Since my presentation to the Caldet meeting on June 27 I have looked at the actual ADC distributions for the Light Injection Snarls to understand whether the effect I presented previously is a shift in the whole distribution or a tail. From the plots below I think it is very clear that the distributions are shifted high for those channels that see the most beam. This effect it very clear for the Near electronics but is not seen in the Far electronics. Having shown these plots to Peter and Alfons they think this is due to conditioning of the M64 PMTs.
The plots shown have 2 histograms. The solid black one is the reference run I have used to normalize the responses. This is run 31131. The red dashed histogram is run 31401. I have looked at other runs and they all look similar. Each distribution has been normalized by the corresponding pin diode value to remove variations in light levels. Each plot shows the mean and RMS for both runs.