Contributions to χ² for a set of 81 best-fit models. Individual contributions (datasets) are shown in the panels (from top left): TTV, RV, SKY2, SED, SYN, VIS, CLO, T3. Every convergence was initialized with a different combination of masses m₁, m₂ (Ac1, Ac2) in the range 15–50 M_S, while m₃ (Aa1) was set to m_sum − m₁ − m₂ − m₄. All parameters were free during convergence. The axes correspond to the masses m₁ and m₂; the colours correspond to χ² (see also tiny numbers), with adapted colour scales: cyan best fits, blue good fits (< 1.2minχ²), orange poor fits (≥1.2minχ²). The factor was 3.0 for TTV, RV, SKY2. The forbidden regions can be clearly seen (e.g., high m₁, m₂ especially due to CLO), as well as correlations between the parameters (TTV −, RV +). The weighted very best fit for all the datasets is denoted by the red circle.