Cumulative probability distributions for the same R = 13 km data as before, except that we have added a sinusoidal component with an absorbed ﹩{dN}/{dE}\propto {E}^{-1.25}﹩ power law, modulated at the stellar rotational frequency with a total number of counts in the 0.4–3 keV range which is the indicated fraction of the thermal counts from the star in this energy range. The solid red line shows the previous no-power-law fit. We fit the synthetic data using our usual model to determine the potential systematic errors introduced by the unmodeled modulated nonthermal component of the spectrum. Because the thermal component is eclipsed by the star, whereas the nonthermal component is an uneclipsed sinusoid, the tendency is to drive the fits toward smaller radii. However, the unmodeled component worsens the fit; for example, the best fit for a power law with 0.3 times the thermal counts has ﹩{\chi }^{2}=457.4﹩ for 387 degrees of freedom, which has a probability less than 1%. Thus, although there would be a bias toward small radii, there would also be an indication that something is missing in the fit. More work needs to be done, but this figure is an encouraging indicator that even a modulated but unmodeled component might not bias the inferred radius without being detectable by a poor fit.