The east-west striped pattern of clouds in Jupiter's weather layer is accompanied by a zonal flow containing 12 eastward-going jet streams alternating in latitude with westwardgoing jet streams. Based on theory, simulation and observations of the Earth's oceans and atmosphere, it is conjectured that Jupiter's weather layer is made of bands of constant potential vorticity (PV), where the interfaces between bands are at the latitudes of the maxima of the eastward-going jet streams. It is speculated that the mixing of PV on Jupiter is analogous to the mixing of salt in the ocean by the Phillips effect, which causes the salt density to form a monotonic 'staircase'. It is hypothesized that the PV in Jupiter's weather layer is also a staircase, decreasing from north to south. PV is a function of vorticity, as well as parameters with unknown values, e.g. the vertical stratification and the zonal flow beneath the observable weather layer. Therefore, these hypotheses cannot be tested directly. Using an atmospheric model that contains these unknown parameters, we solved the inverse problem and found values of the unknown parameters (and their uncertainties) that best fit Jovian observations. The unknown parameters influence how the zonal flow interacts with large vortices, e.g. the Great Red Spot (GRS; the largest and longest-lived Jovian vortex, centred at 23• S) and the Oval BA (the second largest vortex, centred at 33• S). Although we found that the PV distribution is approximately piecewiseconstant and that the peaks of the eastward-going jet streams are at the latitudes of PV interfaces, there is also a PV interface at 20• S, where there is a westward-going jet stream. We find that the zonal PV is not a monotonic staircase due to the 'backwards' interface at 20• S. We show that this backwards interface is necessary to make the GRS nearly round, and that without that interface, the Red Spot would be highly elongated in the east-west direction and probably unstable.