Note: The reflog is an internal part of the way Arq works and normally you'll never need to worry about it, so feel free to skip this section!
Arq stores backup records as a linked list. The most recent backup record has a link to the previous backup record, which has a link to the previous backup record, and so on until the earliest backup record. Arq stores the ID of the most recent backup record at the destination. Each time Arq creates a new backup record (which has a link to the last backup record), it replaces the most-recent-backup-record-ID at the destination. It also writes a separate file in the "reflog folder" containing that ID and the previous ID.
In rare situations the most-recent-backup-record-ID could be lost. In those situations, you can select a different most-recent-backup-record-ID from the reflog and use that backup record list instead of starting over with no backup history.
To view the reflog, select the folder on the left in Arq and then pick "View Reflog..." from the Backups menu.
In the reflog viewer you can expand each reflog entry to see the backup record list it refers to. You can also select a reflog entry and click "Switch" to tell Arq to use that entry's backup record list instead of its current list. In this way you can switch to an earlier backup record list if the most-recent-backup-record-ID pointer was lost.
Caveat: Older reflog entries will not be useful if Arq has since performed a budget-enforcement cycle. In that case Arq may have deleted some of the backup records in the list pointed to by that reflog entry. The reflog is most useful for recovering a backup record list relatively soon after a problem has occurred.