Annu. Rev. Astron. Astrophys. 1996. 34: 511-550 Copyright © 1996 by Annual Reviews. All rights reserved

2.6. The Old Long-Period Variables

More than 1000 Long-Period Variables (LPVs) are known in the field of the LMC. A representative set of articles, with references to the other literature, are found in Bessell et al (1986), Hughes (1989, 1993), Hughes et al (1990, 1991), Reid et al (1995). Most LPVs are luminous asymptotic giant branch (AGB) stars with -7 < M_bol < -4 and come from a population younger than the ancient populations being considered here (Wood et al 1983). The small number of LPVs in Galactic globular clusters tend to have periods between ~ 190^d and ~ 230^d, which places them at the lower end of the range of LPV periods. They are generally found in clusters with [Fe/H] > -1 (Frogel & Elias 1988). The globular cluster LPVs are generally up to a magnitude brighter than the core helium flash termination point of the first ascent red giant branch (RGB) M_bol ~ -3.5, and therefore they are AGB stars close to the point of nuclear fuel exhaustion.

The Milky Way field short-period LPVs have halo kinematics, whereas the longer-period LPVs have disk kinematics (see the introduction to Bessell et al 1986). Under the assumption that the LMC LPVs would have similar properties, Bessell et al (1986), Hughes et al (1990, 1991) define a class of old long- period variables (OLPVs) in the LMC. The theoretical pulsational masses for the OLPVs derived by Wood et al (1983) also support this assumption.

Hughes (1989) gives a large catalog of LPVs from which a sample of OLPVs were drawn. Hughes & Wood (1990) showed that these OLPV stars (defined as LPVs with periods of 150-225 days) in the LMC had absolute magnitudes consistent with those found in Galactic globulars, and Hughes et al (1991) measured velocities and the velocity distribution of the group. They find a trend of velocity dispersion with period (Figure 7 and Table 8 in Hughes et al 1991) that seems to vindicate their assumption that the LPV period is closely associated with the stellar age. The OLPVs have a dispersion of 35 km s^-1, which is significantly larger than that for any other population except the preliminary RR Lyrae result discussed above. This velocity dispersion is not corrected for the problem of observing these LPVs at random phase, but Hughes et al (1991) argue that the random-phase velocity errors are approximately the same as the measurement errors. This velocity dispersion implies that the OLPVs belong to a flattened spheroid and that the mean age derived from the pulsational masses is about 10 Gyr.

How certain are we that the OLPVs are truly an old population? The association with Galactic globulars and the halo-like kinematics of the field OLPVs certainly are consistent with these facts. Yet the OLPVs have some other curious properties. The association of the OLPVs with metal-rich clusters in the Galaxy would imply the Galactic OLPV population should have a kinematical signature that is more like a thick disk than a halo. In addition, almost all LPVs are either carbon (C) stars or M stars. Most of the OPLVs in the lists of Hughes et al (1991) are also C and M stars; many are late M stars. C stars are nonexistent in the Galactic globulars and are found in intermediate-age populations in the Clouds. M stars are found only in the metal-rich clusters ([Fe/H] -1). The OLPV population, therefore, seems signficantly more metal rich than the old clusters listed in Table 1, and if we are interpreting the presence of the C stars correctly, they are also signficantly younger. Finally, there is the disturbing fact that the Galactic globular cluster LPVs do not have the same period-luminosity relationship as the OPLVs of the LMC (Menzies & Whitelock 1991). These authors point out that unless both the Cepheid and RR Lyrae distances scales are in error, this difference implies that the OLPVs in the LMC are more massive than the variables in the Galactic clusters. Whether or not the OLPVs are as old as claimed by Hughes et al (1991), the fact that the velocity dispersion is higher than the old cluster population is provocative.

It would be very interesting to discover and study LPVs in Magellanic Cloud clusters to allow a direct age dating to verify the period-age assumptions and the pulsational masses. Frogel et al (1990) note that a few of their C and M stars found in a grism survey of clusters are clearly dusty LPVs. Periods are not known for these stars.