Around M87 and M49, we find velocity dispersion profiles that rise slightly with radius, suggesting massive extended dark halos, as found by the GC and X-ray studies. Combining all these constraints together will give us a detailed picture of the mass and orbit distributions. It is initially evident that the stellar orbit radial anisotropy increases into the halo.
The more ordinary (∼ L*) ellipticals from our PN.S studies, as well as NGC 4697 (Méndez et al. 2001), show something entirely different: their velocity dispersions decline rapidly with radius (see Fig. 2, left). Simple Jeans models with a moderate degree of anisotropy indicate total masses consistent with the visible stars only: a benchmark parameter ΥB5 (M / L inside 5 Reff) is ∼ 6–15 (Romanowsky et al. 2003), while stellar populations should have ΥB ∼ 3–12 (Gerhard et al. 2001). For NGC 3379, we have constructed more versatile orbit models to allow for the infamous mass-anisotropy degeneracy, and to extract as much information as possible out of the discrete PN velocity data (Romanowsky & Kochanek 2001). These tightly constrain ΥB5 to be 7.1 ± 0.6 (Fig. 2, right). There are still systematic uncertainties in this study, notably the possibility that the galaxy contains a component that is flattened along the line-of-sight; however, independent confirmation of a low M / L comes from an HI ring (Schneider et al. 1989), and from a steeply declining dispersion in the GCs (Beasley et al. 2004; Bergond et al. 2004) — which are unlikely to reside in a flattened system.
Figure 2. Left: Projected velocity dispersion profiles for four elliptical galaxies (see text) scaled and stacked, as a function of radius, in units of Reff. Open points show PN data; solid points show long-slit stellar data. Simple model predictions are shown for comparison: a singular isothermal halo (dashed line) and a constant-M / L galaxy (dotted line). Right: Circular velocity profile of NGC 3379. The solid lines and shaded area show the region permitted by orbit modeling within the 68% confidence limits. Dotted lines show excluded models (bottom: constant-M / L; top three: more dominant dark halos).
For comparison, other results from brighter galaxies typically show ΥB5 ∼ 20–40 (Bahcall et al. 1995; Loewenstein & White 1999). Weak lensing estimates for L* ellipticals, extrapolated inwards in radius, give ΥB5 ∼ 15–25 (Wilson et al. 2001; Seljak 2002). Theoretical CDM predictions at these intermediate radii are not yet firm, but the best estimates give ΥB5 ∼ 15–20 (e.g., Weinberg et al. 2004; Wright et al. 2004), somewhat higher than we infer for the PN galaxies.
Thus, it seems there is a population of elliptical galaxies largely bereft of dark matter. One explanation is that their primordial CDM halos have been stripped away or puffed up by interactions with other systems, as can happen in galaxy clusters (Natarajan et al. 2002). However, these ellipticals are not in such rich environments. An alternative is that this is another manifestation of the generic problem of insufficiently centrally concentrated dark matter; we infer halo concentrations c ∼ 5 while CDM predicts c ∼ 15 before including baryonic effects. This interpretation is supported by studies at smaller radii with gravitational lenses — where the dark matter fraction is lower than CDM expectations, especially given additional priors on H0 (Rusin et al. 2003) — and with dynamical model fits to the fundamental plane (Borriello et al. 2003). Further observational and theoretical investigations are needed to see if this “missing missing mass” presents a major predicament for CDM.