Hmm, from my personal testing/comparison of war DA vs Blu DA (or rather a confirmation that war da is in fact 10% within a .05% error range vs blu's 7%) brings me to a conclusion that Hideka's sample is too small. I recall that 10% DA had an error variance of 2-3% in the 500-700 range, and a 1-2% in the 700-1000 range, that didn't display correctly until past the 1000 rounds mark. This would go a long way in explaining the variance of Hideka's numbers to Byrth's numbers since his are >1000 compared to Hideka's 500~550 range.