The positions contained in the Hipparcos Input Catalogue are derived from a wide variety of sources. The correspondence between the letters used in field 9 and the source of the positions, along with the number of stars whose position was taken from the source, is given in Table B1 .
As far as possible, all positions and proper motions were initially reduced to the FK4 system for equinox and epoch B1950.0. Subsequently, all data were transformed to the FK5 system and to equinox and epoch J2000.0, i.e. the old standard was replaced by the IAU (1976) system, and the systematic corrections FK5-FK4 were applied as given in Fricke et al. (1988). A recent discussion of such transformations can be found in Smith et al. (1989).
The above procedure applies also to the new meridian circle measurements, which were performed differentially with respect to the FK5 system for equinox J2000.0. However, for the basic FK5 (Fricke et al. 1988) and the FK5 extension (Fricke et al. 1991) stars, the J2000.0 data were taken directly from the respective catalogues. Their B1950.0 data (fields 16-19) were then obtained by transforming the J2000.0 FK5 data to equinox and epoch B1950.0 and eliminating all systematic corrections applied in the transition from FK4 to FK5. Therefore, the B1950.0 data of the basic FK5 and FK4 Supplement stars are not identical to those given in the FK4 and FK4 Supplement catalogues since the individual corrections FK5-FK4 are still included.
The data given in fields 4-9 are the result of extensive compilations of previously-available data (Bastian & Lederle 1985; Jahreiß 1988, 1989; Jahreiß et al. 1992), and new observations and measurements (Réquième 1982, 1985, 1986, 1988a, 1988b, 1989; Muiños et al. 1988; Le Poole et al. 1988; Morrison & Gibbs 1985; Turon & Réquième 1984) performed by the INCA Consortium. The new observations were necessary since many of the stars of high astrophysical interest were not contained in the major astrometric catalogues.
The compilation of existing `astrometric data', i.e. position and proper motion data of good quality, was performed at the Astronomisches Rechen-Institut at Heidelberg, and collected within the CDA (Catalogue des Données Astrométriques, Bastian & Lederle 1985; Jahreiß 1988, 1989). Data were taken from the FK5, FK5 extension, FK4 Sup, PPM, AGK3RN, SRS, NPZT74, N30, 2CP50, GC, AGK3, SSSC, CPC2, SAO, CPC (the given order corresponding roughly to decreasing order of positional accuracy). FK5 data were not used for the faint stars of the FK5 Extension for schedule reasons.
In addition to this compilation work, and once the list of stars requiring new positional data was established, extensive campaigns of new observations and measurements were initiated and organised according to the capabilities of the European institutes involved in the INCA Consortium. From south to north, in equatorial coordinates, contributions were made as follows:
In addition, 950 SAO stars with poor positions in the zone were observed with the Carlsberg Automatic Meridian Circle.
All collected data, compiled and newly obtained from ground-based observations or measurements, were then processed in a global and homogeneous way (Jahreiß 1989). The selection of the best data for each star was made following their expected positional accuracy at epoch 1990, the adopted mean observing epoch of Hipparcos. This choice strongly depends on the errors of the available proper motions. Only for a very small number of catalogues (FK5, PPM, AGK3RN, GC, and the new meridian circle observations) are the positional error estimates given individually for each star. For other catalogues and for plate measurements, the central epoch (when individual epochs are not available), the mean error at that epoch, and the proper motion error (when a proper motion is available) had to be evaluated. For regions south of -17.5, plate measurements were made for all proposed stars, yielding a significant overlap with the CDA. This allowed further comparisons and checks between the different sources of data. At the end of the campaigns of new observations and measurements, a few hundred stars remained without reliable positions. Positions (and in a few cases magnitudes) were taken from the Guide Star Catalog (Lasker et al. 1990).
In parallel with the global observing and measuring campaigns, specific measurement programmes were devoted to stars in galactic open clusters, for which careful attention had to be given to the problems of star identification, and to stars in double and multiple systems. Stars in galactic open clusters were measured on plates taken with an astrograph at Hamburg Observatory for the northern hemisphere, and at the Münster University Astronomy Department for the southern hemisphere (Tucholke 1988, 1989). An additional verification was made for stars south of by the comparison of the identifications and positions obtained respectively on astrograph and on Schmidt plates (from measurements made at Bordeaux).
For stars in double and multiple systems, the situation was
more problematical than for single stars for two main reasons. First,
it was not straightforward to cross-identify their components with the
stars in large astrometric catalogues. Second, in many cases
discrepancies exist between the individual positions given in these
catalogues for each component, and the relative positions given in
catalogues of double and multiple systems. As a consequence, an
extensive programme of compilation and analysis of the existing data
was required, before defining the necessary measuring programmes
(Dommanget 1985, 1988, 1989). The work was organised by the team of the
Observatoire Royal de Belgique at Brussels, and the collected and newly
obtained data were incorporated into the CCDM (Catalogue of Components
of Double and Multiple stars, Dommanget et al., in preparation). The
construction of this catalogue was undertaken in Brussels starting from
1981 (Dommanget 1983) and subsequently improved with newly compiled and
observed data. CCDM positions are the result of a compilation from a
large number of other sources: CDA positions were used when there
was no doubt about component identification. Many other sources were
compiled and new measurements obtained within the framework of the
Double and Multiple Star Working Group of the INCA Consortium. This
work has been presented in detail by Dommanget (1989), and is also
summarised in the introduction to Annex 1. Depending on the
characteristics of the stellar system (separation and difference of
magnitude between components) the positional and proper motion data
were processed differently: systems with separation smaller than
3 arcsec, and the bright components of systems where the expected
Hipparcos signal perturbation due to the presence of secondaries was
negligible, were processed as single stars. In most other cases, all
data were taken from a preliminary version of the CCDM. A few hundred
systems with very poor positions were processed individually, in a
collaboration between Meudon and Brussels using new ground-based
measurements, meridian circle observations and, in some cases, the