Hi,

actually, I think that also the 1D in mX does not look too bad. It seems 
to be consistent with what we observe in the 2D plots. Please don't forget 
that the flat behaviour there is seen as a function of q^2 in a given 
range of mX. The ratio as a function of mX is not constant and it can't be 
given the fact the variation we see in 1d-mX. The important thing is that 
we can control this ratio and I think we can.

For 1d-mX: the double ratio for data/MC-signal-enriched is consistent with 
what we observe on the depleted sample. The only difference is that the 
errors for the enriched sample are much larger and hence the point-to-point 
fluctuations are larger. Therefore, I think we can defend fixing the ratio 
using the MC-enriched sample and apply a data-MC correction using the 
depleted sample.

Here is my proposal:
a) Use the signal/peak.BG ratio from MC-signal-enriched
b) Multiply the number with the double-ratio data/MC-signal-depleted
c) Fix the signal/peak.BG ratio obtained in this way in the fit
d) Systematic uncertainty: Error propagation using
   * the uncertainty on the signal/peak.BG(MC-signal-enriched)
   * the uncertainty on double-ratio data/MC-signal-depleted

Things to be discussed in the meeting:

* The P0 fit Antonio is calculating should be consistent to a good 
  approximation with what we find in the 1d-mX case.

* Fixing the endpoints of Argus and CB to the same parameter value:
  Do we have already checked if this leads to a larger or smaller bias wrt 
  the number of expected signal events?

* Is the fitting code already able to fix the ratio?
  If yes, Chukwudi should start running the mm^2 scans today once Antonio 
  has provided the ratios and the code.
  
* Of course we need to start studying P+ as well.

Cheers,
Heiko