1. Present observational constraints
The bulk of the mass of the LMC resides in a nearly face-on disk, with an inclination usually taken to equal the canonical value of (Westerlund 1997), although both lower () and higher (up to ) values have also been derived from morphological or kinematical studies of the LMC . This disk is observed to rotate with a circular velocity km s-1 out to at least from the LMC center (Schommer et al. 1992). If all the stars belong to the same population, with a vertical (i.e. perpendicular to the disk) velocity dispersion , the microlensing optical depth of such a disk upon its own stars is given by (Gould 1995). Considering the measured velocity of LMC carbon stars (Cowley & Hartwick 1991), Gould (1995) assumed km s-1 as a typical velocity dispersion for LMC stars. He thus concluded that , i.e. that self-lensing (first suggested by Sahu 1994and Wu 1994) contributes very little to the observed optical depth towards this line of sight.
Carbon stars however may not be the ultimate probe to infer the velocity dispersion of LMC populations: they actually comprise various ill-defined classes of objects (Menessier 1999), and their prevalence is a complex function of age, metallicity and probably other factors (Gould 1999).
Both observational and theoretical arguments favour the existence of a wide range of velocity dispersions among the various LMC stellar populations. To commence, Meatheringham et al. (1988) have determined the radial velocities of a sample of planetary nebulae (PN) in the LMC . They measured a velocity dispersion of 19.1 km s-1, much larger than the value of 5.4 km s-1 found for the HI. This was interpreted as being suggestive of orbital heating and diffusion operating in the LMC in the same way as it is observed in the solar neighbourhood. Then, the observations of Hughes et al. (1991) show clear evidence for an increase in the velocity dispersion of long period variables (LPV) as a function of their age. For young LPVs, the velocity dispersion is 12 km s-1 whereas for old LPVs, it reaches 35 km s-1. More recently, Zaritsky et al. (1999) found a velocity dispersion of km s-1 for 190 vertical red clump (VRC) stars 1 whereas for the red clump (RC), they measured a value of km s-1 on a sample of 75 objects (throughout this paper, error bars are converted from Zaritsky's 95% confidence levels to standard ). A general trend appears: the velocity dispersion is an increasing function of the age. Just like for our own Milky Way, stars of the LMC disk have been continuously undergoing dynamical scattering by, for instance, molecular clouds or other gravitational inhomogeneities. This results in an increase of the velocity dispersion of a given stellar population with its age, as will be further discussed in Sect. 3. Notice that the main argument in disfavour of a LMC self-lensing explanation is precisely the low value of the measured vertical velocity dispersions. However, the stellar populations so far surveyed predominantly consist of red giants. They are shown in the next section not to be representative of the bulk of the LMC disk stars, and actually biased towards young ages: they are on average 2 Gyr old, to be compared to an LMC age of 12 Gyr.
© European Southern Observatory (ESO) 1999
Online publication: October 14, 1999