Daniele and I have been trying out the proposal from Franz ( http://babar-hn.slac.stanford.edu:5090/HyperNews/get/pubboard54/26/1/1.html ) to discriminate B0->D*lnu events with the same variable used in the partial reco analysises (see for istance BAD #287). This variable (called prmm2 in the following and peaked at 0 for D*lnu events ) seems to be quite powerful: the comparison between Vub and Vcb events on B0 is shown in http://www.slac.stanford.edu/~rfaccini/phys/vub/deltam/comparison7wdeltamallcuts-sig.ps (histo == Vcb) The corresponding plots for charged Bs show that the signal at 0 is genuine: http://www.slac.stanford.edu/~rfaccini/phys/vub/deltam/comparison7wdeltamallcuts-sigbch.ps The data-MC agreement on B0, shown in http://www.slac.stanford.edu/~rfaccini/phys/vub/deltam/comparison7wdeltamallcuts.ps ,is quite good, The effect of a cut at prmm2<-2. GeV^2 on the Mx scan for B0 events (the cut would be applied only to them) is shown in http://www.slac.stanford.edu/~rfaccini/phys/vub/deltam/prmm2effect.eps (open dots after the cut, full dots, before). The effect of the cut is clearly beneficial, the slope being reduced ~40%. It is amazing that our analysis is sitting in the only stable point. Daniele is performing the scan of this variable and the optimization. The final cut is more likely to be ~4GeV^2 ciao ric