Plotted below is the change in the distribution of track momentum after cuts are applied.  The black histogram shows the momentum distribution after all TPC- and BEMC-related cuts have been applied -- i.e., after the standard tower calibration cuts -- but before any BSMD cuts have been applied.  The remaining histograms show the distribution after each cut.  The big change occurs between red and blue, when I require that at least one of the 5 strips surrounding the strip that the track projects to in the BSMDE has ADC>1.5*ped sigma.  This cut greatly reduces the number of low-momentum tracks while not affecting the high-momentum tracks all that much.  This seems to indicate that the low-momentum tracks are actually suffering from significant mip contamination.  The cuts are, in order of application:

All five strips must have good status (red)

At least one of the five strips must have ADC>1.5*ped sigma (blue)

The cluster must have at least two strips (green)

The cluster must be isolated (yellow)

The distance from the track to the cluster centroid must be less than one strip width (cyan)

There must be a cluster passing all these cuts in the BSMDP plane (violet)