Recovery records are used WITHIN a single archive file, to provide data recovery through data parity methods. You can define the amount of parity data the archive stores. Think of this as a CD backup where the CD is scratched or HDD is unreadable at some sector. The recovery records will reconstruct the missing/corrupt data.
Recovery volumes are used ACROSS multi-file archives, to provide archive recovery if one or more of the split volumes gets lost or corrupted. You can define how many recovery volumes are created. Depending on the number of recovery volumes you create, you can "lose" that many original archive volumes... or thereabouts. Think of this as multi-CD backup and one of the CDs gets lost or totally screwed. You can rebuild the missing CD/archive from the recovery volume.
You can also think of recovery volumes as "RAID over CD" if you catch my drift.