As a first step, the wavelength calibrations of the two FTS atlases
mentioned above were tested. This was also done for the Liège
Atlas, which is composed of disc-centre observations performed at
Jungfraujoch Observatory (Swiss Alps) using a double-pass grating
spectrometer, and which has been extensively used in many solar studies.
We have made use of the version available as part of the
KIS
IDL library.
It has been shown (e.g. Balthasar 1984) that once
the Sun-Earth velocity shifts have been corrected for, the maximum
displacement to the red exhibited by solar spectral lines
corresponds to the gravitational redshift due to the difference in
gravity between the solar and terrestrial surfaces. This is the
expected case for the three atlases
under consideration, which use wavelengths on standard
air (dry, 15 
C, 760 mmHg). Solar-Earth doppler shifts were corrected in the FTS flux spectrum, and the same is true for the tables
of solar wavelengths prepared by Pierce & Breckenridge (1973)
at Kitt Peak, which were employed to calibrate the wavelength scale of the FTS disc-centre spectrum. Both FTS spectra were compiled from eight (flux) or seven (disc-centre) carefully overlapped scans. The Liège atlas is supposed to be calibrated following the same reference, but prior to publication.
Following the procedure described in the next section, shifts for neutral iron lines were measured in the atlases. Figure 1 (click here) shows the results as a function of wavelength. It appears that both the FTS disc-centre spectrum and the FTS flux spectrum agree in the absolute scale within the errors and the expected differences between intensity and flux measurements. They do not show any stronger than expected trend (Hamilton 1997) and are in agreement with the expected maximum redshift of 636 m s-1 (solid line), which corresponds to the gravitational effect. On the contrary, the wavelength calibration of the Liège Atlas differs clearly on the absolute scale, exhibiting redshifts larger than 636 m s-1 and showing a strong trend with wavelength, which points towards errors in the spectral calibration procedure. For this reason we have discarded the Liège Atlas in this study.
Figure 1: Line shifts measured in the FTS disc-centre spectrum (triangles) and
in the FTS flux spectrum (plus signs) show no trend with wavelength.
The largest shifts to the red are consistent with the expected gravitational
shift of 636 m s-1 (solid line). The shifts measured in the Liège Atlas (dots) exhibit a wavelength dependence, and go further to the
red than the gravitational redshift, thereby revealing calibration errors