I wrote a quick analyzer that calculates the true number of t6 events in a whole run, not just the replayed portion. You can call it by running e.g.
% goonlana % ./replayt6 2005
It will try to find the run number and number of events by three methods in this order 1. look for stored information for that run in a file temp_t6log.txt 2. look in a text file rootfiles/apex_t6_runnum.txt produced by previous call of replay/apex_t6new.C 3. calls replay/apex_t6new.C, which cuts on (D.evtypebits&64)==64 and reports resulting count.
The script also writes this count,along with the run numbers, to temp_t6log.txt (as long as ian entry for that run isn't already in the file).
I've tabulated the T6's from the following sets ofruns: 1822-1832 1850-1856 1990-2007 2011-2036 (except 2023, which I had trouble reading)
and find 1.75 million events total: 740K in production before optics studies, quality may not be uniform 130K in 70uA friday running 200K in 100uA running (sat. owl, day) 670K in 120-130uA running (sat. day, swing) You can try to account for accidentals by looking at representative runs in each configuration.
Two other small notes: * I also think Jixie's script over-counts T6 events because some events are double-counted in the D.t1 histogram.
* true coincidence fraction varies between 50 and 90% from one run to another, while DAQ deadtime is almost always a much smaller effect (<5%), there's no reason to be this careful instead of just looking at the scalers.