By dating the ages of the copies. They far post-date when the originals would have been penned.
That doesn't really address the problem. All it says is when that particular original or copy was penned. It doesnt tell us if it is an original or copy.
If i made two copies of something I wrote, one immediate after the original and one 10 years later, you would be able to tell that one was 10 years early, but you couldn't tell if it was the original or a copy.
Hence my point. There is no way to know a manuscript that old is a copy or the original. It's not like they stamped "Original" on the manucripts.