Hi, the tag from RecoilAnalysis I was referring to is KT022304 (this was used in 12.5.2). This version has the asymmetric Gaussian for the smearing of the B->D(*,**)lnu branching fractions. Here the seed gets set as I described in the meeting, so we would submit jobs with -Sys i, i>=3, and choose a different i for each job. The >=3 is actually important because of how we use it with VVF. The i determines how many of the random numbers thrown from a fixed seed will be discarded, so we do not use any numbers (or worse, sets of numbers for the different weights) twice. We can change this to use the old method if you prefer. I do not have a strong opinion, this was just my way to make sure we have an ok distribution of the weights in the end. I assume I should use PDG2004 to check if we need to update the branching fractions for B->D and D->XY. If this is not correct, please let me know. I looked through Antonio's plots and have a question: Why is the first bin so different in CM1 and CM2 for the mxhadfitleptoncuts plots? Does this mean we have much more pis in CM2? Is it due to differences in the kinematic fit (the mxhadleptoncuts does not show such a big difference)? Cheers, Kerstin