Stellar wind and stellar disc models of dispersion and rotation measure variations in the PSR B1259 – 63/SS2883 binary system

Two models of the PSR B1259 – 63/SS2883 binary system are evaluated to determine whether they account for the orbital variations in dispersion measure (DM) and rotation measure (RM) observed during the 1994 periastron passage of PSR B1259 – 63. In the first model, the DM and RM variations are produced by an ionized, radiatively driven, coronal wind from SS2883. The electron density n_e and magnetic field B vary as power laws with distance r from SS2883, and the wind terminates where it collides with the relativistic, electron-positron wind from PSR B1259 – 63. It is shown that this model - popular in the literature -cannot reproduce the observed pre-periastron rise in DM unless one has $$n{_\text e}\propto r^{-5}$$, with $$n{_\text e}\sim 10^{14}\enspace\text{cm}^{-3}$$ at the stellar surface, contrary to the predictions of dynamical wind theories. In the second model, the DM and RM variations are due to a cool, dense, circumstellar disc with an exponential density profile in the radial and vertical directions. Good agreement with the observations is achieved for a disc inclined ~10° to the orbital plane, with n_e ~ 10⁸ cm^–3 and B ~ 10 kG at the stellar surface, and radial scale size $$r_\text D\sim12r_\ast$$. The data do not discriminate between a thin disc (scaleheight h_{D ≪}r_D) and a thick disc (h_D ~ r_D). The disc model correctly predicts the duration of the pulsar eclipse if free-free absorption is responsible, and accounts for the depolarization of the pulsar signal by stochastic Faraday rotation given rms electron density inhomogeneities satisfying $$\left \langle \delta n^2_\text e\right \rangle ^{1/2}\sim0.2n_\text e$$. It is shown that the empirical value of r_D is in accord with theory, and that the pulsar-induced component of the wind from SS2883 is weak.