines with respect 
to the positions of the solar proper frequencies? 
Supposing the solar and terrestrial frequencies to be identical 
(absence of Einstein effect), and exactly given by the positions of the 
are lines, the simple comparison of the average limb-centre displa- 
cements with the average centre-arc displacements would teach us 
that there is not much difference in magnitude between the said 
two categories of displacements. The second one gives a somewhat 
smaller mean, as will be shown further on. The great inequality of 
the displacements of different lines is also similar in both cases, and 
points to a common origin and nature of the two phenomena’). 
Moreover, various considerations regarding the constitution of the 
outer layers of the sun lend support to the inference, that the light 
from the central parts of the disk has had on the average not quite 
half as much opportunity to be refracted and scattered, as has the 
light coming towards us from the marginal parts. 
The deductions from the dispersion theory are therefore quite in 
harmony with the observed displacements towards the red (both 
limb-centre and centre-are shifts), if the terrestrial are lines really 
give us with great approximation the solar values of the proper 
frequencies. 
The following table contains a summary of the observational 
material we have used. The sun-are displacements given in the 
tables of GRrrBw and Bacnem (Verh. d. D. Phys. Ges. 21, 454, 1919), 
EversHep (Kodaikanal Bull. 36), Korps (Kod. Bull. 38), Eversnip 
and Royps (Kod. Bull. 39), 446 values in all, most of which are 
means derived from several observations, have been divided by us 
into three groups according to wave-length, and for each group 
(covering 800 A) the mean value dof the displacement has been taken. 
These three values of d are indicated by eross-bloeks in the 
diagram on page 527. The intermediate block-line C shows the 
general progression of the sun-are displacements with wave-length ; 
it remains below the curve F of the limb-centre displacements, as 
already remarked. 
1) It should be kept in mind 1. that with regard to the second category of 
observed displacements the uncertainty about the positions of many arc lines cannot 
be overlooked, and 2. that, according to the dispersion theory, a simple propor- 
tionality of the two categories is not to be expected. 
