The role of magnetic fields in both the initial collapse of molecular clouds to form protostars and their subsequent evolution to the main-squence via the outflow phase, is still unclear. Gravitational collapse may follow the magnetic field lines of the parent molecular cloud, producing a flattened, slowly rotating core, which will continue to collapse as turbulent and magnetic support is lost due to ambipolar diffusion (e.g. Shu, Adams & Lizano 1987). If this is the case, one might expect a correlation between the outflow axis and the magnetic field lines of the parent molecular cloud. Observational evidence imply this may well be the case (e.g. Strom & Strom 1987; Hodapp 1990).

The importance of magnetic fields has been incorporated into several recent outflow models. Pudritz & Norman (1983,1986) have proposed that bipolar outflows are centrifugally driven hydromagnetic winds that originate in the envelopes of rotating magnetised discs around protostars. The disc envelope pressure is maintained by constant flux loss from the dense core due to ambipolar diffusion. This model produces a poloidal magnetic field in the shape of an `hour glass', that co-rotates with the disc. Alternatively, Uchida & Shibata (1985) have proposed that mass is accelerated in the relaxing magnetic twist, created by the wind-up of the magnetic field in the contracting and rotating circumstellar disc. The magnetic field of the disc is coupled to the helical field lines of the parent molecular cloud. The mass outflow follows the field lines (leaving a hollow interior) and gradually accelerates with distance. Unlike the Pudritz & Norman model, the field close to the protostar is toroidal (rather than poloidal). Therefore, the ability to measure the magnetic field direction in the vicinity of outflow sources may reveal the relationship between the disk magnetic field and that of the parent molecular cloud. This should provide direct evidence for or against each of the hydromagnetic wind models.

As optical, near and mid-infrared polarization can be produced by either scattering, dichroic absorption or dichroic emission of radiation, any information deduced from observations about the magnetic fields close to protostars is highly uncertain as they require the `disentangling' of the polarizing effects. The most direct, and therefore potentially most reliable technique for mapping these magnetic fields, is to observe the polarised emission from warm dust in the far-infrared or sub-millimetre. Observations at such long wavelengths have the advantage of being free from contamination by scattered radiation. The position angle of polarization for dust emission is parallel to the long axis of the dust grain, and therefore perpendicular to the magnetic field, assuming the dust grain is aligned by paramagnetic relaxation (Davis & Greenstein 1951).

We report the detection of 800 mm polarization at the flux peak and positions around the outflow sources DR21 and NGC7538-IRS11 (referred to as IRS11 throughout the rest of the paper). The observations were made using the Aberdeen/QMW polarimeter in conjunction with the continuum receiver UKT14. The data acquisition modes and data analysis methods used with the polarimeter are described elsewhere (Murray 1991; Minchin & Murray 1994).

The polarimetric data for both DR21 and IRS11 are presented in Table 1 and shown graphically in Figs. 1 and 2, respectively. For both sources the direction of the polarization vectors is extremely uniform and the dispersion is small. This implies that, within the reslution of our observations, the magnetic field is both uniform in direction and strong. If this were not so disruptive processes would affect the grain alignment, reducing the uniformity of the observed polarization position angles (Chandrasekhar & Fermi 1953).

The direction of the magnetic field around DR21 is not aligned with either the direction of the outflow axis or the major axis of the submm circumstellar dust structure. As the DR21 region is extremely complex, possibly containing 3 outflows in close proximity, the non-alignment may not be significant.

For NGC7538-IRS11 the magnetic field is aligned with the outflow axis, implying that within the resolution of our observations (14 arcsec FWHM) the circumstellar magnetic field is poloidal. The direction of the molecular outflows, magnetic fields and dust ridges for both IRS11 and its neighbour IRS1 (60 arcsec/0.9 parsec to the north) are in identical directions. As IRS1 and IRS11 are linked by an arm of submm emission and form an elongated dust ridge that is orthogonal to the ambient magnetic field of the NGC7538 region, part of the cloud may have collapsed along the magnetic field lines to produce the core from which IRS1 and IRS11 have formed.

There is a marked increase in the observed percentage polarization from DR21 at higher wavelengths (1100 mm and 1300 mm). This implies the grain composition cannot be predominantly silicate, but instead is mainly graphite/metallic, and may require grains of different magnetic susceptibility or varying elongation to also be at different temperatures along the observed line of sight.

Position	RA(")	Dec(")		P (%)			Theta

Number		Offset	Offset					(degrees)

DR21
		
1		0	0		1.8 ± 0.3		17 ± 4

2		7	0		2.1 ± 0.5		29 ± 7

3		7	7		3.0 ± 0.5		15 ± 6

4		0	7		2.7 ± 0.4		20 ± 5

5		-7	7		2.4 ± 0.3		30 ± 4

6		-7	0		1.7 ± 0.3		34 ± 5

7		-7	-7		2.5 ± 0.4		30 ± 5

8		0	-7		1.9 ± 0.8		23 ± 11

9		7	-7		2.2 ± 0.5		35 ± 7

10		0	14		2.6 ± 0.6		23 ± 6

11		0	-14		2.0 ± 0.9		30 ± 12

NGC7538 IRS11

1		0	0		2.5 ± 0.2		58 ± 2

2		7	0		4.0 ± 0.4		66 ± 3

3		7	7		3.7 ± 0.6		56 ± 5

4		0	7		3.4 ± 0.3		57 ± 3

5		-7	0		2.6 ± 0.5		64 ± 6

6		0	14		5.6 ± 0.9		49 ± 4

The percentage polarization at the position of the flux peak for each outflow source is low compared to the offset positions. This has been noted for several other outflow sources and is commonly referred to as a polarization `hole'. It is unlikely that either the effect of flux contamination by unpolarized line emission or reduced grain alignment/changes in the grain size distribution or composition could be responsible. The most plausible explanation is a change in the magnetic field alignment in the vicinity of the outflow source. Poloidal magnetic field lines may become directed closer to the observers line of sight in the vicinity of the outflow source, or the magnetic field lines close to the outflow source may become twisted, possibly due to the presence of a toroidal field in a circumstellar disc.

This is a summarised version of a paper to be published shortly in Astronomy & Astrophysics (Minchin & Murray 1994).

Nigel R. Minchin and Alexander G. Murray,

Queen Mary and Westfield College, London

References:

Chandrasekhar S., Fermi E., 1953, ApJ 118, 113

Davis L., Greenstein J. L., 1951, ApJ 114, 206

Hodapp K. W., 1990, ApJ 352, 184

Minchin N. R., Murray A. G., 1994, A&A in press

Murray A. G., 1991, Ph. D. Thesis, University of Aberdeen

Pudritz R., Norman C. A., 1983, ApJ 274, 677

Pudritz R., Norman C. A., 1986, ApJ 301, 571

Richardson K. J., Sandell G., Krisciunas K., 1989, A&A 224, 199

Sandell G., et al. 1994, in preparation

Shu F. H., Adams F. C., Lizano S., 1987, ARA&A 25, 23

Strom S. E., Strom K. M., 1987, in: Star Forming Regions, IAU Symp. No. 115, eds. Peimbert M., Jugka J., Reidel, Dordrecht, p. 255

Uchida Y., Shibata K., 1985, PASJ 37, 515