4 Discussion

In eight out of ten symbiotic stars of the sample, our polarimetric results show different wavelength dependences of the polarization percentage and/or the position angle and temporal variations. Therefore, we can conclude that intrinsic polarization is a common characteristic of this group of stars. The presence of intrinsic polarization in KX TrA and Hen 1761 is not conclusive. We find wavelength dependences of $P(\%)$ decreasing monotonically from blue to infrared; decreasing in opposite sense; peaking in some particular wavelength or presenting a dip in the visual and then increasing to the blue and to the red. Therefore, there is not any particular wavelength dependence of the polarization on the contrary, different trends are observed for each star along the time. Moreover, no conclusive correlation between the behavior of polarization and position angle can be drawn. In the observed objects it is common to find that the same wavelength dependence in the degree of polarization is associated with different orientations in the position angle.

In several systems we can see that the most important changes in the percentage of polarization along the time are observed towards the short wavelengths, whereas the smaller variations or constant values are observed in the I band, which is the cases of Hen 1103, CL Sco, AR Pav and CD  $-43^{\circ }$14304. As it can be seen, Hen 1103 and CL Sco present the particularity that both the polarization degree and the position angle are practically constant in the infrared. The blueward slope of the $P(\%)$ trend is highly variable even if considering the same object (see V748 Cen, CL Sco, AR Pav, RR Tel and CD  $-43^{\circ }$14304).

It is known that the dust richest D-type symbiotics are intrinsically polarized (Schmid 1997). Since dust scattering is considered the main source for intrinsic linear polarization in symbiotic systems, our positive results obtained for seven S-type objects in the sample, would suggest the presence of dust surrounding asymmetrically the system or the cool component of S-type symbiotics which does not show infrared excess. This would be in agreement with results obtained from polarimetric studies of red variable stars: the polarizing particles can be placed near the star while the particles producing the infrared excess are farther from the star (Shawl 1975) or different grains may be involved in producing the polarization and the infrared excess (Coyne & Magalhães 1977).

In this work RR Tel is the only D-type symbiotic star with a Mira variable as cool component. We confirm the intrinsic polarization of this object, especially in the I-band which seems to be associated to the Mira period, corresponding to increasing polarization with decreasing brightness. More polarimetric observations of this object would be particularly necessary to isolate the mechanisms occurring at shorter wavelengths and to confirm the modulation of the infra-red polarization with the Mira period and, to some extent, the binary period. Some symbiotic systems display extended ionized nebulae with different morphological structures (bipolar, elliptical, irregular, jet-like). The polarimetric study of these objects is particularly interesting because it allows one to correlate the orientation of polarization vector with the known nebular geometry and consequently the scattering region can be defined (for example in BI Cru, Harries 1996). Corradi et al. (1999) have presented an optical search for extended nebulae around symbiotic stars and a compilation of objects with nebulae detected at optical and radio wavelengths, as well as a list of optical negative detection. AG Peg is the only star of this work included in their lists of radio and optical detections. As we mentioned, some features of polarization in this object seem to be related with the bipolar nebulae, especially the large rotations of the position angle. However, scattering regions connected to the binary system might be present and more observations along the period would be necessary in order to draw firm conclusions. Hen 1103, KX TrA and RR Tel have no nebula detections according to Corradi et al. (1999); it means that the observed polarization is originated in gas structures close to the stars or inside the binary (stellar photosphere, gaseous stream, accretion disks). There is no information about the existence of extended nebulae in the remaining S-type symbiotics included in our sample, although Corradi et al. (1999) have confirmed less frequent ocurrence of extended nebulae among S-type objects.

As regards to the correlation between the polarization and the orbital phase analysed in the other symbiotic binaries with known periods (V 748 Cen, CL Sco, AR Pav and CD  $-43^{\circ }$14304), the scarse number of observations allows one only to glance at the behavior of the polarization in particular at positions of both components, respect to the line of sight. V 748 Cen shows an increase of polarization when the hot component is moving towards the second quadrature and a decrease before the primary eclipse when the giant is in front of the observer. In this case, like for CH Cyg, the polarization can be due to scattering of the light of the hot source by the M giant envelope and/or from the stream of material between the stars. CD  $-43^{\circ }$14304 shows also an increasing UB polarization before the first quadrature but we have no observations during the second one. On the other hand, CL Sco and AR Pav show an opposite behavior of polarization with phases. The higher values of $P(\%)$ occur when the components are in conjunction whereas we have a minimum in the polarization when the line of sight is perpendicular to the binary axis. Strong rotations in the position angle are observed in both cases.