Rotational widths for use in the tully-fisher relation. II. The impact of surface brightness

Using a large sample of spiral galaxies for which 21 cm single-dish and/or long-slit optical spectra are available, we make a detailed comparison between various estimates of rotational widths. Different optical width estimators are considered and their limitations discussed, with emphasis on biases associated with rotation curve properties (shape and extent) and disk central surface brightness. The best match with H I rotational velocities is obtained with Polyex widths, which are measured at the optical radius (encompassing a fixed fraction of the total light of the galaxy) from a model fit to the rotation curve. In contrast with Polyex widths, optical rotational velocities measured at 2.15 disk scale lengths r_d deviate from H I widths by an amount that correlates with the central surface brightness of the disk. This bias occurs because the rotation curves of galaxies are in general still rising at 2.15r_d, and the fraction of total mass contained within this radius decreases with increasing disk surface brightness. Statistical corrections, parameterized by the radial extent of the observed rotation curve, are provided to reduce Polyex and H I width measurements into a homogeneous system. This yields a single robust estimate of rotational velocity to be used for applications of disk scaling relations.