## 1. IntroductionRecent observations (Stenflo & Keller 1996, 1997) have revealed a richly structured polarization spectrum of the Sun, known as the "second solar spectrum", since it bears little resemblance to the ordinary, unpolarized intensity spectrum and thus contains at least in part complementary information. The structuring is due to mixed contributions of similar importance from the continuum and lines. The continuous spectrum gets linearly polarized by radiative scattering, mainly by Rayleigh scattering at neutral hydrogen and by Thomson scattering at free electrons. The polarization in spectral lines is due to coherent scattering in atomic bound-bound transitions and is altered by the ubiquitous magnetic fields. In order to fully understand the various physical processes involved we need to disentangle them. In the present paper we start with the continuous spectrum. Apart from a better understanding of the physics such a study is of great use for constraining solar model atmospheres and for the determination of the zero level of the observed polarization scale. With a solar model atmosphere as input the continuum polarization is obtained by numerically solving the transfer equation for polarized radiation. Different model atmospheres give different degrees of polarization. A comparison with empirical data will therefore enable us to select between several solar model atmospheres. Such observations in continuum windows from 4500 Å to 8000 Å with a sensitivity of in the degree of polarization are planned but not yet available. For the diagnostics of turbulent magnetic fields with the Hanle effect it is necessary to know precisely the true zero level of the polarization scale (Stenflo et al. 1998). The Hanle effect, a coherence phenomenon occurring in coherent scattering in the presence of magnetic fields, leads to a depolarization in the line core. Since the polarization in the lines and the continuum are usually of the same order of magnitude one cannot use the continuum level as a reference for line polarization. The true polarization zero level must be the reference. Due to instrumental effects, the true zero point of the polarization scale is not known with sufficient precision. However, knowing the degree of continuum polarization from theoretical considerations, the zero level in the observations can be determined. In Sect. 2 we will describe the relevant physical theory, the numerical technique, and the solar model atmospheres used. In Sect. 3 two tests of the computer code are presented. In Sect. 4 we gain physical insight into the relevant quantities by illustrating the role of the absorption and scattering coefficients and the temperature gradient. It is of special importance to know the layer of formation of the continuum polarization, since it is often assumed to lie well above the layer of formation of the continuum intensity. We will show that the two layers in fact overlap. Finally, in Sect. 5 a simple analytical expression describing the center-to-limb variation (CLV) of the continuum polarization over the whole visible spectral range is derived and fitted to the theoretical data, providing a convenient approximate representation of the full set of computed polarization values. © European Southern Observatory (ESO) 1999 Online publication: December 16, 1998 |