5 Conclusions

 

In this paper we have discussed the overall quality of IUE high resolution spectra processed with the NEWSIPSsystem.

The stability of the wavelength scale ($\approx$ 5 km s^-1) is within the limits imposed by the acquisition and tracking accuracy. No appreciable distortions in the wavelength scale over the full spectral range or during the spacecraft lifetime have been found.

A discrepancy of 9 km s^-1 has been found in the velocities derived from the two components of the Mg II doublet at 2800 Å in the LWP camera. The correct velocity is provided by the K line measured on spectral order 83. No such discrepancy has been found in LWR spectra.

The wavelength scales of NEWSIPS short and long cameras present an inconsistency, which is well above the repeatability errors quoted above. Measurements of narrow interstellar lines have shown that SWP velocities are systematically more negative by -17.7 km s^-1, on average. A similar discrepancy has been detected in long wavelength small aperture spectra, whose velocities are more negative than those from long wavelength large aperture spectra by -13.7 km s^-1.

The determination of the inter-order background has greatly improved with respect to the the IUESIPS system, especially at the shortest wavelengths, as shown by the absence of negative fluxes in the cores of saturated absorption lines and by the greater accuracy of equivalent width measurements.

The INES system derives two spectra from each original NEWSIPS MXHI file. In the first one, the "concatenated'' spectrum, the spectral orders are merged together and the overlap regions are suppressed according to an algorithm which computes suitable cut wavelengths which maximize the signal-to-noise ratio. This spectrum contains an error vector calibrated into absolute flux units, which is not available in NEWSIPS data. Since the INES high resolution spectra are obtained from NEWSIPS MXHI files, all previous considerations about the stability of the wavelength scale, the stability and accuracy of the flux scale and the validity of the background extraction are applicable to them. The second output product is the "rebinned'' spectrum, which is the "concatenated'' spectrum after resampling at the low resolution wavelength step.

To correct for the discrepancies found in the NEWSIPS high resolution wavelength scale, a velocity correction of +17.7 km s^-1 for SWP spectra and of +13.5 km s^-1 for LWP/LWR small aperture spectra has been applied to the INES "concatenated'' spectra. With this correction, the overall INES velocity scale is self-consistent, and agrees to within 8 km s^-1 with the optical velocity scale.

Acknowledgements

We would like to acknowledge the contribution of all VILSPA staff to the development and production of the INES system, and the referee, Dr. J.S. Nichols, for her useful comments.