Fits of C₁₀/μ as a function of μ for the 16 channels. In the top plot, the data and model (the full model of Eq. (20)) have been offset by 0.004 × c, where c is the channel number. We plot both individual measurements and averages at the same intensity, where the error bars reflect the scatter (compatible with the expected shot noise). The down turn at low flux of channels 8 and 12 is accommodated in the fit by the n₁₀ term in the model Eq. (20). The bottom plot reports the binned average residuals to both fits.