6 Summary and discussion

Both binary systems under investigation were found to have active secondaries. In UX For, the absorption spectrum of the secondary was detected and a regular double-lined spectroscopic orbit computed. In the case of AG Dor, we found no photospheric lines from the secondary, but detected its strong H$\alpha$ emission and used it to determine a double-lined orbit. Both secondary stars are later in spectral type than their respective primaries, G6+K0.5 for UX For and K0+K4 for AG Dor, which makes them appear relatively more active due to their deeper convective envelopes. Both stars in both binary systems rotate synchronously to their respective orbital motion.

Our spectrum-disentangling technique as well as the Hipparcos parallax suggest that all components are on the main sequence or at least very close to it. Different approaches to determine absolute stellar parameters yielded radii, temperatures, and luminosities that are formally in agreement with each other, but yield a slightly inconsistent inclination for AG Dor with respect to our Doppler imaging analysis. Taking the quite reliable orbital elements (primary-lines velocity residual of 410 ms^-1), we may estimate masses to within 4%, mainly depending upon inclination. An inclination of 55$^\circ$ $\pm$ 10$^\circ$ is suggested from Doppler imaging and would give $\approx$dG0+dK5 for AG Dor. On the other hand, an early K primary - expected from photometric colours - requires an inclination of between 65$^\circ$ to 70$^\circ$, thus being just marginally consistent with the inclination suggested from Doppler imaging. Geometrical constraints involving $P_{\rm rot}$, $v\sin i$ and i also favour an early or mid-G type main-sequence primary, while Doppler imaging, spectrum synthesis, broad-band colours and the Hipparcos distance emphasize a K1$\pm$1 primary and a K4 or K5 secondary. The discrepancy possibly indicates a slightly off-main-sequence evolutionary stage for the primary. The low lithium abundance of approximately half of the solar abundance likely excludes a pre-main-sequence status. H$\alpha$-line variations are detected in all four stars but none of them appeared periodic. The variations can be as large as a factor of two in equivalent width for the UX For primary but were below 30-40% for the AG Dor primary. On the other hand, the secondary star of UX For appeared more constant than the secondary in AG Dor, a maximum of 50% variation compared to 100% for the AG Dor secondary, but their equivalent widths are also more uncertain than those of the primaries. The residual H$\alpha$-line profiles of the two UX For components normalized to the spectrum with the weakest emission equivalent width show a transient emission during quadrature at the relative zero-velocity wavelength, i.e. near the inner Lagrangian point. Since the emission is only detected in three of our 40 spectra and is only observed at one particular epoch (1996.88), we can not draw any firm conclusions about it. However, if real, we suspect that it is caused by neutral hydrogen expelled from both active stars and stored in the inner Lagrangian point, where it is heated by the stellar coronal X-ray and UV emission.

Despite the fact that the two UX For components show more rapid rotation than AG Dor, our data allowed just the primary of AG Dor to be Doppler imaged. AG Dor shows a single-lined absorption spectrum and thus less blending. Its spectra have, at the same time, a S/N ratio higher by a factor of two compared to UX For. Phase smearing due to the orbital motion during our relatively long integration times is also prohibitive for Doppler imaging of UX For, but no significant problem for AG Dor. However, mapping AG Dor still presents a challenge because only one photospheric line is available (Fe I 6546 Å) and the stellar rotational velocity of $v\sin i=18$ kms^-1 is at the limit of current mapping techniques, given the instrumental resolution of 50000. Therefore, we included several tests to ensure that we minimized surface artifacts down to the level comparable with the intrinsic quality of the data. A moderately sized polar spot along with four, possibly five, low-to-medium latitude spots are reconstructed from Fe I 6546 Å. The polar feature is quite asymmetric and has one, possibly two, appendages connecting it with lower-latitude spots. This seems to be a commonly observed feature in spotted stars of various masses and ages (post and pre-main sequence). Amado et al. (1999) used multi-colour broad-band photometry of AG Dor to reconstruct a spot map for winter 1996/97, and found a spot distribution very similar to ours. Modelling optical and infrared colour curves, they obtained effective temperatures as well as spot locations and sizes and found two low-latitude spots as well as a non-modulating component. They reconstructed it with a cool polar spot, which is also supported by their TiO molecular-band measurements. We may also note that the Doppler image of AG Dor appears quite similar to the Doppler image of the single, young, solar-type star EKDraconis (Strassmeier & Rice 1998), as well as to the pre-main-sequence binary V824 Ara (Hatzes & Kürster 1999; Strassmeier & Rice 2000). It has been discussed whether polar features are connected to rapid rotation (e.g. Vogt et al. 1999). AG Dor and EK Dra are the two slowest rotators Doppler imaged so far (both are main-sequence stars and possibly not directly comparable to the active giants in RS CVn binaries), and both of them show a polar feature whose center is displaced from the rotational axis along with several low-latitude or equatorial spots. This could be evidence that the surfacing latitudes of magnetic flux tubes coming from the bottom of the convection zone are being rapidly redistributed by surface velocity fields such as differential rotation, meridional flows, and sporadic plasma rearrangements triggered by magnetic reconnections. Several recent papers presented evidence for the existence of such surface velocity fields on spotted stars (see the review by Collier Cameron 2000; and also Weber & Strassmeier 2000 for a most recent application). Binarity per se seems to play just a secondary role despite of these star's additional angular momentum due to spin-orbit coupling. For a discussion of the impact of tidal forces on the magnetic flux tubes in a close binary system, see Holzwarth & Schüssler (2000).

Finally, we note that there seems to be no apparent correlation between the H$\alpha$ variations on both binary components of AG Dor and our Doppler map of the primary.

Acknowledgements

We are very grateful to the Austrian Fonds zur Förderung der wissenschaftlichen Forschung for support under grant S7302-AST and S7301-AST and to the Deutsche Forschungsgemeinschaft for grant HU 532/8-1. We also thank M. Kürster (then MPE, now ESO) for his kind help with the remote CAT observations from Garching and G. Cutispoto for recalculating the colour indices of AG Dor. Special thanks go to Michael Weber and János Bartus for their personal support and computer assistance.