1996A&A...311..513L

While some Ap variables display a linear polarization variation very similar to that computed for a pure magnetic dipole, several Ap stars show conspicuous peculiarities which must be interpreted in terms of departures from the standard, oblique rotator model (we have shown previously that abundances anomalies are not sufficient to explain odd polarization diagrams). We have designed an inversion method, based on a residues minimization process, which allows us to build the map of the magnetic peculiarities at the surface of non-dipolar stars. As the linear polarization is but weakly sensitive to the variations of the field modulus, we interpret the polarization anomalies in terms of inclination changes of the lines of force within their meridian plane. Keeping the magnetic equator as a plane of symmetry, we show that it is sufficient to assume slightly expanded lines of force, over some parts of the magnetic equator, to explain most peculiar polarization curves (Figs. 2 to 7). Such regions, where the lines of force expand outwards, seem to occur preferentially in the vicinity of the rotation poles for those stars having a β angle not far from 90deg. In the case of β CrB, which was studied previously in detail (Leroy, 1995), this region nearly coincides with the equatorial patch of enhanced field strength, which must be postulated to explain the surface field measurements. The present study, which also provides unambiguous determinations of the i and β angles for 15 stars, marks the provisional end of our investigation based on broadband linear polarization measurements. We expect that similar measurements, having a good spectral resolution, will be available soon: they will yield more severe observational constraints enabling a more detailed modeling work. However, we think that the series of articles which ends with the present paper has demonstrated the great value of linear polarization data and may have opened fruitful research tracks bearing on the magnetic structure of Ap stars.