The above plots show the Pt separated mass plots for the data (black) and the mixed event combinatoric background (green.)  For more information on how the background was modeled, see this blog post.  The background shape has been normalized to the data between 0.8 and 1.2 GeV.  In the first three bins, the background follows the high mass tail very well and also seems to account for most (if not all) of the mid-mass background between the pion and eta peaks. 

**There is one change I made to the background distribution for these plots.  Normally, I allow any event with a Zgg less than 0.8.  For these plots I also set a lower bound at .35 for Zgg.  This is supposed to protect against mixing a trigger-tripping photon from one event with a trigger-tripping photon from the second event.  My thinking is that two photons that have similar energy (above threshold) they would most likely both be trigger photons.  These would have a small Zgg.  For example...

The above left plot is Zgg from the mixed mass background.  The right is real data.  Notice the real data is heavily weighted towards high Zgg values while the mixed mass plot is bimodal.  Taking a Zgg cut at 3.5 eliminates the low Zgg events which do not show up in data.