When dealing with some timezone bugs, I read a rant about historical timezone reckoning in the timezone library source. Now that it's transitioned to IANA, it's toned down considerably (though retains mention of the early five minute difference in clock time inside of, and outside of, Paris train stations in the late nineteenth century).
https://data.iana.org/time-zones/theory.html
We had a database that threw an assertion when trying to deal with times in Kingston, Jamaica from 1910, and found that the timezone offset from UTC was not on a 15 minute boundary -- it wasn't even an integral number of minutes. Sailors are good at determining when local noon is.
data.iana.orgTheory and pragmatics of the tz code and data
