4 Discussion

4.1 Comparision with theoretical results

We have estimated masses for the components by comparison with theoretical PMS models for a subsample that contains 17 out of our 34 systems. In these cases, we obtained resolved J band photometry. This spectral band is supposed to be least affected by circumstellar excess emission and can thus be taken as an indicator of stellar luminosity. Moreover, in these 17 systems, there are no additional companions known, so the mass sum of the components derived from theoretical models will match the dynamical system masses. We placed the components of these systems into the HRD, estimating the stellar luminosity from the resolved J band magnitudes, assigning the optical spectral type of the system (taken from Kenyon & Hartmann 1995 for Taurus-Auriga and Walter et al. 1994 for Scorpius-Centaurus) to the primary and assuming that all components within one system are coeval. Masses of the components were then derived using the PMS evolutionary tracks of D'Antona & Mazzitelli (1998), Swenson et al. (1994) and Baraffe et al. (1998).

The mean mass obtained for this subsample from the HRD is $0.88~M_{\odot}$ for the D'Antona & Mazzitelli (1998) tracks and $1.17~M_{\odot}$ for the Swenson et al. (1994) tracks. The PMS model from Baraffe et al. (1998) yields a mean mass of $1.28~M_{\odot}$. The uncertainties of the mass estimates that originate from observational data used for placing the components into the HRD are $\approx$ $0.1~M_{\odot}$. The average empirical mass derived from Eq. (11) for that subsample is $1.22\pm 0.50~M_{\odot}$. Within the large formal error, the predictions of all three PMS models match our empirical result. However, our dynamical <M> is much closer to the mean masses derived using the Baraffe et al. (1998) and Swenson et al. (1994) tracks than to the value calculated from the D'Antona & Mazzitelli (1998) PMS model, which seems to underestimate T Tauri star masses. This finding has been recognized by other authors. Bonnell et al. (1998) estimated T Tauri stars' masses from infall velocities of accreted material and also conclude that the empirical masses are generally larger than those predicted by the D'Antona & Mazzitelli (1998) model. A similar result was reported by Simon et al. (2000), who calculated T Tauri stars' masses from Keplerian motion in circumstellar and circumbinary disks.

4.2 Implications for binary statistics

There is significant relative motion in most systems, and this motion is in almost all cases consistent with orbital motion (Sect. 3). As already pointed out by G95, this demonstrates that the large majority of all close companions detected in the multiplicity surveys mentioned in Sect. 2.1 really are gravitationally-bound stars. No binary component discussed here has to be reclassified as a Herbig-Haro object (Sect. 3.2.4), and only 2 out of 34 companions may be chance projected background stars (Sect. 3.2.2).

This finding is particularly important because there is a companion overabundance among T Tauri stars in the SFRs discussed here when compared to main-sequence stars in the solar neighbourhood (Leinert et al. 1993; Ghez et al. 1993; Ghez et al. 1997a; Köhler & Leinert 1998; Köhler et al. 2000). In Taurus-Auriga, almost all T Tauri stars seem to be components of multiple systems. To further confirm this result, Köhler & Leinert (1998) performed stellar counts in the vicinity of their survey objects and concluded that in Taurus-Auriga, statistically 4.3 out of 44 apparent companions are projected background objects. Köhler et al. (2000), in a similar way, derived a number of 7.8 chance projections per 46 companions in Scorpius-Centaurus. We found one candidate for a chance-projected background star out of 21 objects in Taurus-Auriga and one candidate out of 11 objects in Scorpius-Centaurus. The results are not directly comparable because we have only studied the closest pairs for which chance projections are least probable. The percentage of background stars projected by chance among the observed companions is, however, of the same order of magnitude in both studies. Thus, it can be concluded that chance projections do not affect the binary statistics significantly in the SFRs discussed here.

Köhler et al. (2000) excluded six close companions in Scorpius-Centaurus from a restricted sample. Their observed separations are less than the strict diffraction limit $\lambda /D = 0\hbox{$.\!\!^{\prime\prime}$ }13$ of a 3.5m-telescope in the K-band, so they cannot be definitely distinguished from elongated single objects. In three of these cases, namely RXJ1601.8-2445, RXJ1603.9-2031B and RXJ1604.3-2130B we derive a relative velocity that is consistent with orbital motion (see Sect. 3). Thus, we propose to classify these objects as binary systems in further studies of multiplicity in the OB association, Scorpius-Centaurus.