Figure 1 gives an example of observations of M51 at 4 different wavelengths, all smoothed to the same linear resolution of 75" HPBW. The vectors are rotated by 90° but not corrected for Faraday rotation. The figure illustrates nicely the different effects of Faraday rotation and depolarization effects depending on the observing wavelength: the observed vectors at 2.8 cm and 6 cm are mainly parallel to the optical spiral arms as expected in spiral galaxies (see below), Faraday rotation is small at centimeter wavelengths. However, the pattern looks very different at 18/20 cm where Faraday rotation is expected to be strong. Further, we see a region in the northeastern part of M51 with complete depolarization.
Figure 1. Maps of the E-vectors rotated by 90° of M51 observed at 2.8 cm, 6.2 cm, 18.0 cm, and 20.5 cm. The length of the vectors is proportional to the polarized intensity. They are shown superimposed onto an optical picture (Lick Observatory).
After substraction of the thermal fraction of the emission we distinguish between beam-dependent and wavelength-dependent depolarization. The difference in depolarization at different wavelengths in maps with the same linear resolution should be purely wavelength dependent where two different wavelength-dependent depolarization effects are important to consider: the differential Faraday rotation and Faraday dispersion as despcribed by Burn (1966) and Sokoloff et al. (1998). The latter effect is due to turbulent magnetic fields within the source and between the source and us, whereas the Faraday rotation depends on the regular magnetic field within the emitting source. The differential Faraday rotation has a strong wavelength dependence as shown e.g. in Fig. 1 in Sokoloff et al. (1998) leading to a complete depolarization at 20 cm already at a RM 40 rad / m2, with again decreasing depolarization for higher RMs. Such an effect has first been detected in small isolated areas in M51 (Horellou et al. 1992). At 6 cm the depolarization is much smaller, increasing smoothly to zero at RM 400 rad / m2 (the first zero point is at RM = / (2 . 2)).
Hence, the galaxies may not be transparent in linear polarization at decimeter wavelengths so that we may observe just an upper layer of the whole disk. At centimeter wavelengths we do not expect complete depolarization even in galaxies viewed edge-on, i.e. centimeter wavelengths are best suitable to trace the magnetic field structure.