I have been investigating the problem of neutral candidates that originate from tracks to whom they should be matched but are not. Two variables are used to quantify the distance between a neutral cluster and a tracks: dph and dth, the difference between the phi(theta) of the track extrapolated to the EMC and the cluster centroid one. The scatter plot of the two (zoomed around (0,0)) is shown in http://www.slac.stanford.edu/~rfaccini/phys/vub/aug02/dthdph.eps You can see that random track-photon associations would show a depletion around (0,0) , but close to the origin there is a bump. This can be better seen in http://www.slac.stanford.edu/~rfaccini/phys/vub/aug02/dph-dth32-dtmc.eps where dph is shown for -0.03<dth<0.02 rad (top: cocktail, bott: data) and in http://www.slac.stanford.edu/~rfaccini/phys/vub/aug02/dth-dph44-dtmc.eps where dth is shown for |dph|<0.04 rad There is clear evidence of unmatched neutral candidates. The data seem to be reasonably reproduced by MC, but it is probably adviceable to remove them. In conclusion,I would suggest we throw away neutral candidates if there exists a track in the event with |dph|<0.04 -0.03<dth<0.02 rad ciao ric