Measuring Stellar Velocity Dispersions in Active Galaxies

We present stellar velocity dispersion (σ*) measurements for a significant sample of 40 broad-line (type 1) active galaxies for use in testing the well-known relation black hole mass and stellar velocity dispersion. The objects are selected to contain Ca II triplet, Mg I b triplet, and Ca H+K stellar absorption features in their optical spectra so that we may use them to perform extensive tests of the systematic biases introduced by both template mismatch and contamination from the active galactic nucleus (AGN). We use the Ca II triplet as a benchmark to evaluate the utility of the other spectral regions in the presence of AGN contamination. Broad Fe II emission, extending from ~5050 to 5520 Å, in combination with narrow coronal emission lines, can seriously bias σ* measurements from the Mg I b region, highlighting the need for extreme caution in its use. However, we argue that at luminosities constituting a moderate fraction of the Eddington limit, when the Fe II lines are both weak and smooth relative to the stellar lines, it is possible to derive meaningful measurements with careful selection of the fitting region. In particular, to avoid the contamination of coronal lines, we advocate the use of the region 5250-5820 Å, which is rich in Fe absorption features. At higher AGN contaminations, the Ca H+K region may provide the only recourse for estimating σ*. These features are notoriously unreliable, due to a strong dependence on spectral type, a steep local continuum, and large intrinsic broadening. Indeed, we find a strong systematic trend in comparisons of Ca H+K with other spectral regions. Luckily the offset is well described by a simple linear fit as a function of σ*, which enables us to remove the bias and thus extract unbiased σ* measurements from this region. We lay the groundwork for an extensive comparison between black hole mass and bulge velocity dispersion in active galaxies, as described in a companion paper by Greene & Ho.