The spectra of the 5 sources observed are presented in Figs. 2 and 3 (see captions for details). In every source, the primary (defined as the brightest component in V) is referred to as A, and the secondary as B. Out of the 5 pairs, 3 show similar spectral types for both components: DK Tau, HK Tau and Haro 6-37 (see Fig. 2). In HN Tau and UX Tau, the primary appears significantly hotter than the secondary (see Fig. 3).
For detected emission lines, the equivalent widths are given in Table 4, together with the spectral types and the corresponding effective temperature, from Cohen & Kuhi (1979). Errors on the equivalent widths are typically 5%, except for the [OI]Å line in HN Tau B, where it reaches 30%. Errors on the spectral types are 0.5 subclass for types later than M0, and 1 subclass for the others. They translate into errors of 0.01-0.02 dex on , from M to mid-K, to which 0.02 dex should be added due to a systematic overestimate caused by the choice of luminosity class (IV for PMS stars, instead of V tabulated by Cohen & Kuhi).
4.2. Polarimetry of the wide binaries
The polarization data were collected from the literature. They are reported in Table 5. The data are from Tamura & Sato (1989) for V773/FM Tau, from Moneti et al. (1984) for V807/GH Tau, and from Ménard et al. (1998) for all others. We have quoted the value of the polarization level P for each component of the binary, together with the uncertainty . When the ratio was larger than 3, we have also quoted the polarization position angle and its uncertainty .
Polarization data for the wide binaries. The results are listed two by two in the same order as in Table 2 (to the exception of the triple star HP/G2/G3) with the primary component listed first. Moneti et al. (1984) did not provide uncertainties for their V807/GH Tau polarization measurements.
The separations range between and , or AU and AU, assuming a distance of 140pc for Taurus. We consider the HP Tau group as two binaries, HP Tau/HP Tau G2 and HP Tau G3/HP Tau G2. The separations are given in that order in Table 2.
The goal of this section is to check whether the orientations of wide binary components are similar or not. Two conditions are necessary: the polarization must be detected, and it must be intrinsic to the objects, i.e., not of interstellar origin. We argued in Sect. 2.1 that CTTS/CTTS pairs, viewed with a large inclination are the best candidates for that study.
If the polarization were of interstellar origin, both components would appear with parallel polarizations because in clouds like Taurus, the interstellar polarization changes on scales much larger than the separations considered here. A few other clouds however, might have more than one polarization component with different interstellar polarization position angle for a given line-of-sight. This may lead to difficulties in evaluating the interstellar polarization. This is not the case for our data in Taurus. We have used the data of Goodman et al. (1990) and Tamura & Sato (1989) to evaluate the position angle of the interstellar polarization at the location of the wide binaries and compare it to the stars' polarization position angle.
The results are presented in Fig. 4 and show a smooth correlation, with many objects having a measured polarization similar to the interstellar one to better than . Among these objects, many are part of the same pair, indicating that the interstellar medium plays a role, at least partly, in this apparent alignment.
Fig. 5a presents the results of a search for a correlation between the polarization position angles of each component. The positive result in all cases but one, suggests that the stellar symmetry axes are mostly parallel in the binaries we considered. In the following, we check each individual result against the interstellar contamination.
The CTTS binary V807 Tau/GH Tau (HBC 404 /55) appears in the bins of both Fig. 5a and Fig. 4 and its polarization is low. We conclude that this apparent correlation is dominated by interstellar polarization. This is also the case for the WTTS binary HBC 358/359 , as expected for WTTS stars. Although we cannot check the interstellar polarization very close to the other WTTS binary of the sample HBC 352 /353, we believe its apparent parallel orientation is also dominated by interstellar polarization.
The HP Tau group (HBC 66/415 /414) contains stars that are highly polarized but show only a moderate polarization position angle difference with the local interstellar value. Moreover, the two companions (G2 and G3) are WTTS stars. This group is probably affected by interstellar polarization, although it is at the upper end of the values measured in Taurus.
The pair FY/FZ Tau (HBC 401/402 , both CTTS) is highly polarized at angles similar for both objects but otherwise very different from the interstellar polarization value. Similarly, the CTTS pair GI/GK Tau (HBC 56 /57) exhibits polarization position angles that are very different for both components, ruling out an interstellar origin, at least for GI Tau.
V773/FM Tau, (HBC 367 /23) have parallel polarizations within the error bars. This orientation is perpendicular to the local interstellar value. This object is peculiar because the only published simultaneous polarization measurements were obtained in the infrared, at m. The difference is not an optical depth effect (because of the longer wavelength) because polarization position angles measured in the optical are similar. However, the polarization position angle of V773 Tau is extremely variable (Ménard and Bastien, 1992). It rules out an interstellar origin but makes it difficult to assess the orientation reliably. We have used the simultaneous m results in our histograms of Figs. 5 and 4.
Finally, we also searched for a link between the polarization position angle of each member and the apparent position angle of the binary, (Fig. 5b). No correlation is found between the primary's polarization position angle and the apparent binary position angle. It suggests that the polarization is probably not circumbinary in origin. However, the lack of a correlation does not mean that the disks are not coplanar with the orbital plane. In particular, the position angle of the binary may not necessary be aligned with the position angle of a possible circumbinary envelope.
4.3. Polarimetric imaging on close binaries
The results show that HN Tau A is significantly polarized: 2.7% +/- 0.4%. This is in agreement with measurements by Ménard et al. (1998) and Tamura & Sato (1989). DK Tau B is also marginally detected.
However, no significant detection of the polarization was made on any other object. This appears surprising in view of our simulations in Sect. 2.1 and the large S/N ratios effectively reached in the observations. The S/N for the primaries were , they were for the secondaries. We would have expected reliable estimates of the polarizations from these numbers. However, the S/N reached for the reference stars are below the criterion given in Sect. 2.2 (). For such reference star S/N values, our simulations in Table 1 show that the polarization is hardly measured with a precision better than .
Moreover, these results raise questions regarding the ability of rotating polaroid sheet/wire grid imaging polarimeters to perform high accuracy polarimetry. We suspect non-photometric conditions to be partly responsible for the large error bars and the lack of detections, as a photometric error on one of the 3 frames at , and translates into a slightly larger polarization error on P. From the photometry point of view, this is not a large error, but from the polarimetry point of view, this is a large error. Suggestions for improvement will be presented in Sect. 5.2.
Since we are interested in the relative orientation of the components, we can go further and try to calculate relative polarizations, which are easier to measure precisely. We have applied the same direct method to compute the relative polarization of the secondary in every pair, using the high S/N primary as a reference, and arbitrarily setting its polarization to zero. The corresponding results are presented in Table 6 and should be interpreted as follows. When significant relative polarization is "detected", it means that the polarization of the secondary is different than the primary's, either in level, or in angle.
Although the S/N ratio is better, it still falls short of the accuracy we should have reached. HN Tau remains the only system where the polarizations of the two components are clearly different. In all other cases, the relative "detection" of the secondary against the primary is not significant at a level.
© European Southern Observatory (ESO) 1998
Online publication: September 30, 1998