Reminds me of a chat I had with a ballistics engineer at Lake City Army Ammo Plant in the late 1960's regarding a lot of M118 7.62 Match ammo that had an occasional light powder charge; 4 or 5 rounds out of a box of 20 would shoot 10 to 20 inches low at 600 yards. We talked about their testing process where they shot from 100 to 200 rounds per test group then calculated the mean radius; a very good way to assess accuracy.A group of twenty shots is reasonably relivant, but a tiny group of three or four or five shots is more likely a statistical fluke meaning nothing at all. Averaging a series of five shot group sizes also means nothing except in BR competition, only groups that include ALL of our hits has any real world significance!
He mentioned one of their new hires was learning about this test helping him do QA on a lot of M80 ball ammo. After shooting many dozens of rounds from their machine rested test barrel, they went down range to check the 600-yard target. That many-shot group was about 18 inches and he commented to the new guy: "Look at all those 5-shot clusters that are an inch or less in spread." The new guy replied by saying: "Yes, there's several of them. We should tell everyone how accurate that lot of machine gun ammo is. But it'll be hard to prove it by firing only 5 rounds." The old engineer then said: "You're a good man; you'll do well here at Lake City."