When I read about a 5%powder charge variance being OK I was a little skeptical, but I'm coming around. My real concern was in the allowable error of the other components. Since I was having trouble getting my Tikka t3 in .223 to shoot well with hand loads I started weighing brass cases and j-bullets. The results were interesting. So here is the last group of cases that I'm going to weigh. If we sort by headstamp we can easily achieve close enough tolerances in case capacity. All cases were trimmed to length the same , de-primed ,and tumble cleaned. I'm using case weight to indicate case capacity.
PMC brass,.223 had a 3.5% weight variance
LC military mixed lot but recent years, 2.4% weight variance
here's the lulu: 20 different headstamps mixed military and civilian, 10.9% weight variance
Lapua .223 2.0% variance
Hornady cases from their match ammo 6.8%
So what is becoming evident is that 5% is both OK and easily achieved, and if you want to do better that too is easily achieved up to certain point and then consistency becomes ever more difficult.
I hope this adds to the body of knowledge, Ross