I'm guessing.
If it is a marriage
register, then the later date will be the date the marriage was registered with the county as having actually taken place.
If it is a record of marriage licenses, then the later date would be the marriage.
The steps in the US are: 1) Get a license, 2) Get married, 3) The minister or official registers the marriage.
Note: Not everyone who got a license went on to be married.
This page
https://www.ancestry.com/search/collections/60548/offers 2 collections for Beaufort Co, NC. 1851-1866 and 1851-1982. If you can find an event that is in both registers and has 2 different dates, you might be able to sort it out. The earlier register was hand written and says "so and so were married by me on this date", with no date of registration.
That said, I couldn't find a record in both registers that had different dates. I think the later register just put down the same date in both columns for the early records.
Another way of possibly sorting it out is to see if you can find the same minister/official registering a few marriages, one right after the other. If the date in one column is the same for all of them then that is probably the date he went to the registry office and wrote them all in the book. -- Even if the couples were married on different days.
The problem here is that the sections of that database that I am looking at are
indexes to the records themselves, and they are indexed by surname of the groom. So it might be difficult/impossible to find the same minister's records. That method works better if you are looking at the registration book itself, where the marriages were recorded chronologically.
(I hope I am clear here, but I doubt it)
Bottom line, since it is a Registration book, the two dates are probably for the marriage and the later date of registration.