The brightest, or first-ranked, galaxies (BCGs) in rich clusters show a very small dispersion in luminosity, making them excellent standard candles. This small dispersion raises questions about the nature of BCGs, Are they simply the extremes of normal galaxies formed via a stochastic process, or do they belong to a special class of atypical objects? If they do, are all BCGs special, or do normal galaxies compete for the first rank? To answer these questions, we undertake a statistical study of BCG magnitudes using results from extreme value theory. Two-population models do better than one-population models. A simple scenario where a random boost in the magnitude of a fraction of bright normal galaxies forms a class of atypical galaxies best describes the observed distribution of BCG magnitudes.