Hi all,

 I repeated the study to evaluate the theoretical uncertainty.

 First, these are the plots that show the uncertainty due to the Mx cut
(the same of the BAD)

 http://www.slac.stanford.edu/~daniele/vub/sfDependency-0.15-allcuts.eps

 (second raw is new and rapresent the systematic effect due to the a
parameter)

 http://www.slac.stanford.edu/~daniele/vub/sfError-0.15-allcuts.eps

 (second raw is new and rapresent the error due to the a parameter)

 Then, in conclusion, systematics on the eps_mx efficiency is around
15.2%.

 Unfortunately this is not all we have. In July an additional effect was
neglected. Since the efficiency is very dependent on Mx_true (see
http://www.slac.stanford.edu/~daniele/vub/mxhadgenvubtheefficiency.eps)
a reweighting will affect the overall efficiency (named eps_u)

 In order to take into account properly both the effects (eps_u and
eps_mx) I implemented this reweighting in the fit.

 The results are

 mb=4.950 (a=1.290)    BRBR  -25%

 mb=4.650 (a=1.290)    BRBR  +30%

 a=3.6 (mb=4.800)      BRBR  +9%

 a=0.380 (mb=4.800)    BRBR  -8.5%


 Then in conclusion the theo. uncertainty should be +31% -26% on the
ratio(BR). In July we quoted a factor two less (15%). This difference
derives, as mentioned above, from the uncertainty on eps_u that is around
+-15%

 With this uncertainties we are fully dominated by sistematics. A looser
cuts on Mx will not help too much since the uncertaity is due to the
overall efficiency.
Then two questions:
 - may the multiplicity fit help (in principle it is less affected by
efficiency)? I will check this.
 - may we use a more precise values of mb and a?

 Daniele