I agree, that number does sound kind of high. The number of deaths per rifle would likely follow a heavy-tail distribution: a few rifles would have extremely hi

I agree, that number does sound kind of high. The number of deaths per rifle would likely follow a heavy-tail distribution: a few rifles would have extremely high numbers, and most would be have extremely low numbers. So the average would be a lot higher than the median. That doesn't rule out 2/rifle on average as a possibility. But the average is a misleading statistic if you're talking about the common case.