I thought I'd write down the question BobK asked this morning before I forget. Heiko explained to me that there is still a difficulty with mES fit. My understanding is that it is difficult, in mES fits on small samples, to separate the signal and the peaking background reliably. Bob's questions boil down to the following: 1) Why is it not possible to fix the peaking background using MC prediction, and assign systematic error on this by varying inputs to the MC? Does it lead to excessive systematic errors? If so, 2) Does it make sense to tighten the purity cut of the Breco tags to reduce the background, and thus the systematics? You lose statistics, of course, but isn't it possible to find an optimum that is tighter than what you use now? Masahiro -- Masahiro Morii <[log in to unmask]> Harvard University, Department of Physics 17 Oxford Street, Cambridge, MA 02421, USA