The same as Figure 10 but for the best three-circle fit that does not include a modulated power-law component. Because the spectral contribution from the modulated power law does not exist in this fit, the total background must account for these counts, and thus the data minus background and model minus background count totals are lower than they were in Figure 10. We also see that, without the modulated power law, the fit in the higher-energy channels is considerably worse than it was in Figure 10.