Unravelling the mystery of the M31 bar

The inclination of M31 is too close to edge-on for a bar component to be easily recognized and is not sufficiently edge-on for a boxy/peanut bulge to protrude clearly out of the equatorial plane. Nevertheless, a sufficient number of clues allow us to argue that this galaxy is barred. We use fully self-consistent N-body simulations of barred galaxies and compare them with both photometric and kinematic observational data for M31. In particular, we rely on the near-infrared photometry presented in a companion paper. We compare isodensity contours to isophotal contours and the light profile along cuts parallel to the galaxy major axis and offset towards the north, or the south, to mass profiles along similar cuts on the model. All these comparisons, as well as position-velocity diagrams for the gaseous component, give us strong arguments that M31 is barred. We compare four fiducial N-body models to the data and thus set constraints on the parameters of the M31 bar, as its strength, length and orientation. Our 'best' models, although not meant to be exact models of M31, reproduce in a very satisfactory way the main relevant observations. We present arguments that M31 has both a classical and a boxy/peanut bulge. Its pseudo-ring-like structure at roughly 50 arcmin is near the outer Lindblad resonance of the bar and could thus be an outer ring, as often observed in barred galaxies. The shape of the isophotes also argues that the vertically thin part of the M31 bar extends considerably further out than its boxy bulge, that is, that the boxy bulge is only part of the bar, thus confirming predictions from orbital structure studies and from previous N-body simulations. It seems very likely that the backbone of M31's boxy bulge is families of periodic orbits, members of the x(1)-tree and bifurcating from the x(1) family at its higher order vertical resonances, such as the x1v3 or x1v4 families.