since the phone meeting was cut, here are some remarks which I wanted to
make [hoping that they are not too obvious to you ;-)]

Daniele's list is quite exhaustive, but I find more clear to factorize
the systematic uncertainties into 4 categories:

1) MC parameterization of B->D l nu decays, in terms of absolute/relative
D, D*, D** BRs and distributions
2) MC parameterization of D decays (cascades and topological BRs)
3) MC signal composition (resonant vs. non-resonant relative normalization
and distributions)
4) detector effects (PID, resolutions in Mx, etc.)

All the quantities used in the analysis (BG_u, BG_sl, the
efficiencies) are affected by effects from the above categories, each
one in different proportions. For instance, Eff^u_sel is mostly affected
by 3) whereas BG_u is affected by 1), 2) and 4). I find that thinking in
terms of these categories allows also to take correlations into account
and avoid double counting/overshooting in a more natural way.

Morevoer, concerning 1) and 2), we already have some guidance and
estimates, based on the V_cb analysis [BAD 218].

I have not mentioned the uncertainty from the M_es fits, but it is
trivial, isn't it? :-)

Ciao, Concezio.