My weird sink drain

This is totally OT for this group -- except the group contains a lot of
obviously clever people who are professionally interested in aerodynamic
and fluid flows and pressures, pipes, valves, and the like, and maybe
someone will be entertained by the following odd bathroom sink behavior.

Our newly remodeled bathroom has a circular glass above-counter sink
like a hemispherical glass salad bowl, about 18" in diameter at the top
and 7" deep, with a push down-pop up drain plug in the bottom center,
and a pretty high capacity faucet above it. Standard elbow fitting and
drain pipe going into the wall underneath the sink and counter top.

With the drain open, turn the faucet on full force: water gets dumped
into the sink considerable faster than it can drain out and the water
level in the sink rises rapidly up to the rim, on the verge of
overflowing, in 20 or 30 seconds.

At the last second turn the faucet part way off -- down to roughly 50%
of full flow, more or less -- then trim the flow until inflow rate just
equals outflow, so the water level stays just 1/4 inch or so below the
rim. Then leave it in this steady-state condition, and wait.

For approximately *eight minutes* (by the watch) the resulting situation
remains perfectly stable, with water level hovering just below the
overflow point. Then, all of sudden, water level starts dropping.

Turn faucet back up to full flow. Water level continues dropping, keeps
dropping faster in fact, until sink is essentially empty, and the full
force input that initially caused the sink to fill now roars down the
drain with only 1/2' or so of water swirling around the drain in the
bottom. This continues as long as I want to watch.

This doesn't seem to result from just blowing some temporary clog out of
the drain: I've repeated it three times, several hours apart, with
essentially identical behavior.

I'm at a loss to explain how it happens, except to hypothesize that
maybe there's some point underground and quite a ways further down the
drain where the drain pipe has a long slow rise, then a drop, and in
some way the initial slower flow has to fill the rising section until a
siphon action gets going over the top?

That doesn't really sound persuasive, however. Anyone have any other
ideas?