1 Introduction

Polarimetric observations are a unique tool to single out different physical processes. In the context of gamma-ray burst (GRB) afterglow emission, some degree of polarization is expected to emerge in the optical flux as a signature of synchrotron radiation (Mészáros & Rees 1997). The observation of power-law decaying lightcurves (e.g. Wijers et al. 1997) and of power-law spectral energy distribution (e.g. Wijers & Galama 1999; Panaitescu & Kumar 2001) give also further support to the external shock synchrotron emission scenario.

The first successful polarization measurement was achieved for the optical afterglow (OA) of GRB 990510 (Covino et al. 1999; Wijers et al. 1999). Some months later, Rol et al. (2000) could perform three distinct observations for GRB 990712, showing a possible variation in the polarization degree, but with constant position angle. More recently, GRB 020813 showed definitely a highly significant variation in the polarization level, again with constant position angle (Barth et al. 2002; Covino et al. 2002a). Last, for GRB 021004, different measurements were performed (Covino et al. 2002b,c; Rol et al. 2002), but the results are still ambiguous because of the large Galactic-induced polarization. For all these observations, the polarization degree was always in the range $(0.8 \div 3)\%$. For three further GRBs, GRB 990123 (Hjorth et al. 1999), GRB 011211 (Covino et al. 2002d) and GRB 010222 (Björnsson et al. 2002), upper limits are again consistent with a maximum value of $\sim$$3\%$ ($95\%$confidence limit).

As a general rule, some degree of asymmetry in the expanding fireball is necessary to produce some degree of polarized flux. Gruzinov & Waxman (1999) argued that if the magnetic field is globally random but with a large number of patches where the magnetic field is instead coherent, a polarization degree up to $\sim$$10\%$ is expected, especially at early times. Ghisellini & Lazzati (1999) and, independently, Sari (1999) considered a geometrical setup in which a beamed fireball is observed slightly off-axis. This break of symmetry again results in a significant polarization. This model also predicts a testable variation of the polarization degree and position angle associated with the evolution of the afterglow lightcurve.

GRB 020405 was localized on 2002 April 5 at 00:41:26 UT by the interplanetary network (IPN) (Hurley et al. 2002). The burst showed a duration of $\sim$40 s and therefore belongs to the class of long duration bursts (Hurley et al. 1992). The optical counterpart was identified by Price et al. (2002a,b) 17.3 hours after the burst as an $R \sim 18.9$ source located at the coordinates  $\alpha_{2000} = 13^{\rm h}58^{\rm m}03\hbox{$.\!\!^{\rm s}$ }12$, $\delta_{2000} = -31\hbox{$^\circ$ }22\hbox{$^\prime$ }22\hbox{$.\!\!^{\prime\prime}$ }2$.

VLT observations allowed to determine the redshift of $z=0.695\pm0.005$(Masetti et al. 2002a) and to discover the bright host galaxy (Masetti et al. 2002b). A new radio source was found at the above coordinates by the VLA (Berger et al. 2002), with a flux of 0.49 mJy at 8.46 GHz.

In addition to those presented here, polarimetric observations were performed by Masetti (2002c) with the VLT and by Bersier et al. (2002) with the Multiple Mirror Telescope, beginning 1.2 and 1.3 days after the GRB, respectively. Even if these two measurements were almost simultaneous, their results are in remarkable contrast. The first group found a level of polarization $P = (1.5\pm0.4)\%$ (hereafter 1-$\sigma$ uncertainties are reported) with position angle $\vartheta = (172 \pm 8)\hbox{$^\circ$ }$, similar to other GRBs, while the second group reported the unprecedented high value $P = (9.89\pm1.3)\%$ at $\vartheta = (179.9\pm3.8)\hbox{$^\circ$ }$. We note however that the results of both groups are not yet published in a refereed journal.