2.1. Proper motion catalogues
Astrometric data with the highest currently available accuracy is provided in the Hipparcos Catalogue (ESA 1997). It contains about 120 000 stars and the typical error of the proper motions is about 1 mas yr-1. The Hipparcos Catalogue was the major output of the ESA Hipparcos space astrometry satellite mission, and proper motions were determined by fitting all astrometric parameters (positions, proper motions, parallax) simultaneously to the data points collected over the about 3 years time of operation for every individual star.
Proper motions in the ACT Reference Catalogue (Urban et al. 1997) and STARNET (Röser 1996) however were determined by comparing the positions of stars with an epoch difference of about 80 years. Both catalogues use the Astrographic Catalogue (AC) with a mean epoch of 1907 as the first position measurement. For the ACT the Tycho Catalogue (ESA 1997) provides the second epoch, yielding proper motions for about 1 million stars with an accuracy of about 3 mas yr-1. STARNET uses the Guide Star Catalogue (GSC 1.2) as second epoch and provides proper motions for 4.3 million stars with an accuracy of about 5 mas yr-1. Thus it is the most extensive proper motion catalogue available so far, containing stars with magnitudes up to 12 mag and mean errors still acceptable for kinematic studies.
The proper motions discussed in the following sections are taken from these three catalogues, which are all on the ICRS astrometric system defined by Hipparcos. The PPM proper motions, with a similar accuracy as the ACT for 400 000 stars, were used for comparison only, see Sect. 2.3.
2.2. The samples
In Table 1 we list all the stars which were known or suspected to be connected to the Chamaeleon association before the ROSAT mission, along with their proper motions. Altogether these are 14 stars: the 2 well-known late B type stars HD 97048 and HD 97300, with entries from Hipparcos, and 6 classical and 6 weak-line T Tauri stars. We have included T Cha, although it maybe located foreground to the Chamaeleon clouds as indicated by its Hipparcos parallax (Wichmann et al. 1998). Only one of these stars (BF Cha) is associated with the Cha II cloud, whereas the other stars are located close to Cha I . There is a third cloud in the Chamaeleon region termed Cha III which apparently also shows star formation activity (Pfau et al. 1996), but the sources are on average 2 mag fainter than in the other two clouds and none could be identified in STARNET.
Stars which could be identified either in the Hipparcos (HIP), ACT (A) or STARNET (S) catalogue and which were known to be associated with the Cha I or the Cha II cloud before the ROSAT mission. Additional designations for the stars are given in the last column. Please note that the mean errors of the proper motions in right ascension are given with the factor .
As in other nearby star forming regions, optical follow-up observations of X-ray sources discovered by ROSAT led to the identification of 77 probable new pre-main sequence stars in the Chamaeleon region (Alcalá et al. 1995, 1997). These new T Tauri stars are not only located close to the clouds like most of the T Tauri stars known before, but also up to 10o away from any known site of star formation.
Precise determinations of the lithium line (670.7 nm) strength by means of high resolution spectroscopy and comparison with the typical lithium equivalent width of young main sequence stars (like the Pleiades) of the same spectral type confirmed the pre-main sequence nature for more than half of these stars (C97). We could identify 31 stars of the total sample in the Hipparcos, ACT and STARNET catalogues (Table 2).
Table 2. Stars with proper motions from the ROSAT sample investigated by Alcalá et al. (1995, 1997) and C97. The data are again taken from the Hipparcos, ACT and STARNET catalogues. The classification in T Tauri stars and ZAMS and other stars is based on the lithium criterium as applied by C97. Note however that some of the stars classified as ZAMS stars or stars of unknown nature fall well above the main sequence when placing them in the HR diagram with the help of the Hipparcos parallax (Neuhäuser & Brandner 1998).
Unfortunately, of the 31 stars newly discovered with ROSAT for which we can find proper motions only 8 are confirmed low-mass PMS stars, whereas in the whole sample of C97 the fraction of bona-fide PMS to non-PMS stars is about twice (40 out of 81). This is a consequence of the fact that most of the confirmed low-mass PMS stars in the C97 sample having spectral types later than G5 are normally fainter than about 11.5 mag and hence are not included in the Hipparcos, ACT and/or STARNET catalogues, while the other objects classified as ZAMS stars or with dubious PMS nature by C97 have on average earlier spectral types and hence are sufficiently bright to be present in the aforementioned catalogues.
Similarly, only sources detected with the ROSAT All-Sky Survey and none of the sources detected only in ROSAT PSPC pointed observations could be identified in any of the proper motion catalogues. This means that in our sample there is no artificial spatial clustering due to possibly locally varying sensitivities present within the region indicated in Fig. 2.
2.3. Discordant proper motions
For the majority of the stars in Tables 1 and 2 more than one proper motion measurement is available, so that we are able to compare its values in different catalogues. For most of our stars we find no significant differences and we list the most accurate determination.
Stars with discordant proper motions in two or more catalogues are listed together with all available proper motion determinations in Table 3. The most probable reason for differences in the proper motions are non-resolved binary or multiple systems: in general it is not clear whether the photocentre or the brighter component was observed, and sometimes this may be different for the positions of the first and second epoch, especially for variable stars. This may lead to spurious proper motions in the ACT and STARNET catalogues.
Table 3. Stars with discordant proper motions in one or more catalogues. All proper motions have been transformed to the astrometric reference system defined by Hipparcos, so no systematic differences should be present.
Orbital motion further complicates the determination of the mean proper motion for the whole system. The largest effect is expected for the Hipparcos proper motions, since 3 years of data collection covers only a short fraction of the orbits of long period binaries. Thus the instantaneous motion of the photocentre seen by Hipparcos does not reflect the mean motion of the centre of mass for these kind of systems (Lindegren et al. 1997; Wielen 1997).
Two stars of Table 3 are present in the Double and Multiple Systems Annex of the Hipparcos Catalogue, where the observational effects of duplicity have been taken into account. Sz 19 is perhaps an astrometric binary with a short period which could not be resolved by Hipparcos. Indeed Schwartz (1977) notes a close companion to Sz 19 in the south, confirmed by Reipurth & Zinnecker (1993) and Ghez et al. (1997), and the secondary is variable with magnitude differences of at least 2.5 mag (Brandner 1992). For RXJ 1125.8-8456 a non-linear model of the motion including acceleration terms was fitted to the Hipparcos observations, which has no meaning outside the mission interval. Although this is formally a single-star solution, we may deal with an unresolved system with a period in the range 10-100 years (Lindegren et al. 1997).
Similar effects may be responsible for the inconsistencies in the proper motions of the other two stars from Table 3: RXJ 0837.0-7856 was not observed by Hipparcos, but is flagged as dubious astrometric reference star in the Tycho catalogue, as is RXJ 1159.7-7601. The latter star is additionally flagged as 'perhaps non-single' in the ACT catalogue.
Another source of errors in the ACT and STARNET catalogues may be wrong identifications of stars, favoured by large epoch differences and large proper motions. All these effects may explain the large differences between proper motions in different catalogues, although in most of the cases the actual error source is difficult to find out.
© European Southern Observatory (ESO) 1998
Online publication: September 14, 1998