Was there really a time when most Americans were dying in their 40s (as opposed to 40 being the life expectancy from birth, which is a wildly different thing)?
I had no idea the two were different, so thanks for making that distinction. For anyone else's sake, I'll share what I learned:
Life expectancy from birth weights times of poor medicine and prenatal care towards the low end of the life expectancy scale because of much higher infant mortality rates, whereas in modern times life expectancy from birth would be weighted the opposite.
So basically, if you filter out infant mortality and other "first years of life" deaths, you get a more accurate picture of how long someone could actually expect to live if they make it past the initial thresholds to human existence.
Yeah, the problem is basically that historically you're most likely to die either very young or fairly old, but averaging out the ages at which people die creates the opposite impression — that you're likely to die sometime in your late 30s to early 40s, which was never actually the case.
Yeah, something weird is going on there. 72 years at birth for the Amish sounds absolutely crazy too. Maybe the Amish one is life expectancy at 5 or 20 and the other at birth.
It's possible that they simply have much lower rates of infant mortality. The linked discussion is from pay-walled researchers, so hard to see where they discuss their methodology.
Low incidence of alcohol use, strong communities with a tradition of midwifery, narrow genetic pool (which would tend to exaggerate any gene pool effects) are all things that could contribute to lower rates of infant mortality.
