In the Monday EMC meeting, a concern was raised regarding a possible eta dependence to the new BEMC calibration systematic calculation.  The fear being that such an effect would mean the quoted 1.3% systematic bias would be an underestimate for any systematically extreme eta rings.  

The difficulty in constraining any such effects is that, as the data sample is partitioned into eta bins, the statistical variance of each partition will grow.  Increased variance alone, then, cannot be considered sufficient evidence for a systematic effect.  What is needed is an increase that exceeds the expected statistical growth.

To this end we studied the RMS of the E/p peaks as the calibration data set was partitioned into eta bins.  If no systematic effect is present, then  the RMS should grow as the square root of the number of partitions only (technically the growth can fall below the square root due to the already known systematic bias).

In Figure 1 we see the growth of the E/p RMS for the nominal calibration sample.  We have also included a second sample specially modified with a systematic shift to the E/p peaks, -1% for |eta| < 0.5 and +1% for |eta| > 0.5.  The eclipse of the square root growth is clearly evident in the modified sample, where as the calibration sample follows the expected behavior.

Hence we can confidently claim that there is no evidence of an eta dependence to the calibration uncertainty.