On the ephemerides of Uranus

We compute a new set of Uranus's initial conditions in order to investigate the nature of its observed systematic residuals. In addition, we pay special attention to the masses of the known perturbing planets. However, we find that the only way to fit accurately the observations is through the partition of the whole period of observations into two: one from the discovery up to 1897 and another from this last date up to the present. The problem is then the existence of a discontinuity in the orbital elements. Some speculations are made about the impossibility to fit accurately the whole orbit.