3.2 NGC 3115: M_BH 10^{9.0 ± 0.3} M

One of the best stellar-dynamical BH cases is the prototypical S0 galaxy NGC 3115 (Fig. 1). It is especially suitable for the BH search because it is very symmetrical and almost exactly edge-on. NGC 3115 provides a good illustration of how the BH search makes progress. Unlike some discoveries, finding a supermassive BH is rarely a unique event. Rather, an initial dynamical case for a central dark object gets stronger as observations improve. Eventually, the case becomes definitive. This has happened in NGC 3115 through the study of the central star cluster - a tiny, dense cusp of stars like those expected around a BH (Figure 1). Later, still better observations may accomplish the next step, which is to strengthen astrophysical constraints enough so that all plausible BH alternatives (clusters of dark stars) are eliminated. This has happened for our Galaxy (Section 3.4) but not yet for NGC 3115.

Figure 1. HST WFPC2 images of NGC 3115. The left panel shows a color image made from 1050 s V- and I-band images. The right panel shows a model of the nuclear disk. The center panel shows the difference; it emphasizes the compact nuclear star cluster. Brightness is proportional to the square root of intensity. All panels are 11".6 square. [This figure is taken from Kormendy et al. 1996, Astrophys. J. Lett., 459, L57.]

The kinematics of NGC 3115 show the signature of a central dark object (Fig. 2). The original detection was based on the blue crosses. Already at resolution _* = 0".44, the central kinematic gradients are steep. The apparent central dispersion, 300 km s^-1, is much higher than normal for a galaxy of absolute magnitude M_B = -20.0. Therefore, isotropic dynamical models imply that NGC 3115 contains a dark mass M_BH 10^{9 ± 0.3} M. Maximally anisotropic models allow smaller masses, M_BH ~ 10⁸ M, but isotropy is more likely given the rapid rotation.

Figure 2. Rotation velocities (lower panel) and velocity dispersions (upper panel) along the major axis of NGC 3115 as observed at three different spatial resolutions. Resolution * is the Gaussian dispersion radius of the PSF; in the case of the HST observations, this is negligible compared to the aperture size of 0".21. [This figure is adapted from Kormendy et al. 1996, Astrophys. J. Lett., 459, L57.]

Since that time, two generations of improved observations have become available. The green points in Figure 2 were obtained with the Subarcsecond Imaging Spectrograph (SIS) and the Canada-France-Hawaii Telescope (CFHT). This incorporates tip-tilt optics to improve the atmospheric PSF. The observations with the HST Faint Object Spectrograph (FOS) have still higher resolution. If the BH detection is correct, then the apparent rotation and dispersion profiles should look steeper when they are observed at higher resolution. This is exactly what is observed. If the original dynamical models are ``reobserved'' at the improved resolution, the ones that agree with the new data have M_BH = (1 to 2) x 10⁹ M.}

Finally, a definitive detection is provided by the HST observations of the nuclear star cluster. Its true velocity dispersion is underestimated in Figure 2, because the projected value includes bulge light from in front of and behind the center. When this light is subtracted, the velocity dispersion of the nuclear cluster proves to be = 600 ± 37 km s^-1. This is the highest dispersion measured in any galactic center. The velocity of escape from the nucleus would be much smaller, V_esc 352 km s^-1, if it consisted only of stars. Without extra mass to bind it, the cluster would fling itself apart in ~ 2 x 10⁴ yr. Independent of any velocity anisotropy, the nucleus must contain an unseen object of mass M_BH 10⁹ M. This is consistent with the modeling results. The dark object is more than 25 times as massive as the visible star cluster. We know of no way to make a star cluster that is so nearly dark, especially without overenriching the visible stars with heavy elements. The most plausible explanation is a BH. This would easily have been massive enough to power a quasar.