Beyond a certain time, data collection wasn't common enough, so my understanding is that people pick a baseline from a period of time when people were already c

Beyond a certain time, data collection wasn't common enough, so my understanding is that people pick a baseline from a period of time when people were already collecting data in substantial amounts. I guess 2nd half of 19th century, with all the industrialization, was when people both had enough resources, and were also interested in trying to actually collect it. I think this is similar to how the baselines for all satellite-measured data (like arctic sea ice, I think) start in the 70s. But the other comment is right too, which baseline doesn't matter so much, so long as the choice is reasonable, and used consistently. This is because, in the end, absolute values aren't meaningful on their own, only differences between them (similar to how one always reads about global temperature anomalies, rather than global temperature).