>>As requested, with a few examples - in each case I believe the goodtbl{n} and
>>badtbl{n} should give the same results.In all cases, the goodtbl{n} results
>>are what I expect. In the case of badtbl2 & badtbl3, I think the results
>>should be identical.
>
>Len, FWIW, if you SET ANSI ON, you get the same results for each select -- but I don't think they're the results you want. :-)

Consistently wrong is possibly better than occasionally right. At least you know you've got the wrong (well, let's be fair, right but incomplete) results, rather than having to wonder whether you've got the full set of data this time.