The calibrations constants for each ADC were generated by cutting on 400<=Ephoton<600 as there seems to be little or no contamination from the unknown "junk" (probably electrons). The veto information seems quite useless but I cut on tveto<-100 anyway as this removes 0.1% of the garbage. Applying the new calibration one gets the above image which is normalized so that there are an equal number of events under each curve. They do look very similar now but the JLab data is slightly wider "squashed" and extends further into the higher and lower energies presumably due to the lower resolution as the simulation data is idealized where we only have to concern ourselves with the shower fluctuations and the sampling fraction of each shower due to the sampling calorimeter design of the BCal.
It seems that the mean of the JLab data fit is systematically 10-12 MeV smaller than the simulation data. This does not seem to be a function of the scale factor as there is no energy dependence, i.e the difference does not increase as the energy increases as one would expect if the scale factor was too small.
