Standard analyses of ZEUS data usually remove all calorimeter cells with a large imbalance (see (5.1)) from the event data. MUFFIN instead keeps all those cells. The reason is that the probability for a coincidence of a spark and a background muon is very low. So if there is a cell with large imbalance this is more likely due to the muon than to a spark, for example because of a muon bremsstrahlung near the wave length shifter.