As mentioned already it was the optical polarization observations that gave us the first information about magnetic fields in external galaxies. Öhman (1942) gave us in addition to the results for M31 the detailed description of all the many problems of the observing technique. The advantages and/or the problems of surface polarimetry compared to the observations of discrete sources (stars, globular clusters) were discussed. The interpretation for the reasons for polarized light had to wait until Davis and Greenstein (1951). The optical polarization studies of galaxies (e.g. Hiltner, 1958; Elvius & Hall, 1965; Appenzeller, 1967; Bingham et al., 1976; Scarrott et al., 1977; Elvius, 1978; Martin & Shawl, 1982; Scarrott et al., 1987) are characterized by ever increasing sensitivity. The photomultiplier has been replaced by a CCD detector. The polarization analyzer remained essentially the same;- polarization foil or a Wollaston prism. Savart plates are also used for studies of stars. A new era of optical polarization observations seems to be at hand in view of the relative availability of medium sized telescopes and sensitive CCD detectors.
The radio observations needed some time to develop sensitive methods to measure polarization in galaxies. The first published result for a galaxy was for M51 by Mathewson et al. (1972) using the then commissioned Westerbork synthesis radio telescope. A follow-up observation of Segalovitz et al. (1976) gave us information about M51 and M81. The Effelsberg 100-m dish has been intensively dedicated to the study of magnetic fields in galaxies since the first results on M31 were published by Beck et al., (1978, 1980). Since that time practically all the large northern galaxies have been mapped in Effelsberg at wavelengths 11 to 2.8 cm. In the quest of angular resolution the Very Large Array (VLA) has been used, in particular in the D array mode, at lower frequencies. More recently the Parkes radio telescope has been used for polarization mapping of the Magellanic Clouds (Haynes et al., 1986, 1990) and for large southern galaxies (Harnett et al., 1989, 1990). The present data base needs to be expanded both in respect to angular resolution (without loss of sensitivity) and to higher frequencies. Some progress with existing radio telescopes is possible. Given longer integration times for the C and B array mapping at the VLA we should get better information. The 100-m telescope in Effelsberg with a multibeam receiver at 9mm wavelength will allow 25" angular resolution practically free of Faraday effects. The commissioning of the Australia Telescope should usher in a new era of studies of southern galaxies.