On Sep 30, 5:15 pm, I wrote:
I'm not sure, but I think someone pointed out a flaw with GEOMEAN
not too long ago. If my recollection is correct, it is more reliable to
compute the geometric mean with the following array formula (ctrl-
shift-Enter):
=exp(average(ln(1+$A$1:A3))) - 1

Not so much a flaw in GEOMEAN as a limitation of binary computer
arithmetic. In a thread on June 23, Ron Rosenfeld speculated about
why GEOMEAN returned #NUM after a certain number of data points:
"With 174 entries, my guess is that your formula is producing a value
outside of the range allowed by Excel". It is unclear what the range
of the terms of GEOMEAN were in that context.

Also, in a thread on Aug 21, Harlan Grove offered a formulation of the
above that does not require an array formula, namely (modified to fit
Dana's example):

=exp(sumproduct(ln(1+$A$1:A3))/count($A$1:A3)) - 1