1 Introduction

The mass is the most fundamental parameter of a star, because, to a large extent, it determines its structure and evolution. Therefore, it is a major problem for our understanding of pre-main sequence (PMS) evolution that at this time there are no reliable empirical mass determinations for young stars with $M < 1~M_{\odot}$. Mass estimates for T Tauri stars are usually based on comparisons of their position in the Hertzsprung-Russell diagram (HRD) with theoretical PMS evolutionary models, which means that they are affected by the (unknown) uncertainties within these models. Moreover, it is not possible to rate the quality of different sets of PMS models by comparison to observational data.

For this reason, empirical mass determinations for young stars are highly desirable. Binary stars offer a unique possibility to do this, because the system mass is known as soon as the orbital parameters are determined. There are many binaries among T Tauri stars in nearby star-forming regions (SFRs). Most of them have been detected during the last decade by high-angular resolution surveys in the near infrared (NIR) (for an overview of this topic, see the review by Mathieu et al. (2000) and references therein).

The first reliable empirical masses of PMS stars were given by Casey et al. (1998) for the components of the eclipsing double-lined spectroscopic binary (ESB2) TYCrA. These masses are $M_1 = 3.16\ \pm\ 0.02~M_{\odot}$ and $M_2 = 1.64\ \pm\ 0.01~M_{\odot}$. The secondary mass is consistent with the predictions of PMS models from D'Antona & Mazzitelli (1994) and also Swenson et al. (1994). The primary is already close to the main sequence. The lowest-mass PMS stars with empirically determined masses thus far known are the components of RXJ0529.4+0041. For this ESB2, Covino et al. (2000) determined the masses $M_1 = 1.25 \pm 0.05~M_{\odot}$ and $M_2 = 0.91 \pm 0.05~M_{\odot}$. They concluded that these masses are in good agreement with with the Baraffe et al. (1998) and Swenson et al. (1994) models, but less consistent with sets of PMS tracks from D'Antona & Mazzitelli (1994) and Palla & Stahler(1999). Because of the relatively high masses, these results cannot be used to check the PMS models for K- or M-dwarfs and objects with masses below the hydrogen burning mass limit at $\approx$ $0.075~M_{\odot}$.

In this paper we will follow the approach of Ghez et al. (1995, herafter G95). Using NIR speckle interferometry, they obtained repeated measurements for the relative astrometry of the components in 20 T Tauri binary systems. In this way they showed that in most of these systems, orbital motion can be determined. From short pieces of orbital data and the statistical distribution of orbital parameters, they have derived the average system mass of $1.7~M_{\odot}$, which is in the order of magnitude expected for the mass of two T Tauri stars.

We present similar data for 34 T Tauri binary systems and in this way increase the object list and also the observational time base. An overview of our observations and data reduction is given in Sect. 2. The results are presented in Sect. 3, discussed in Sect. 4 and summarized in Sect. 5.