Thanks, I was working on a different premise and missed the mid-list blanks
problem.

***********
Best Regards,
Ron

XL2003, WinXP


"Harlan Grove" wrote:

"Ron Coderre" wrote...
....
I found more effiiciencies in the row reference of INDEX with this array
formula:

B3:
=IF(COUNT(MATCH(A$2:A$20,B$2:B2,0))<COUNT(1/(A$2:A$20<"")),
INDEX(A$2:A$20,MATCH(1,--ISNA(MATCH(A$2:A$20,B$2:B2,0)),0)),"")
....

Doesn't quite do the same thing. I had assumed your earlier formula

=IF(SUMPRODUCT(($A$2:$A$20<"")*1/COUNTIF($A$2:$A$20,$A$2:$A$20&""))
<=COUNTA($B$1:$B1&"")-1,"",INDEX($A$2:$A$20,MATCH(TRUE,
ISERROR(MATCH($A$2:$A$20,$B$1:$B1,0)),0),1))

included the &"" bits in order to exclude blank cells. Now that I've tested
it, I see it doesn't. Neither does your latest formula above. Maybe it's the
correct thing to do to include blank cells in the results, but both your
formulas would then skip the last nonblank entry in col A. My formula,

=IF(COUNT(MATCH(A$2:A$20,B$2:B2,0))<COUNT(1/(A$2:A$20<"")),
INDEX(A$2:A$20,MATCH(0,(A$2:A$20<"")-ISNA(MATCH(A$2:A$20,B$2:B2,0)),0)),
"")

returns all nonblank entries in col A. If it should include blank entries,
then change it to

=IF(COUNT(MATCH(A$2:A$20&"",B$2:B2,0))<MIN(ROWS(A$ 2:A$20),
COUNTA(A$2:A$20)+1),INDEX(A$2:A$20&"",
MATCH(0,1-ISNA(MATCH(A$2:A$20&"",B$2:B2,0)),0)),"")