The observed polarization for the programme stars ranges from a negligibly small value of 0.05 % for star 22 to a value as large as 4% for star 5. The polarization position angle also shows a large dispersion when all the stars are considered together. However, when the association members and field stars are considered separately the two groups of stars can be distinguished by their polarization behaviour. The association members (including star 10 that was considered a possible member by Westerlund (1963)) are characterized by a moderately high value of polarization. These stars have an average value of polarization with a standard deviation . The average position angle with a standard deviation . Only 3 stars (numbered 8, 20 and 26) of the 24 have polarization but their position angles are similar to the other member stars. In contrast, the field stars have polarization values all with ; and ; . This large standard deviation in the position angles is due to only one star (number 23). If star 23 is not included then the average position angle for field stars is with . Thus the association members and the field stars form two distinct groups. The histograms of the observed degree of polarization and position angles presented in Figs. 1 and 2 show this separation very clearly. The histograms are double peaked with the member stars and the field stars predominantly occupying the high polarization (with position angles in the range ) and the low polarization (with position angles in the range ) bins respectively.
The double peaked histograms discussed above indicate that in the direction of the Puppis OB III association there are perhaps two distinct regions of obscuring interstellar matter. If the observed polarization is caused by dust grains aligned by the Davis - Greenstien mechanism, then the two regions of the interstellar medium are threaded by magnetic fields that are oriented (in the plane of the sky) very differently. The nearby field stars are polarized by a relatively thin region of obscuring matter with a magnetic field oriented at and a thicker region (perhaps local to the OB association) with the magnetic field oriented at . The two field directions are almost perpendicular to each other. A polarization map (which also represents the geometry of the projected magnetic field in the region) is presented in Fig. 3, with the polarization vectors of lengths proportional to the degree of polarization measured and the direction parallel to the electric vector. The Galactic equator is also drawn for reference.
A plot of the observed polarization against the distance could give information on the location of the polarizing medium. However, the distances to individual stars are not well determined. The distance modulii () that can be determined from the photometry for these stars given in Westerlund (1963) indicate that the field stars are all within . The association members must all be nearly at the same distance. However, we compute formal distance modulii (which can have considerable scatter) from the data given in Westerlund (1963). In Fig. 4 we plot the degree of polarization measured against the formal distance modulus for the programme stars as well as 14 additional stars within an angular radius of around the star RS Pup for which measurements are available in the catalogue of polarization measurements by Mathewson, Ford, Klare, Neckel & Krautter (1978) available in the Simbad data base. It is evident from Fig. 4 that the degree of polarization jumps up at around . This implies the existence of a polarizing medium at , perhaps a system of dust clouds including RCW 19, 20 and the CO cloud detected by May et al (1988) related to the Puppis OB III association itself.
As expected for interstellar polarization the oberved degree of polarization shows a positive correlation with the reddening for these stars. Fig. 5 shows a plot of the degree of polarization measured by us against the colour excess for the stars given in Westerlund (1963). There is much scatter, but larger values of polarization are generally observed for stars with larger values of reddening. For interstellar polarization a similar plot (Serkowski et al 1975) of polarization in the V band against is well bounded on the polarization axis by a line . In Fig. 5 also the points are bounded on the polarization axis. However, polarization measurements in the present work were made in the R band. For normal interstellar polarization, the wavelength dependence of polarization is well represented by the Serkowski law (Serkowski 1973): , with corresponding to the V band. For interstellar polarization, the Serkowski law gives , and the line (drawn in Fig. 5 as a solid line) is expected to bound the points in the P versus plot. However, the observed points in Fig. 5 are well bounded by the line drawn in the figure as a dotted line. Thus the average polarization to extinction ratio is 3 times smaller than the normal interstellar value. This may indicate a poor grain alignment efficiency in the direction of the Puppis OB III association. Alternatively, the low value of the polarization/reddening ratio could be a result of the magnetic field configuartion being predominantly longitudinal. At the galactic longitude () of the Puppis OB III association, this situation could arise if the magnetic field is parallel to the galactic spiral arm in that region. However, as seen in Fig. 5, the nearby field stars for which the projected magnetic field is nearly perpendicular to the galactic plane, also show low polarization /reddening ratio.
© European Southern Observatory (ESO) 1998
Online publication: February 16, 1998