I would have come to the same conclusion. After all, a square hole with width x has a greater cross sectional area than a round hole with diameter x.

However, if you had a round and a square port with the same cross sectional area the round port will flow better. There is less dimensionally induced turbulence in the circular hole.

Interestingly when you read the turbo books, they suggest that you induce turbulence right after the turbine outlet. This is because turbulent air travels a shorter distance to get out of the exhaust, than what swirling air would, as the swirling air would travel along the outside edges of the exhaust...

Theoretically it makes sense, but in practice you simply turn the boost up slightly, and forget that you exhaust gas swirls...

Benny, Thats interesting about the D-shaped port. Did it say that shape is best at the edge of the head or throughout the port length? I'll look for that book. Right now I'm in the middle of "Race Car Vehicle Dynamics". Its an awsome book too.

Thanks for replies everyone. Sounds like you went to some effort to get an answer, Tom. Thanks.

So, has anyone actually ported the exhaust ports different shapes and tested it??

Back in the 70's when the Chrysler Direct Connection [now called Mopar Performance] people developed the W series of performance heads for the LA series of small block engines they found that, for a given cross sectional area, the D shape, with a flat floor flowed more exhaust gas than any other shape, & this shape was adopted for the exhaust ports in these heads.
I believe that some Chevrolet heads, particularly big block heads, copied this basic shape for some of their performance models.
[can anyone confirm]

The Datsun A series engine has a flat floor in the exhaust port too. Perhaps those clever little Datsun Engineers were much more clever than we have yet given them credit for.

EDIT
OK, just read Benny's post. It looks like more than one group has come up with the same result.

Hmmm...!!!