Two analyses, look for a signal, meaure the fractions of 1/9 and 4/9 MIP events to 1 MIP events.

To look for the signal, need to see the "Bumps" in uncorrected distributions of the average MIP signal from the WITNESS counters. May need some cuts, but data is pretty clean anyway.

To quote the fractions, need 3 numbers from the DATA samples

These numbers could be the number of events seen in windows near +/- 0.1 MIPs around the 1/9, 4/9 and 1.0 bins.

To measure the integrated luminosity in single MIP events

Use the scalers which counted the T1&T2&T3&T4 counts. As the cut off for the DATA events eats in to the MIP signal - we can get the number of MIPs from the scalers. The T1&T2&T3&T4 scaler was recorded per spill (and per run). This count is proportional to the number of MIPs seen by the experiment. The constant of proportionality can be measured for the MIP runs and applied to the DATA runs to measure the number of MIPs seen per run. We can also fold in the computer busy to this luminosity.

Uncorrected plot

Simply plot the average of the WITNESS counters in MIPS for all the data runs. There are no "bad events". Do we need to define the fiducial volume, cut on the "hole" scintillator ? (asking for a signal in each of the WITNESS counters is enough).
Total Number of DATA events
7 Witness Counters > 0.05 MIP
Average of WITNESS counters between 0.90 and 1.20

Method to extract the Luminosity in MIPS

MIPS runs

155,231 Events, Mean 1.038.

DATA runs

417,108 events, Mean=0.499

To Do

DATA runs for runs over 17

121,594 events, Mean=0.528

Removing the first 2 runs - when the detector was moved down, away from the beam line, there are fewer events which pass the cuts compared to the number of events in the runs. The distribution now has fewer bumps in it - mainly a bump at 1MIP, a central mound at 0.5 MIPs, and a small shoulder at .1 MIP.