It's about all the mundane people who didn't die as babies. In earlier centuries, the life expectancy at birth (which is the metric that's about 35 years for that period) is dominated by the relatively high childhood mortality, all the many babies dying within the first days or weeks, and is very different from the expected lifespan of adults.

For example, here's a dataset on USA historical life expectancy https://www.infoplease.com/us/health-statistics/life-expecta... - if we look at the 1850 data, then the average life expectancy at birth is 38 years but that does not mean that people live until 38, as a the life expectancy for a random 20 year old man is 40 more years, so people (who didn't die in early childhood) lived (on average) to the age of 60, not 38.

There's data from Roman Empire where life expectancy was 25 years at birth, but life expectancy is average 53 years upon reaching age 25 - so the intuitive metric "how long do adults live" is twice as large as the technical life expectancy at birth.

Here's an archived resource that goes into more detail https://web.archive.org/web/20070713083310/http://www.plimot...