Hi, here the recipe I would like to propose to address the hadronization uncertainty using the infromation contained in the data: 1) a) Measure the data-MC agreement concerning multiplicity in the recoil for the depleted sample. Under the assumption that we understand the physics, i.e. the production of charm in semileptonic decays we can assign any difference between data and MC to e.g. beam background. b) If one does not trust the MC concerning the production of high mass charmed states one could do the same for only D(*)lnu candidates which are consistent with a semileptonic decays. Any remaining additional particles contain information about things like beam background. 2) Knowing this kind of data-MC difference one can then make a data-MC comparison for the signal enhanced sample. Any significant difference observed in the multiplicity between data and MC are related to the hadronization model. Of course one should study this by varying the BF's for the exclusive modes in the MC. 3) To take possible differences into account one could then apply a reweighting of the MC as it is done by Kerstin for the unfolding. The systematic error which would then be quoted is driven by the statistical error on the multiplicities measured on (mainly) data. What do you think? Cheers, Heiko