| Abstract | The brightest stars in galaxies have attracted considerable interest and observational effort as potential distance indicators. We present a thorough assessment of the problems and uncertainties associated with the method. We find that the inherent error associated with any distance estimate using the method has been underestimated, a comment that also applies to several recent claims using the method in support of the `short' distance scale. In this paper we re-examine most of the existing observational data. Comparisons between different sets of data are discussed in an attempt to place all the data on a common scale so that it can be properly combined. For the first time, an error is associated with the quantity M_*_(3), and a rigorous procedure is established for the accounting of errors. The appropriate use of the technique in the infrared (K) band and non- calibrating bands is fully discussed. The resulting calibrations are compared with those obtained by other researchers in the field: Humphreys and Sandage. The well-known dependence of the brightest blue supergiant luminosity on parent galaxy magnitude is confirmed. However, & similar dependence is found for the brightest red supergiants in both the V and K bands. The resulting error in any distance estimate obtained using the method a found to be much higher than previously thought for all calibration techniques, and in no case is δμ less than 0.55 mag. The method's use in two programmes observing the Virgo group and other galaxies is critically assessed. The prospects for reducing the errors associated with the method are assessed and we conclude that the method is unable to distinguish between the `long' and `short' distance scales on the basis of observations to single galaxies. |
