THE SHAPE OF THE GALAXY

We propose that a rotating triaxial spheroid is the source of the large-scale asymmetries observed in the l-upsilon-diagram of the H I distribution in the outer Milky Way. We show that a quadrupole moment in the gravitational potential of the Galaxy is required to provide a dynamical explanation for the kinematics and the distribution of the gas. We construct numerical models for gas motions in a nonaxisymmetric potential, project these models onto the l-upsilon-diagram, and compare them with observations. The asymmetries in the l-upsilon-diagram in the outer Galaxy imply that the gas there moves on nearly circular orbits and that the local standard of rest (LSR) is moving outward at approximately 14 km s-1. The character of the outer gas contours and the kinematics of stars in the local neighborhood suggest that this outward motion is not due to a local perturbation (e.g., spiral arm) but rather is due to a global oval distortion. The near-symmetry of the terminal velocity curve, observations of the 21 cm absorption feature toward the Galactic center, and the magnitude of the outward LSR motion imply that the distortion originates inside the solar circle and is slowly rotating. We show that the overall distribution of gas in the outer Galaxy is asymmetric; there is systematically more gas in the second quadrant than in the third, and more gas in the fourth quadrant than in the first. All of the asymmetries in the gas distribution can be understood in the context of a model with a quadrupole term that is a decreasing function of galactic radius. The best fits of the three free parameters of the model are 0.02 for the amplitude of the quadrupole term, 6 km s-1 kpc-1 for the figure rotation speed, and 45-degrees +/- 20-degrees for the position angle of the long axis. The position angle of the streamlines of the gas resulting from the model are the same, within the uncertainties, as those found for the H I within 3 kpc of the Galactic center, suggesting a common dynamical origin for the kinematics of all of the atomic gas. The model is shown to be consistent with a large body of observational data, some of which had previously been a puzzle, including the near-zero radial velocity of the 21 cm absorption feature toward the Galactic center, the approximately 15 km s-1 offsets of the centroids of the molecular emission in the inner core of the Galaxy, the streaming motion of the spheroid stars, and the values of LSR corrections obtained from relatively local molecular and stellar observations. The model also implies that estimates of the circular speed are uncertain by as much as 20% and that kinematic distances are uncertain by as much as 0.5 kpc. We provide an explicit prescription of correcting kinematic distances estimated in an axisymmetric model. The model implies that the outer parts of the disk are quite round and that the dark matter halo is very nearly axisymmetric at large radii, b/a > 0.87, which in turn implies that the velocity distribution of dark matter particles, whether they are brown dwarfs or exotica, cannot be extremely anisotropic.