We calculated the Photomultipler Gain  HV at the high voltage HV applied by measuring the response at different led amplitudes.
For each HV, the fit of the variance vs pulse height peak (The so called "Response line") gives us two parameters:
For each HV, the slope S is proportional to the Gain   HV:
The factor depends on delta, where delta comes from fit of pm single photoelectron distribution. BUT:
In absence of a correct method for fine mesh single photoelectron spectra interpretation, we have decided to normalize our GainHV=2000 to the nominal gain given by Hamamatsu factory.
(If you want to know how data were analysed, please read PM data analysis )