Comments on the cluster main-sequence fitting method. III. Empirical UBV reddening lines for early-type stars