Alain Fournier wrote:
The Panama Canal is a potential delimiter only if divide the Americas
into North and South without having Central America. The way I was
taught it (I'm not sure how to which point this is universally accepted)
North America is Canada, the US and most of Mexico, the part of Mexico
east of the isthmus of Tehuantepec (that is the thinnest part of Mexico)
is in Central America.

Yeah, but that's a artificial boundary that was put in as a arbitrary
division, like that between Europe and Asia.
Before the Suez canal, the same held true for Asia and Africa.
I don't consider a land mass you can walk to from another landmass to be
a true separate continent.
Which means that back in the ice age when Asia and North America were
joined by the land bridge, the Earth only had three true continents:
Australia, Antarctica, and this monster that was everything else.

Even if you do accept the above as the definition of the southern
limit of North America finding the exact centre of North America
is a daunting task. There are some of the shores of the arctic which
are poorly mapped and for which it is not obvious what would really
be land and what is merely dirty ice that doesn't quite melt in the
summer.

You could figure that out in detail via seismic explorations using
blasting and reflected shockwaves from under the ice (in fact, I'll bet
it's already been done).
The question as to whether ice that doesn't melt on its shoreline is
considered part of a continental landmass is a good one. I wouldn't
think it would be as long as it was floating and didn't go all the way
to the seabed.
If it did go all the way down to the seabed though, like a glacier that
terminated in the ocean, you could make a good argument that it was
indeed part of the continental landmass, as ice is technically a
mineral: http://www.galleries.com/Minerals/OXIDES/ice/ice.htm

Pat