5 Conclusions

 More than 4 years of operation have proven the usefulness of the VALD concept as it was described in Paperi. Two mirror sites and more than 240 users from all around the world confirm the need for such a data base. Most of the proposed improvements listed at the end of Paperi have been implemented to various degrees in VALD-2. Only the compilation of telluric line lists has not yet been started. In particular, we have reported the following achievements:

• More than 16000 individual lines have been processed in preparation of VALD-2. It took an enormous amount of work not only to collect the data but to compare and establish the relative ranking between different lists, as described in Sect.4.
• References to primary sources for any measured or computed parameter value extracted by VALD are included to the output. Besides the proper referencing to sources, which is fundamental in scientific publication policy, this new feature also provides an important incentive to contributors of these much needed data to communicate their results to VALD.
• First steps towards including complete information on Zeeman pattern, Stark and van der Waals broadening constants have been made. In collaboration with specialized institutions (e.g. M.S. Dimitrijevic and L. Popovic, Belgrade Observatory; S. Sahal-Bréchot, Observatoire de Paris-Meudon; R.K. Janev, International Atomic Energy Agency) VALD will significantly improve quality and quantity of such data.
• Access to VALD has been simplified by providing two WWW interfaces,
  HTTP://WWW.ASTRO.UU.SE/ $^\sim$VALD
  HTTP://WWW.ASTRO.UNIVIE.AC.AT/ $^\sim$VALD
  and turn around time has been reduced by implementing two mirror sites
  VALD@ASTRO.UU.SE
  VALD@HYPATIA.GSFC.NASA.GOV
  in Uppsala and at NASA-GSFC which are synchronized every 12 hours with the main site
  VALD@ASTRO.UNIVIE.AC.AT
  The Vienna site also continues to take responsibility for registration (at VALDADM@ASTRO.UNIVIE.AC.AT).
• Flexibility of data extraction has been increased by providing personal configuration files which allow optimization of extractions for specific scientific requirements of each user. In addition, usage of personal configuration files will ensure identical outputs even for future versions of the data base and new default settings for data extraction chosen by the VALD team.

Extensive plans have been made for the future development of VALD (VALD-3) and we already started working with the prototype software and test datasets. The main goals are to extend the applicability of VALD to stars with lower effective temperatures and to include additional line data such as Stark broadening and wavelength shift parameters which can be crucial e.g. for proper synthesis of Helium lines. A similar approach will be used for hyperfine structure and isotopic splittings as well as for Zeeman patterns (when the LS coupling approximation is not applicable).

A major new step towards VALD-3 is the inclusion of molecules. Recent progress in molecular spectroscopy resulted in line lists of diatomic molecules that can be compared with high resolution astronomical spectroscopic observations on a line-by-line basis (Valenti et al. 1998). VALD-3 will in the beginning contain a few best lists of diatomic molecules that can be reliably used for spectrum synthesis. The new internal and the extracted data format provide the capability to accommodate this new data while a molecular equilibrium solver has been developed for the selection procedures.

Last, but not least we call on the community of users to provide us with critical comments on what should be changed and suggestions on what should be included in the next version of VALD. In particular, we are grateful for any references to data we have not yet been aware.

Acknowledgements

We are very grateful to C. Cowley and D. Bord (University of Michigan), to S. Bergeson, K. Mullman, and J. Lawler (University of Wisconsin) for sending us their data before publication, as well as for very useful discussions and comments on the transition probabilities. We thank V. Pavlova, E. Davydova, and A. Pavlov for their help in preparing the input line lists. This research was supported by the Fonds zur Förderung der wissenschaftlichen Forschung (projects S 7303-AMS and P11882-PHY) and by the Swedish Natural Science Research Council (Naturvetenskapliga Forskningsradet) grant F-AA/FU 11680-300. TR received partial financial support through grant 98-02-27050 of the Russian Federal program "Astronomy''.