... errors

Detailed analysis of the data will be presented elsewhere (Wozna et al. 2002, in preparation).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... accurate

For presentation of detailed numerical results see e.g. Harding et al. (1997).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... occurs 

It is easy to reproduce the behaviour of $\theta_0/\theta_{\rm pc}$as a function of P in Fig. 7 with analytic formula: the maximal value of $B_{\perp}(r)/B_{\rm pc}$encountered in a static dipolar field by a photon emitted along the local field line at $\theta_0$is approximately equal to $0.1~ \theta_0$ (Sturrock 1971, also Fig. 1 in Rudak & Ritter 1994). This occurs always at $r_0 = 4/3~ R_{\rm NS}$, regardless the value of $\theta_0$. The angle $\psi$ between the photon propagation direction and the local field line at r₀ is $\psi_0 % \approx B_{\perp}(r_0)/ B(r_0)$ and therefore $\psi_0 \approx 0.1~ \theta_0 (4/3)^3$. To minimize $B_{\perp}^\prime (r_0)$ as much as possible in the case of rotation, the aberration angle due to local linear velocity $\beta(r_0) = 4/3~R_{\rm NS}/R_{\rm lc}$ should be close to $\psi_0$. Since $\beta(r_0)\ll 1$, the aberration angle is $\sim \beta(r_0)$. Therefore, we obtain the condition $0.1~(4/3)^2~\theta_0 \approx \theta_{\rm pc}^2$ which gives

$$\frac{\theta_0}{\theta_{\rm pc}} \approx 0.08~ P^{-1/2}.$$ (9)

This formula overestimates the locations of maxima in Fig. 7 by a factor $\sim$2 only.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.