I've made some progress toward validating the BHT3coincidence requirement's stability vs the dijet rate.  I took the largest mudst from each run included in the W run list, and counted the number of events that passed the dijet trigger (the L2jet algorithm with the requirement that the dijet bit be set in the L2Result) as well as the number of BHT3coin triggers (the L0 BHT3 trigger with the requirement of away-side energy with an opening of 10 bins (=1/3 of the phi range), as was first defined).

There is no BHT3 prescale

The number of dijets is multiplied by the JP1 prescale (50 across all runs of interest, apparently -- I thought it was changing dynamically?)

At the moment I have only verified the prescale for the standard JP1 - trigger ID 230410.  There are a few other trigger IDs that could be added in.  it's possible that one of these is responsible for the low patch in the plot below.

 Specifically, the runs that fall below a ratio of 0.065 are:

Low: R10087059, ratio=0.063958

Low: R10087110, ratio=0.055000

Low: R10088122, ratio=0.064500

Low: R10089080, ratio=0.061351

Low: R10095120, ratio=0.042972

Low: R10095121, ratio=0.050401

Low: R10096001, ratio=0.050464

Low: R10096002, ratio=0.047394

Low: R10096005, ratio=0.055789

Low: R10096006, ratio=0.051583

Low: R10096007, ratio=0.046429

Low: R10096008, ratio=0.052723

Low: R10096009, ratio=0.053578

Low: R10096014, ratio=0.055065

Low: R10096015, ratio=0.051432

Low: R10096019, ratio=0.055869

Low: R10096025, ratio=0.050281

Low: R10096026, ratio=0.052851

Low: R10096027, ratio=0.054968

Low: R10096139, ratio=0.033178

Low: R10096140, ratio=0.040067

Low: R10097038, ratio=0.038329

Low: R10097044, ratio=0.037293

Low: R10097086, ratio=0.023682

Low: R10097088, ratio=0.044306

Low: R10097090, ratio=0.057333