The photometric database consists of several directories containing the photometric datafiles, the references, the co-ordinates and various useful files, like cross-indexes and application programmes and scripts. The package used is /rdb (Manis et al. 1988), which is a Unix relational database management system. The advantage of /rdb is its simplicity and the possibility to include the database commands within Unix shell programming or command lines by using pipes and input/output redirection because this software uses simple ASCII files with a tabulator as field separator.
We have prepared one file for each photometric system. Each one has the same structure and contains: the Lausanne identificator (LID), which uses the code numbering system designed by Mermilliod (1978), a remark on duplicity (component observed) or variability, the data, and the data source. We have found some 78 photometric systems, by considering that the use of different filters results in a different system. We have restored the full information on the components observed in visual double stars, like AB, BC or ABC. It is however clear that due to the frequent lack of information in the publications it is sometimes difficult to decide what has been really observed. Below 7 to 10 arcsec, we have considered that the double stars have usually not been separated, while above 25 to 30 arcsec, they should be separated. For separation between 10 and 25 arcsec, it is often difficult to decide what has been done, depending on the magnitude difference between the components. In case of doubts no information has been indicated when nothing was specified in the original paper.
Table 1 (click here) gives the number of entries for the photometric systems in which more than 1000 stars have been observed: the system designation, the total number of measurements, the real number of stars and the number of papers where
data were found. For basic references on these photometric systems, please refer to those given by Hauck et al. (1990) paper or on the WWW site described below.
|
No | System |  |  |  |
|
01 | UBV | 169826 | 107996 | 1633 |
| 04 | uvby | 105891 | 66398 | 520 |
| 13 | Geneva | 39900 | 1 | |
| 78 | WBVR | 13588 | 13588 | 1 |
| 11 | Walraven | 12872 | 12115 | 55 |
| 14 | UBV Eggen | 10489 | 10489 | 80 |
| 54 | RI Cousins | 10811 | 9506 | 153 |
| 12 | DDO | 10277 | 8077 | 100 |
| 08 | UBVRI | 9008 | 7397 | 118 |
| 02 | UBV Cape | 8309 | 7352 | 43 |
| 17 | RI Eggen | 7887 | 6777 | 83 |
| 21 | Vilnius | 7559 | 6763 | 81 |
| 16 | uvby Eggen | 6980 | 5861 | 45 |
| 36 | Celescope | 5719 | 1 | |
| 19 | RI Kron | 5427 | 4489 | 45 |
| 09 | IJHKLMN | 3358 | 2353 | 44 |
| 77 | Oja | 2752 | 2752 | 2 |
| 81 | CaII H&K | 1990 | 1990 | 2 |
| 75 | Maitzen  | 1837 | 1780 | 20 |
| 18 | 13-Color | 1911 | 1730 | 8 |
| 61 | Andrews | 1715 | 1715 | 3 |
| 20 | Gyldenkerne | 1453 | 1453 | 5 |
| 45 | Washington | 1445 | 1318 | 30 |
| 06 | UVBGRI | 1704 | 1299 | 26 |
| 63 | Alexander | 1218 | 1218 | 1 |
| 72 | Moffet | 1161 | 1152 | 11 |
The total number of entries is now about 400000, coming from 2526 papers. There are however some redundancies, because the UBV catalogue collects data from systems including UBV filters, like the various UBVRI systems. The corresponding UBV data are usually reduced to Johnson's standard UBV system.
When several observers are measuring stars in one particular system, they generally use differing instruments and techniques for observations and data reduction. This inevitably results in disparities in the published data (systematic errors, different accuracies). The main sources of these disparities are to be found in the photomultipliers used, the filter sets, the method of reduction outside the atmosphere, the choice and use of standard stars and the optical properties of the telescope. We have computed average data for two of the most used systems: the UBV (Mermilliod & Mermilliod 1994) and uvby (Hauck & Mermilliod 1996). The method developed for this purpose is explained in the introduction to the UBV catalogue (Mermilliod & Mermilliod 1994) and its main goal is to attribute a lower weight to discrepant observations. Briefly, a first weighted mean is computed, where the weights are determined by the number of observations in each data source. Then the differences between the mean value and the individual observations are computed. A second mean value is then computed, where the weights are proportional to the inverse of these differences. In this way, discrepant data receive a smaller weight and the average value is centred around the most common values.
The computation of mean photometric data has sometimes been criticised (Manfroid & Heck 1985), but we think that averaged data are very useful for many purposes. As we do not have access to the original measurements, it is of course impossible to re-reduce the data. The processing of the published data is also sometimes problematic in the absence of stars in common with other lists or because the list of the observed standards has not been published. We therefore think that the method used is the best possible in the present circumstances. In any case, the data for each source are fully available through the WWW server and anybody can retrieve them and decide which data he or she prefers to use.
Following an agreement with Geneva Observatory, only the mean values of the Geneva 7-colour system are stored in the database and accessible through the public WWW server described below.
As a rule all astronomical journals are surveyed for new photometric data. The bibliographic references (data sources) are also collected in individual files, one for each system. The reference numbers are generally attributed sequentially in each system. The information entered in the files is the source number, the list of authors' names, the journal information (year, name, volume, page), the complete title and the ADS/CDS bibliographic code to enter the ADS Abstract Service system.
As described in Hauck et al. (1990), we have built a Master Index, which indicates in which systems each star has been observed. This is important for locating more easily the available information on each star. The new Index is different because it does not contain the photometry code as before, but the number of independent data sources in each photometry. The corresponding fields are left blank if no data at all exist for some photometric systems.
According to the relational database structure adopted, the Master Index contains the Lausanne identificator (LID), the remarks on duplicity or variability, and the number of data sources in each photometric system. This file contains more than 207500 stars. The merging of all these data has of course produced a considerable number of problems. Many of them have already been examined, the other ones will be looked at gradually.
The database contains a number of additional files collecting general information for the stars (cross-identifications) and especially co-ordinates. Basic astrophysical catalogues containing non-photometric data have also been transformed into /rdb files to have additional on-line information and the possibility to perform selection on various criteria and later obtain the corresponding photometric data.
One of our main contributions concerns the stellar identifications. We have transformed the published identifications into a uniform system: various names are used in the literature, generally depending on the kind of sample investigated (proper motion stars, luminous stars, nearby stars, and so on) and the most time-consuming work is related to the necessity of collecting all the data for each star under the same identification, otherwise the advantage of working on large compilations partly vanishes. The basic principles adopted for this work have been explained in the description of the code numbering system (Mermilliod 1978) and the solution designed for star clusters is described by Mermilliod (1976).
Coordinates are very important for performing selections on the position in the sky. We have made a big effort to collect co-ordinates from the literature, the HST GSC, or to determine them from photographs of cluster or field areas. In spite of our efforts, several thousands of faint field stars (generally fainter than the limit of the GSC) still lack co-ordinates and we are not in a position to obtain them easily. This may result in the loss of faint stars that could be used, for exemple, to calibrate Schmidt plates or CCD frames.
Adapted software in the form of Unix shell scripts and Perl programmes has been developed to facilitate the handling of new data and database update. An effort will be made to automate the treatment of data files and the detection of errors or discrepancies for all sets of new data before they are introduced in the catalogue files. Graphics facilities are also being implemented to plot photometric diagrams or compare indices from different photometric systems. Eventually existing calibration codes will be implemented on the database to determine the astrophysical parameters of stars.