>Craig,
>
>As you can see in the sample code, he *did* assign (sql) aliases. It does not seem unreasonable to assume that SQL would use the self-same alias as has been supplied, when one *has* been supplied.
>
>I know of the Help warning, but I have to doubt that *this* is the reason.

I've burned my fingers on this already. It uses its own aliases, and if you supply an alias, it's the existing one, which is unused by the SQL, so, yes, Recno("cust") will give you the record number of the current record in Cust, and not in the temporary alias used during the Select. This record number doesn't change at all during the select, therefore one number is returned all the time.

Sounds logical, but sure isn't too clear from the help. I guess the real mechanism is a bit too complicated, and the gurus didn't want to scare off the... OK, so we had already agreed a good white paper on the internals of Rushmore and Fox's SQL would be welcome anytime, no need to repeat.