I don't know anything about the history of chemistry (because chemistry is balls), but if observations can't be consistent without a particular thing being true, those observations would count as empirical evidence for that thing.