I have repeated the recoil vs reco study with higher statistics and with a reweighting that improves the agreement in the energy spectra of the two sides (reweighting by the ratio of the two spectra: here is the outcome: http://www.slac.stanford.edu/~rfaccini/phys/vub/aug02/EgamPhot_high.ps ). The plots are linked to the usual page http://www.slac.stanford.edu/~rfaccini/phys/vub/aug02/midSum.html (I replaced the old ones) In particular the lower cut on LAT (>0.05) is justified by the low energy bin: http://www.slac.stanford.edu/~rfaccini/phys/vub/aug02/LmomPhot_low.ps while the upper one (<0.5) by the high energy bin http://www.slac.stanford.edu/~rfaccini/phys/vub/aug02/LmomPhot_high.ps In order to quantify the impact of this cut on the photon selection, I have computed the fraction of photon candidates that pass the selection for the high and low energy sample and for the reco and the recoil candidates separately (by fitting individual Mes distributions): low energy high energy reco 75.3+/-0.7 94.8+/-0.4 recoil 69.8+/-0.3 83.3+/-0.3 The cleanup is therefore more effective at high momenta than at low ones. S9S25 is instead shown in http://www.slac.stanford.edu/~rfaccini/phys/vub/aug02/S9S25Phot_low.ps and http://www.slac.stanford.edu/~rfaccini/phys/vub/aug02/S9S25Phot_high.ps When fitting Mes for low and high S9S25 (the cut is at 0.9) one gets low energy high energy reco 97.5+/-0.3 93.1+/-0.4 recoil 85.5+/-0.3 88.7+/-0.3 The low energy sample is cleaned up by this cut, which does not show up in the plots because of the underflows (will enlarge the plot for the BAD) ciao ric