1 5 4 D O C U M E N T 8 M A Y 1 9 2 0
The following reasons, in brief, speak additionally for this interpretation: (1) the
generally small width of the Fr. lines and the great similarity between spectra from
the center and limb of the solar disk; (2) the type of relation between Fraunhofer
lines in the limb spectrum and bright lines of the chromospheric spectrum; (3) reg-
ularities in the line distortions in solar-spot spectra, generated by a radially posi-
tioned slit, which phenomena are only very unsatisfactorily explained by means of
the Doppler principle as a consequence of radial emissions, but on the contrary,
quite easily by the effect of refraction; and (4) the influence that total limb-center
shifts experience by the presence of neighboring lines. A neighbor on the red side
reduces the redshift, a neighbor on the violet side magnifies the redshift, namely,
ceteris paribus, the first effect is larger than the second, exactly as required by dis-
persion
theory.[10]
(Among 656 lines measured by Adams, Evershed, and
Royds,[11]
we found 44 lines with red companions, 49 with violet ones. The former
gave as the mean shift 0.0044 Å, the latter 0.0077 Å, while the mean value for all
656 lines was 0.0063.) For solar-center-arc shifts, the same seems to apply, as
emerges from the detailed paper by Albrecht (Astroph. Journ. 41, 333 [1915] and
44, 1
[1916]).[12]
Royds, who contests Albrecht’s results (Kodaik. Bull.
48),[13]
nonetheless himself finds that 17 lines with red companions furnish a redshift ( -
arc) of only 0.0032 Å; by contrast, 30 lines with violet companions yield such a
shift of 0.0079 Å. He does not provide the total mean of all lines he had taken into
account.
[St. John, Evershed, and Larmor took pains to show that no mutual influence of
adjacent lines exists and, theoretically (Larmor), is not expected to any noticeable
degree;[14]
however, I believe I can attribute the reason for their statements to an
incorrect conception of the nature of dispersion lines.– We at the laboratory are
now occupied with the problem of thoroughly analyzing, visually as well as micro-
photometrically, closely lying line pairs, using artificial
lines.][15]
Thus, if it can be shown that spectrum lines actually do exhibit noticeable dis-
placements by anomalous dispersion in the Sun, both toward the violet as well as
toward the red, it ought to be possible to explain the lack of a gravitational shift of
the relevant magnitude in many lines by a violet shift due to dispersion, which par-
tially conceals the gravitational effect in precisely these lines.
That is why I hoped it would be feasible to interpret all the material on the mea-
sured redshifts quite completely as a superpositioning of two main effects: a grav-
itational displacement—increasing proportionately with wavelength—of the
eigenfrequencies and a—strongly fluctuating—displacement of the enveloping
dispersion lines.
·