Hi, I added a couple of plots into the same file: http://www.slac.stanford.edu/~kerstin/vubfit/VR122104.pdf I added: * Resolution plots produced with generator mix signal MC * A second version of the resolution plots, using an additional cut on Qtot=0. For signal MC I see some inprovement in the bias and the resolution, for B->Xclnu from generic I see some improvement between using the cut on Qtot and not using it. I recognized that the binning is quite different from the plots in 540v9, taking this into account there does not seem to be such a big difference. * The covariance and correlation matrices for the statistical uncertainties (including MC statistics) on the unfolded spectra using mxhad and mxhadfit. In both cases the first five bins and bin seven to the last bin are positively correlated among each other. So this fits together with the pattern we see in the unfolded spectra. I had a look at the last filled bin in the bias and resolution plots for nonres signal MC. This bin is only populated by 1 event, so whatever value for the bias and the resolution this event has, it will be just consistent with zero (unless my determination of the uncertainty is wrong). I am working on the plots Daniele suggested (split up by neutral multiplicities), I will add these once they are done. Cheers, Kerstin