1999AJ....118.1963C

We have made high-quality measurements of the polarization spectra of 13 FR II radio galaxies and taken polarization images for 11 of these with the Keck telescopes. Seven of the eight narrow-line radio galaxies (NLRGs) are polarized, and six of the seven show prominent broad Balmer lines in polarized light. The broad lines are also weakly visible in total flux. Some of the NLRGs show bipolar regions with roughly circumferential polarization vectors, revealing a large reflection nebula illuminated by a central source. Our observations powerfully support the hidden quasar hypothesis for some NLRGs. According to this hypothesis, the continuum and broad lines are blocked by a dusty molecular torus, but can be seen by reflected, hence polarized, light. Classification as a NLRG, a broad-line radio galaxy (BLRG), or a quasar therefore depends on orientation. However, not all objects fit into this unification scheme. Our sample is biased toward objects known in advance to be polarized, but the combination of our results with the 1996 findings of Hill, Goodrich, & DePoy show that at least six out of a complete, volume and flux-limited sample of nine FR II NLRGs have broad lines, seen either in polarization or Pα.

The BLRGs in our sample range from 3C 382, which has a quasar-like spectrum, to the highly reddened IRAS source FSC 2217+259. This reddening sequence suggests a continuous transition from unobscured quasar to reddened BLRG to NLRG. Apparently the obscuring torus does not have a distinct edge. The BLRGs have polarization images that are consistent with a point source broadened by seeing and diluted by starlight. We do not detect extended nebular or scattered emission, perhaps because it is swamped by the nuclear source. Our starlight-corrected BLRG spectra can be explained with a two-component model: a quasar viewed through dust and quasar light scattered by dust. The direct flux is more reddened than the scattered flux, causing the polarization to rise steeply to the blue. Strong rotations of the electric vector position angle across Hα in 3C 227 and 3C 445 may be explained by systematic orbital motions in an equatorial broad-line region.