"George" george@least wrote in message
...
Start with a concept of load.

A biscuit installed parallel to the load has the shear strength of the
glue
that holds it. A dowel - loose tenon - has far greater shear strength.

Do you have some data for this? I'm sure you could calc the shear strength
of a hardwood dowel.
But where is your data showing that the glue has less shear strength? And is
it really the shear of the glue or the shear of the biscuit itself which is
at issue? I think I saw some photos of destructive testing of biscuit joints
out there. I seem to recall that they did not fail at the biscuit. Loose
tenons are a different story.

A biscuit installed perpendicular to the load has much greater strength,
but
is still only 1/8" thick, vs. 3/8 or 1/2.

And is much wider. The cross section of the typical biscuit is much larger
than the cross section of the dowel.
A #20 biscuit has a cross sectional area of 0.35 sq in. You would need a
dowel 2/3 inches in diameter to match that cross section.
Not easy to use that size of dowel in 4/4 stock. :-) This is why the loose
tenon was invented.

Then think withdrawal. Here the initial nod also goes to the dowel for as
long as the glue remains sound.

I suggest you calculate the surface area before you make this conclusion.
I'm not going to do the math for you though. It looks like it is close.

Now return to a real M/T and notice that it's also a great shear joint,
but
cross-grained, and subject to the same, if lesser problems than its round
cousin. We pin the tenon, and there's a new shear kid in town. We have
to
shear the pin to withdraw the joint. This is made extremely difficult by
the registry of the shoulder of the tenon with the face of the mortise.