Hi all, I studied a bit more in detail the problem of the instability of the result as a function of the hadronic mass cut. As you know, our biggest systematics is the theoretical one and it is very dependent on the Mx cut itself. This implies that it can introduce a slope in the Mx scan. I tried to reweight the MC model shifting the theoretical parameters mb and a in order to check the impact on this scan. The hope is that the scan is just telling us that mb and a are not the ones in the MC model. In http://www.slac.stanford.edu/~daniele/vub/theoscan.html you find the scans in many configurations. The first five tables show the data for the following configurations: - default - mb = 4.65 (-0.15) , a = 1.290 (default) - mb = 4.95 (+0.15) , a = 1.290 (default) - mb = 4.60 (-0.20) , a = 1.290 (default) - mb = 4.80 (default) , a = 3.1 (+2.31) - mb = 4.80 (default) , a = 0.91 (-0;19) As you can notice, moving mb down and a up, fixs part of the discrepancy at low Mx low values. As a crosscheck, I verified the impact on the generic MC, that is shown in the second set of tables. (in order to reproduce the effect on data, you should imagine to flip the slopes and the shifts since here I am reweighting the model not the fitted sample) My comments: 1) moving the theoretical paramenters (down mb and up a, but in particular a) recovers part of the problem at low Mx. 2) generic MC is able to reproduce the effect at low Mx. 3) generic MC shows that the B0s are more sensitive to this systematics. Probably this is due to the slightly better S/B?? 4) our measurement is so clean and we have so many events that we are sensitive to these theoretical parameters. We can start thinking about fitting a and mb. 5) I am a bit less worried about this slope now Daniele