Steve,

PMFJI. The following code shows that result of the query may be ambigious because of a memo field.

CREATE CURSOR test ( kf1 i, mf2 M)
INSERT INTO test VALUES(1, "ABC")
INSERT INTO test VALUES(1, "DEF")
INSERT INTO test VALUES(2, "GHI")
SET ENGINEBEHAVIOR 70
SELECT kf1, mf2 ;
	FROM test ;
	GROUP BY kf1

It's the same problem as with the fields that aren't in the GROUP BY clause.

>Larry,
>
>I see how your solution will work. Thanks.
>
>I agree, VFP should adhere to the standard, as junk data is just that.
>
>However, it seems to me that memo type data fields can't affect the result set so are irrelevant to the query. If that it true, then it logically follows that the data type should be ignored in the Group By clause. I don't know if this is true in SQL Server or Oracle, but if so, then the standard imposes an unnecessary restriction. I realize, that pure theory suggest that the standard should restrict any possible contamination to a result set, but pure theory is not always the best model to adopt in real world scenarios.
>
>Just my $0.02 <s>
>