Representation of Φ0(MG, (B − R)), as in Figure 10, but augmented by 1σ upper limits across the entire (MG, (B − R)) plane, where the sample does not contain any entries. Those limits were derived by calculating the density that even one single object in this (MG, (B − R)) pixel would imply. Knowing the selection function for counterfactual properties q permits us to place constraints in models where we do not see any data in the sample.