5.1. Physical considerations
For objects without dominant interstellar contamination, one of the results of this paper is that at least on one occasion, for GI Tau/GK Tau, the components' axes are not parallel. On the other hand, we find that 2 other wide binaries show polarization vectors of their components that are aligned: FY Tau/FZ Tau, V773 Tau/FM Tau. There is another pair, DI Tau/DH Tau for which we lack information on the secondary but where the primary most likely has an intrinsic polarization. The relative orientations discussed above are projections in the plane of the sky. The other angle needed to have the complete 3-dimensional orientation, the inclination i, is available only for a very limited number of sources. We have used rotation data from Bouvier et al. (1986), Hartmann et al. (1986) and Vrba et al. (1989) to compute this parameter. However, measuring the stellar luminosity of a PMS star is a difficult challenge, and subsequent computations leading to the value can induce even more uncertainties. If we use the recent stellar data from Kenyon and Hartmann (1995), we find that GI Tau and GK Tau have inclinations of and respectively. These values could even be considered compatible (say ), given the uncertainties. However, these two stars have very different polarization position angles, suggesting that rotation axes are not parallel in space. We also find that HP Tau and HP Tau/G2 have similar inclination angles. Since their polarizations are parallel it is tempting to conclude that these stars have indeed parallel rotation axes. However, their polarization is likely contaminated by interstellar (or intracluster) polarizations. Careful observations at many wavelengths would help extract the intrinsic part of the polarization and provide confirmation for this finding.
It is difficult to compare our results with those obtained for main sequence stars by Weis (1974) and Hale (1994). Two out of the four pre-main sequence pairs for which the polarization method is applicable have parallel axes. We cannot draw conclusions, but we note that these PMS and MS distributions are compatible, with the warning that strictly speaking, we do not have access to the coplanarity information for our PMS stars, since the orbital elements are unknown.
Concerning the binaries found with parallel axes at large separations, we favour the hypothesis that this orientation reflects their initial formation conditions, rather than resulting from evolutionary effects. On the other hand, we also find that there exist at least one system where the components' axes are not parallel. This is a very exciting result, as most of the current fragmentation models predict parallel axes if not coplanar systems (e.g., Bonnell et al. 1992).
Of interest are also our results on the respective determination of the spectra of both components in our close PMS binaries sample. If we distinguish between CTTS and WTTS according to whether their equivalent width is more or less than Å (see Table 4), we find that if one of the binary component is a classical TTS, so is the other when the separation is small. There is no mixed pair (CTTS+WTTS) in our close binary sample, to the possible exception of UX Tau where the primary can be classified as a CTTS according to its flux. These results confirm the trend already observed by Prato and Simon (1997). On the other hand, not every primary is the more active component, i.e., presents the larger flux. In Haro 6-37 for instance, the primary is approximately 3 times brighter than the secondary and the latter has a 10 times larger equivalent width than the former so that its flux is more than 3 times larger.
The approach we proposed here to get insights into the process of star formation and the geometry of the collapse seems promising in view of our preliminary results. Hopefully, campaigns to study larger samples will be carried out in the near future.
5.2. Technical considerations
The scope of our study on close binaries was limited by our inability to obtain high accuracy polarimetry with a simple optical polaroid-sheet imaging polarimeter. With this class of polarimeters, we showed that high S/N ratio images are not sufficient to reach an accuracy limited by photon noise and therefore to guarantee high accuracy polarimetry. From a sample of 122 stars, Ménard & Bastien (1992) calculated an average optical integrated polarization of % for T Tauri stars located over the whole sky. In Taurus, the average is 1.6% for CTTS and 0.7% for WTTS (Ménard et al. 1998). Current single beam polarimetric imagers provide of the order of ; this is unsufficient and more accurate instruments are needed to study the bulk of this binary sample.
The dominant limiting factors are the photometric fluctuations (i.e., atmospheric transparency fluctuations, and to a lesser extent seeing variations) of the atmosphere on short time scales, between each exposures. These variations lead to different intensities that propagate into large polarizations and/or large errors on the measurements. In practice, because of the atmosphere, errors on the polarization (P)% are difficult to achieve without extra careful monitoring of the atmospheric transparency. Furthermore, the photometric data reduction process itself limits the accuracy to (P)% in the optical, mainly because of limitations on (polarization dependent) flat field accuracies and of the error propagation in the combination of the many different exposures needed to extract the polarization. This number should be increased when an infrared detector is used.
Better polarimetric accuracies could be obtained with dual-beam polarimeters measuring the intensities in two orthogonal directions simultaneously. Then the two beams are affected by the same atmospheric fluctuations and the measured normalized Stokes parameter, given by , is free from photometric variations, yielding higher accuracies under most observing conditions. Accuracies (P)% should routinely be achieved.
© European Southern Observatory (ESO) 1998
Online publication: September 30, 1998