Hi Sue.
First thing -- your second command has -x (with loadID) AND an SQL Statement. In the MP world, you only need one of these.
Second, I believe I've seen this behavior before in a server consolidation back in 2006... When you load the extracted file, as long as the generic index entries are next to each other (ie, the generic index is sorted by object id, then byte offset), arsload should be smart enough to NOT store the duplicates. Give it a try and see if this works in your environment.
-JD.