Annu. Rev. Astron. Astrophys. 1996. 34: 155-206 Copyright © 1996 by Annual Reviews. All rights reserved

3.8. The Milky Way

Observations in the Milky Way offer a unique opportunity for studying interstellar magnetic fields in a detail unobtainable for even nearby external galaxies. However, the plethora of local detail, which obscures any grand-design features of the magnetic field in the Milky Way, still prevents a reliable picture from being obtained.

3.8.1 MAGNETIC fiELD IN THE SOLAR VICINITY     The most reliable estimates concerning the large-scale magnetic field near the Sun are obtained from statistical analyses of Faraday rotation measures of nearby pulsars (within 2-3 kpc from the Sun) and high-latitude extragalactic radio sources, because larger samples involve lines of sight passing through remote regions in the Galaxy for which the inferred magnetic field configuration is less reliable (see Simard-Normandin & Kronberg 1980, Rand & Kulkarni 1989, and references therein). The regular field strength is 2 µG, probably stronger within the arm.

The field is directed towards a galactic longitude of about 90° (see, e.g. Ruzmaikin et al 1978, Rand & Lyne 1994), with an accuracy of 10-20°. The scatter between different determinations makes it difficult to say whether it is aligned with the local spiral arm (pitch angle of about -15°) or not. A tentative upper limit on the magnetic pitch angle, |p| 15°, implies that |_r| 0.3 ||.

The best agreement with observations is provided by models with the horizontal global magnetic field similarly directed above and below the midplane (S-type field) (Gardner et al 1969;, Vallée & Kronberg 1973, 1975). Claims of an odd symmetry (Morris & Berge 1964;, Andreassian 1980, 1982) probably result from contamination by strong local distortions in the magnetic field. A similar problem prevents the reliable detection of the vertical magnetic field _z near the Sun: It is so weak that it cannot be separated from local magnetic inhomogeneities, |_z| << |_r|, ||.

Because the warm interstellar medium is the main contributor to the electron density in the diffuse ISM, rotation measures sample mainly this phase of the ISM. Heiles (1996) has argued that the warm interstellar medium occupies only 20% of the total volume in the Milky Way, so that the resulting does not reflect the true volume-averaged field. This argument would apply also to external galaxies, where Faraday rotation is also used to study . However, the observed coherency of RM patterns over large regions in many nearby galaxies indicates that the inferred magnetic field is global rather than restricted to a small fraction of the volume (see also Section 3.1).

3.8.2 REVERSALS OF THE MAGNETIC fiELD AND ITS AZIMUTHAL STRUCTURE     The property of the magnetic field in the Milky Way that distinguishes it from probably most other galaxies investigated up to now is the reversals of the regular field along the radius. The reversal closest to the Sun between the local (Orion) and the next arm to the center (Sagittarius) was first detected by Simard-Normandin & Kronberg (1979). The reversal is located in the interarm region at about 0.4-0.5 kpc inside the solar circle (see Rand & Lyne 1994).

There are some indications of more reversals at both smaller and larger galactocentric distances, but this evidence is much more controversial because distant spiral arms occupy smaller areas on the sky. Simard-Normandin & Kronberg (1980), Vallée (1983) argued that there is no reversal between the local and the next outer (Perseus) arms, whereas other authors found some evidence for an outer reversal (Agafonov et al 1988, Rand & Kulkarni 1989, Lyne & Smith 1989, Clegg et al 1992). Two additional reversals were claimed for the inner Galaxy by Sofue & Fujimoto (1983) and Han & Qiao (1994), but most analyses more conservatively imply only one more, at a galactocentric radius of 5.5 kpc (Vallée et al 1988, Vallée 1991, Rand & Lyne 1994). The controversy about the number of reversals is partly due to difficulties in the analysis of Faraday rotation measures. There are natural complications associated with strong local distortions of magnetic field, e.g. the North Polar Spur or the Gum Nebula. However, there are also many pitfalls in the statistical analyses. Many results rely on simple "naked-eye" fitting of the observational data (e.g. Simard-Normandin & Kronberg 1980, Sofue & Fujimoto 1983), which is especially dangerous when the global structure is investigated; others are based on nonrigorous applications of statistical tests (e.g. Han & Qiao 1994). Vallée (1996) discusses some of these problems. More rigorous studies imply an axisymmetric field with two reversals (Rand & Kulkarni 1989, Rand & Lyne 1994), although more cannot be excluded. The radial distribution of the magnetic field strength is shown in Figure 5 (see also Heiles 1996).

Figure 5. The strength of the large-scale magnetic field in the Milky Way (full circles with error bars) and positions of its reversals (crosses), as inferred from pulsar rotation measures (Rand & Lyne 1994). Note a gradual increase of || towards smaller radii (a positive corresponds to the field direction towards the first and second Galactic quadrants). Error bars shown correspond to 30% uncertainty and are chosen tentatively to indicate a scatter of the available estimates at r = 8.5 kpc, the Galactic radius of the sun. The solid line shows the strength of the total magnetic field, averaged in azimuth as obtained by EM Berkhuijsen (in preparation) from the deconvolved surface brightness of synchrotron emission at 408 MHz (Beuermann et al 1985), assuming energy equipartition between magnetic field and cosmic rays; the accuracy of this estimate is probably 30%.

The available statistical analyses adopt either a bisymmetric structure of the global magnetic field (Simard-Normandin & Kronberg 1980, Sofue & Fujimoto 1983, Han & Qiao 1994 or a concentric-ring model in which magnetic field lines are directed exactly in the azimuthal direction. Comparison between these two models often shows that the latter provides a better fit to the data (e.g. Rand & Kulkarni 1989); however, the concentric-ring model is unrealistically simplistic. The regular magnetic field cannot have a zero pitch angle everywhere (see Section 8.3), even if it does near the Sun. The model is consistent neither with theoretical ideas about galactic magnetic fields nor with observations of external galaxies (Section 3.3). The pitch angle of the magnetic field should be a model parameter, possibly a function of position, obtained from fits to data rather than fixed to be zero (or any other value) beforehand. Another problem is that the magnetic field may really correspond to a superposition of different azimuthal modes, so that attempts at fitting a purely axisymmetric or bisymmetric model may lead to erroneous results.

The presence of reversals in the Milky Way is often interpreted as an unambiguous indication of the bisymmetric global structure of the magnetic field. As we discuss in Section 8.5, axisymmetric magnetic structures may also contain reversals, and mean-field dynamo models for the Milky Way favor an axisymmetric field structure.

Field reversals have rarely been observed in external galaxies, only in BSS candidates (see Table 2) and possibly in a galaxy at redshift 0.395 (Kronberg et al 1992; see Section 3.7). In some galaxies, the resolution of the observations is high enough to detect reversals if they were present: This is the case for M31 observed with a resolution of 1 kpc near the major axis (Beck 1982, Ruzmaikin et al 1990). In other galaxies the resolution of Faraday rotation data is lower (e.g. Krause et al 1989a, Buczilowski & Beck 1991, Ehle & Beck 1993) and reversals cannot be excluded. However, the number of reversals within the telescope beam cannot be large as this would average out any Faraday rotation.

Because the Sun is located fairly close to a reversal, the strength of the regular magnetic field at r 8.5 kpc is less likely to be a representative value for the bulk of spiral galaxies or even for the Milky Way itself. Values of order 4-6 µG seem to be more typical.