Showing the best fitting models, in red, for the complete (top), over-complete (middle), and under-complete (bottom) model cases as described in Section 9. The best fitting models are offset along the ordinate for clarity and over-plotted on the “observed” spectrum in gray. The global Bayesian Evidences, log(E), are given for each case, quantifying the adequacy of each model given the data. As expected the evidence strongly favours the correct, complete model.