I am looking for a dataset that contains the languages spoken in each country and their respective percentages of the population, going back as far back into the historical record as possible.
I know the CIA World Factbook has entries of the form Afghanistan: Languages: Dari 80%, Pashto 40%, etc.
Their archive has percentages going back as far as about 2008 and their archive in general only goes back to about 2000. Anything that has data like this for before 2000 would be great. Even if not all countries are included, e.g. just the big ones like China, US, Germany ,etc.