Residuals as a function of the orbital phase for the same ephemeris as Fig. 4. Top panel: we display the residuals predicted by the model but without taking into account the Shapiro delay. Lower panel: we see the same while fitting for other Keplerian parameters. As we see, some of the Shapiro delay is absorbed by this fit, but some still remains; this remnant represents the ‘measurable’ part of the Shapiro delay; this is given, in the limit of perfect orbital sampling, by Eq. (31) of Freire & Wex (2010). In this plot, the Nançay residuals are displayed in black, the Effelsberg residuals in red, and all others in grey, all without error bars for the sake of clarity.