For the observed 18 pairs of double neutron star (DNS) systems, we find that DNS mass distribution is very narrow and its mean value (about 1.34 solar mass) is less than the mean of all measured pulsars of about 1.4 solar mass. To interpret the special DNS mass characteristics, we analyze the DNS formation process, via the phases of HMXBs, by investigating the evolution of massive binary stars. Moreover, in DNSs, two classes of NSs are taken into account, formed by supernova (SN) and electron capture (EC), respectively, and generally the NS mass by SN is bigger than that by EC. Quantitatively, with various initial conditions of binary stars, the observed special DNS distribution can be satisfactorily explained.

A summary is given of the present state of our knowledge of High-Mass X-ray Binaries (HMXBs), their formation and expected future evolution. Among the HMXB-systems that contain neutron stars, only those that have orbital periods upwards of one year will survive the Common-Envelope (CE) evolution that follows the HMXB phase. These systems may produce close double neutron stars with eccentric orbits. The HMXBs that contain black holes do not necessarily evolve into a CE phase. Systems with relatively short orbital periods will evolve by stable Roche-lobe overflow to short-period Wolf-Rayet (WR) X-ray binaries containing a black hole. Two other ways for the formation of WR X-ray binaries with black holes are identified: CE-evolution of wide HMXBs and homogeneous evolution of very close systems. In all three cases, the final product of the WR X-ray binary will be a double black hole or a black hole neutron star binary.