On the left are two plots showing the number of pions per triggered event (L2-G trigger) as a function of run index.  The top plot is for all of the runs when the L2-G trigger had production status, while the bottom plot exhibits the runs for which the L2-G trigger had 'test' status (about the first 100 runs or so.)  On the right are histograms of these two plots.  To remove outliers from the runlist, I made a two-sigma cut around the mean of these histograms and removed any run that fell outside this range.  Below are plots showing pion number per triggered event after the outlying runs have been removed.  Note the change in scale from the original.  A list of excluded runs and thier properties can be found here.




As you can see, there is some funny structre in the above plot.  It's almost bi(or tri)-modal, with the mean number of pions per trigger jumping between .06 and .08 (or even .1).  I think this is a consequence of the SMD.  See below.


The above is a plot of the number of triggered photons that have GOOD SMD status normalized by the total number of triggered photons, as a function of run number.  While this is crude, it gives an approximate measure of the percent of active SMD strips in the barell for each run.  As you can see, the peaks and vallys in this plot mirror the peaks and vallys in pion yield above.  Since the SMD is not required to satisfy the trigger but IS required to reconstruct a pion, we would expect that the pion reconstruction efficiency would decrease as the number of active SMD strips decreases.  Indeed this is what appears to be happening.

I used Frank's pion trees to make these plots, imposing the following cuts on any pion candidate:

• Photon E_t > 0.1 GeV
• PT > 5.2 GeV
• Asymmetry < .8
• Mass between .08 and .25 GeV
• No charged track association.
• BBC Timebin = 6, 7, 8, or 9.
• 'Good' SMD status.