We have studied the ratio of linear fits to FM/wave8 and FL/wave8 in two runs with the same attenuation using an air scatter "flat" field. This should be insensitive to any wave8 effects. The current gain correction factors are 33:1. (I suspect this is just the ration of the design capacitances and does not take having the higher-gain capacitance always in place into account.) Looking at the inner four modules, where there was sufficient intensity to get near or above saturation in FM but being careful to stay away from the saturation region, we get individual fits with estimated uncertainty <1%. The ratio of FM/FL has a width of ~1.5% for each module. Three of the modules have a ratio about 3% higher than expected from the current correction factor (34/33 = 1.03 is suggestive that we have FM vs FM+FL) but one module is at 1.08. Each module has four asics which I assume are chosen at random. Any sensor effect should divide out. So the question is how a module can have a different FM/FL gain ratio than any other. Could that module have come from an entirely different batch of wafers where the capacitance ratio was off from design? Could it have been uniformly affected by radiation damage (no illumination is uniform at LCLS).