From a purist point of view, duplicate data is a no-no, as there is a lot of room for inconsistencies and other nasties to creep in. To save deleting from the hash, why not assign $VAR1 as a ref to $VAR2{KEY1}?

rdfield