I probably would've gone with what I did in my test 2.

... the super complex array <g ?
filled down only in 1 col x 3000 rows
(and which also contains the same volatile INDIRECT ?)

2 observations:

1. The returns are different, ref explanation in my response to you earlier
(pasted below), because of different interps of what the OP wanted. My
suggestion simply lines up all the returns from the 6 sheets for each of the
3000 lines in 6 cols. Yours return results in 1 col, with an implicit
precedence order assumed in the 6 data sheets. Nothing wrong there, just
different interps.

Btw, think there was a slight interp difference in that I presumed what
the
OP would like to have was to match & extract separate returns from each of
the 6 sheets for all the 3000 lines. Then he could assess further what he
wants done.
I didn't assume any "precedence" order in the 6 data sheets.

As you can see, the calc times were pretty much the same ..

2. Wondering whether the calc times would be any different if say, your
array formula were to be modified to return similarly as mine the results in
6 cols ?

Just some thoughts, Biff <g
--
Rgds
Max
xl 97
---
Singapore, GMT+8
xdemechanik
http://savefile.com/projects/236895
--