Hβ light curves of 3C 120 with different non-variable contributions C between 0% and 50% subtracted and re-normalised (red). The best model fit is shown in black and the best corresponding data in green. The inset shows the reduced chi-square as a function of C. The best chi-square is for C ~ 25%, perfectly consistent with the C-value from the long-term trend analysis in Fig. 6.