1999ApJ...520..696A

From considerations of spin evolution and kinematics in the Galactic potential, we argue that the pulsars B1913+16, B1534+12, and B2127+11C may be younger than previously assumed, and we find that a lower bound on the formation and merger rate of close double neutron star binaries is 10^–6.7f_b^–1f_d^–1 yr^–1, where f_b is the beaming fraction and f_d accounts for the possibility that the known double neutron star binaries are atypical of the underlying population (e.g., if most such binaries are born with shorter orbital periods). If we assume no prior knowledge of the detectable lifetimes of such systems, the rate could be as large as ≃ 10^–5.0f_b^–1f_d^–1 yr^–1. From both plausible bounds on f_b and f_d, and a revision of the independently derived limit proposed by Bailes, we argue that a firm upper bound is 10^–4 yr^–1. We also present a unifying empirical overview of the spin-up of massive binary pulsars based on their distribution in spin period P and spin-down rate P{dot}, finding evidence for two distinct spin-up processes, one dominated by disk accretion, the other by wind accretion. We argue that the positions of binary pulsars in the P-P{dot} diagram can be understood if (1) there exists a Galactic population of pulsars in double neutron star systems with combinations of spin and orbital periods that have prevented their detection in surveys to date; (2) recycled pulsars in wide-orbit binaries are not born near the canonical spin-up line in the P-P{dot} diagram because they were predominantly spun up through wind accretion; and (3) there exists a disfavored evolutionary endpoint for radio pulsars with spin periods 5-30 ms and P{dot}>10^–19 s.s^–1.