sanders,
OK, for the InquiryInfo, the PrimaryKey should be CustomerID AND InquiryNo.
Then for InquiryDate the Key should be CustomerID AND InquiryNo AND:
Either InquiryDate if you'll only have one inquiry in this table for each date,
OR
InquirySequence--a new field you create that is just a number 1,2,3, etc, based on which record for that particular Cust/Inq#
..either way there will be 3 fields making up the PK of the INquiryDAte table.
Now, for the Compound index, in the index editor (the lighning bolt icon in table design). You are right, it is not a very intuitive interface. The Index Name is in the left column, and to make a compound index, enter a Name, then tab to the Field, and add a field, then just add another field directly below but *leave the Name blank*. Access will just assume that any fields entered directly below a field, are part of that same index.
Now, let's assume the InquiryDate (many side) has 3 fields as key:
CustomerID
InquiryNo
InquirySeq (you could use autonumber for this but I highly discourage Autonumber for *anything*)
Now, getting into some nerdy nitty-gritty with Query Execution plans, (and this is partly why I discourage the compund keys where possible)...you can ignore this if you like, it's just some extra low-level info which usually wont' amount to a hill of beans in this crazy world...
...when you join the InquiryInfo to InquiryDate using CustomerID and InquiryNo, if you have criteria on the Many side Access will execute this is to *first* read the Many table, then for each record there Access will fetch the One side using the One sides PK...
BUT...
...if you have criteria on the One side, then Access will change it's plan and now scan the One table first, but now there is no matching index--so access will use the Many table's PK, which index that Access will use will be the PK, but this is not as efficient as it would be if the first 2 components were indexed as well (in a compound index)--which Access will use if it exists, you can use SHOWPLAN to verify this.
And if you have no criteria, Access will use a Merge Join on the matching indexes, which is going to be faster than a regular join.
This is almost splitting hairs, however, since the performance difference usually will be minimal...If the compound index (aside from the PK) on the Many side doesn't exist, then Access scans the PK of the Many side (which is slightly larger), but it must do extra logic and look at the first 2 fields of the pk to match with the 2 key fields of the One table, whereas if it used a separate compound index on the first 2 fields of the Many side, it would just match the full index value with the PK on the One side.
--Jim