4 Conclusions

We have shown that a reliable correction for the echelle blaze function can be obtained for IUE high resolution spectra processed with NEWSIPS. We find that not only the Kparameter, but also the $\alpha$ parameter varies with order number, and that the central peaks of the blaze function deviate from the reference positions in a way which depends on the value of the THDA temperature for the SWP camera, and on the observation date for the LWP and LWR cameras. Having taken into account all these dependences (and not only the dependence of Kon order number as in previous investigations), allows the present algorithm to successfully correct most spectra. According to our experience, the algorithm might fail in a few cases (probably less then 5-10%). The identified cases of failure correspond to images in which the interorder background level is not properly evaluated with the NEWSIPS automatic extraction procedure or to cases in which the target is not acquired (purposely or not) in the center of the large entrance apertures of the IUE spectrographs. In this respect, we note that a displacement of 1 arcsec along the dispersion line corresponds to a velocity shift of the wavelength scale of about 5 km s^-1, irrespective of the camera used. For example, a displacement of 5 arcsec, or 0.125 Å at 1500 Å would produce a flux mismatch in the overlap region midway between adjacent orders of about 4%.

As for the absolute calibration of IUE high resolution spectra, we have shown that its internal accuracy is about 4% and that it applies both to continuum and emission line sources. The flatter shape of the $C(\lambda$) function found here is due to the improved background extraction procedure in NEWSIPS (Smith 1999).

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. Joy S. Nichols, for useful comments.