- The track-based primary vertex does not appear to have had any effect
  on the calculation of jet E_T!  At least, the number of events accepted
  by JT_95TT and JT_70TT_JT95PV are identical and the same is true for
  JT_125TT and JT_95TT_JT_125PV.  (Chris, Volker?)

- The aim of running an unprescaled wide region single muon trigger
  is acheivable with MU_W_L2M5_TRK10 as soon as the trigger/DAQ can
  stand the rates without producing unacceptable front end busy
  rates. (The rejection factors are: 12 at L2 and 60
  at L3; the L1 rate is ~450 Hz at 2*E31, so the rate to tape would be
  less than 1 Hz.)  The equivalent "di-track" trigger MU_W_L2M0_2TRK3
  has rejection factors: 5.5 at L2 and 11 at L3, so that could run
  with a small prescale too.

- The di-mu L2 trigger 2MU_A_2L2 has a rejection of 20, which would give
  a rate to tape of 35 Hz at 2E31, so we need some rejection at L3.
  Somebody should try requiring 1 or two soft L3 tracks in the simulator.

- It's not possible to get the .or. of the 5 single electron filter
  scripts with the current monitor.  Moacyr is working on making this
  available in the L3 monitor.  However, the overall rejection
  on CEM(1,10) is somewhere in the range 10-20 (depending on the
  overlap) and the overall rejection on CEM(2,5) is somewhere in
  the range 15-40.  Ulla, could you look offline and get the actual
  rejections?

- The Gap triggers had very poor statistics.  The L3 called/fired rates
  were:
    JT_15TT_GapN   53/2
    JT_15TT_GapS   26/1
 (Unfortunately the summary file was incomplete and didn't have the
  GapNS numbers.)  I think that the new prescale file will give a bit
  more data for these triggers, but those prescales could probably be
  reduced still further.  The rejection of these filters (with gap) is
  MUCH larger than the factor of 1.4 that is observed on straight
  CJT(2,3)CJT(1,5).

- A large fraction of the ~40 Hz rate to tape is given by unprescaled
  triggers (good) that have little or no rejection at L2/L3 (bad).
  E.g.,:
  Trigger              L2 rej    L3 rej  Rate to tape at 2E31 (HZ)
  CEM(1,15)             1         1.9    10.5
  mu2ptatxx_fz          1         1       7.6
  mu1ptatxx_CEM(1,5)    1         1       7.0
  mu1ptatxx_CJT(1,5)    1.3      ~1.5     5.0

  If CEM(1,15) is going to persist (if we can't cure the Front End busy
  problems such that CEM(1,10) can run unprescaled) then we should add
  the new standard L3 single electron filters to CEM(1,15).
  We should look again at the absence (or low rejection) of L3 filters
  on some of the others.

  The L3 rejections on the CEM(2,2.5) and CEM(3,2.5) triggers are
  only around 4 and 2 respectively.  These rejections are
  much lower than needed to make these triggers viable.  Surely we should
  raise the E_t, p_t thresholds at L3 to get a higher rejection in
  order to lower the prescales.  (If we're requiring 2.5 GeV at L1 then
  surely we should be requiring at least a bit more at L3!?)