SIMBAD references

Horizontal branch (HB) stars in globular clusters offer us a probe of the mass loss mechanisms taking place in red giants. For M3 (NGC 5272), in particular, different shapes for the HB mass distribution have been suggested in the literature, including Gaussian and sharply bimodal alternatives. Here, we study the mass distribution of HB stars in M3 by comparing evolutionary tracks for a suitable chemical composition with photometric observations. Our approach is thus of a semi-empirical nature, describing a mass distribution favored from the standpoint of canonical stellar evolutionary predictions for the distribution of stars across the color-magnitude diagram. More specifically, we locate, for each individual HB star in M3, the evolutionary track whose distance from the star's observed color and magnitude is a minimum. We carry out tests that reveal that our method would be able to detect a bimodal mass distribution resembling that previously suggested in the literature, if present. We also study the impact of different procedures for taking into account the evolutionary speed, and conclude that they have only a small effect on the inferred mass distribution. We find that a Gaussian shape, though providing a reasonable first approximation, fails to account for the detailed shape of M3's HB mass distribution. Indeed, this mass distribution may have skewness and kurtosis that deviate slightly from a perfectly Gaussian solution. Alternatively, the excess of stars towards the wings of the distribution may also be accounted for in terms of a bimodal distribution in which both the low- and the high-mass modes are normal, the former being significantly wider than the latter. We also show that the inferred distribution of evolutionary times is inconsistent with theoretical expectations. This result is confirmed on the basis of three independent sets of HB models, suggesting that the latter underestimate the effects of evolution away from the zero-age HB, and warning against considering our inferred mass distribution as definitive.