Just to be more precise, I show the signal which should measure the vertical position of PR mirror. It is not in the data: it is computed adding Sc_PR_F7_Y (vertical position of F7) and Sa_PR_F7_LVDT_V multipied by 2 (vertical displacement of the last stage; the factor of 2 is there to compensate a wrong calibration, which is anyway quite uncertain). In fact, after the manual recovery and a slow adjustment of ITF loops, the residual jump was 40 microns. This is not explained, but not huge; it should not determine a relevant change in the ITF working point.