We have retrieved all the IUE high-resolution spectra
taken between 1978 and 1993 from the IUE VILSPA data-bank. We have however
disregarded some spectra taken
with quite short exposures in which the signal was good for a few lines only.
For every year, the numbers of SWP spectra used and the corresponding exposure
times are reported on the top of Table 1.
We have employed the standard procedures for spectral analysis contained in
the ESO Midas package to measure wavelengths, FWHMs and emission intensities.
We recall (Turnrose & Thompson 1984) that the final error in the
wavelength measurement of a given spectral feature in various IUE spectra is a
combination of various errors due to a number of intrinsic and
extrinsic factors and is on the order of 0.04-0.05 Å around the
mean value although the uncertainty in the positioning of the target
in the large aperture can produce larger errors.
The ability to centroid a spectral feature in the spectrum is
also a limitation to the accuracy of the assigned wavelengths.
The spectral resolution in the IUE high resolution mode, as
indicated by the FWHM values of the Pt-Ne calibration spectra, varies from
about 0.10 Å at 
to 0.19 Å at 
Å in
the SWP camera. For an isolated
narrow feature the best estimate of the measurement accuracy is on the order
of one tenth of the instrumental resolution, that is on the order of 0.02 Å,
but in many cases the spectral data are not of sufficient quality to
approach this limit and slightly larger errors are found.
In our measurements we have found a confirmation for these results,
and only small differences, on the order of
a few hundreds of an Angstrom, have been found in the wavelength
determination for the same
feature in different spectra. This confirms the result by Penston et al.
(1983) of a rms scatter on the order of 0.05 Å of the line wavelengths about
their laboratory value.
For this reason and for sake of simplicity
we have reported in Table 1 only one value (the average) for 
for
each set of spectra. Needless to say, gives the
wavelength value as directly measured on the spectrum, not corrected for the
radial velocity of RR Tel relative to the
Sun, that has a value close to 
(Thackeray 1977).
Instead, note that both 
in Penston et al. (1983) and
in ADB are already corrected for the star radial velocity.
We recall that the corrections for the earth and
satellite motions, and therefore the reduction to an heliocentric
frame of reference, are made during the image processing procedure.
As customary, the wavelengths listed in the SWP spectra are vacuum
wavelengths.
The spectra have been calibrated using the recent high dispersion absolute calibration of Cassatella et al. (1994) and have been corrected for a reddening value E(B-V)=0.09 using the curve of Mathis (1990). This correction is very close to that employed in other studies.
One of the main shortcoming in the observations of emission line objects with IUE is the limited dynamic range in the response of the IUE cameras which leads easily to saturation effects even at exposure levels rather low as compared with the BKG. As a consequence it is generally necessary to employ spectra taken with different exposure times in order to obtain an optimal emission line spectrum. We have carefully checked the line quality for the presence of saturation effects and for any flag associated with data points corrupted by cosmic ray events.
Only spectra taken with the large aperture have been used in order to provide reliable measurements of the emission intensities: in spectra taken with the small aperture the throughput is on the order of 50 percent on the average but the scatter is quite large.
The emission intensities have not been corrected for the degradation in the sensitivity with time of the SWP camera. The reason is that the most recent information (Garhart 1992) comes from a study based on low resolution data in three limited spectral ranges that do not fully cover the whole SWP region. These data show, however, that there is an average total degradation close to 10 percent for the time interval 1978-1992, with a maximum of 12 percent in the 1225-1375 Å range. These values are close to the photometric accuracy of IUE which is on the order of 10 percent.