Example of a successful circlipse fit and Kerr hypothesis test. Left: every 5°, we determine the possible diameters d_φ of the n = 2 ring by maximizing the goodness-of-fit measured g(d_φ) (Eq. (36)). For each angle φ, we obtain a periodic set of diameters separated by Δd_φ ≈ 1/u_w, with likelihood proportional to g(d_φ) and indicated by the coloring of the data point. Right: circlipse fit for each of the possible rings. The darkest circlipse is most likely to be the true shape of the image ring, since the corresponding “joint goodness of fit” far exceeds that of the other solutions.