Histogram showing the number of reliable directions for the 215 sources in our sample. The criterion for a reliable direction is to have σ_Θ < 30°. The color refers to the number of components in the models derived for a threshold of 0.9. Upper panel: histogram before leaving out the duplicate directions. Lower panel: same histogram after leaving out the duplicate directions. The criterion for a direction to be considered a duplicate of a previously identified direction is to present a normalized direction difference smaller than 3 with respect to this other direction (see details in the text).