Hmm... there could be a lesson in here somewhere if I could sort it
out.

My 90 Accord had 196k and a burned #1 exhaust valve when it came to
me.  The owner is a friend and we just assumed he had let the valve
lash get out of hand.  He also had a puff of blue smoke at startup
which could have been valve stem seals or the spark plug tube seals.
I tore it down.  The piston and valves were clean, even the one with
the hole, the usual varnish but no coke or hard carbon.  I know what
they say about high mileage motors but there wasn't any evidence of
ring wear, so I took a chance.  I gave a new valve stem seal set to
the guy who did the valve job and it was supposedly installed.

After putting it back together it burned oil.  1qt in 300 or 500
miles.  The O2 sensor circuit also failed at some later point, and I
never got it to work again (long story).  I kept adding oil and
driving it.  The motor always ran smooth with good power and no
missing.

Yesterday after between 9,000 and 10,000 miles since the repair, I was
on the interstate and lost power.  I pulled the #1 wire and it was
sparking.  I unplugged the fuel injector and drove the 300 miles home
(got to love Honda -- 25mpg on 3 cylinders).

After a compression test, #1 has zero compression again.  #2-4 have
180.  #1 has coke spots around the insulator in the pattern of a moldy
piece of bread.  #2-4 are clean.

I guess all the oil consumption was in #1.  I also guess the oil
burning coked up the valve and burned it again.

The most reasonable thing to do at this point is change the motor,
obviously, because I can't afford the time to fix something that won't
stay fixed.

But it bugs me that I don't understand the nature of why this build
failed.  I had the impression that I thought it through thoroughly.

Weak valves supposedly keep a worn out bottom end on life support.  As
it usually goes, having a valve job done ends up pulling even more oil
past the rings, burning oil and causing the rings to stick / valves to
burn / fouling plugs.

I thought the absence of coke in that cylinder when I tore it down was
a sign that it was not burning oil already.  A compression test on the
other cylinders showed 180 across the board as they still show now.
The absense of oil in the bad cylinder combined with the good
condition of the other cylinders made me think the rings were in fine
condition.  The plugs were worn but looked otherwise okay.  Along with
being an easier one person job, that is why I went with the valve job
and reassemble, rather than replace the motor.

Also strange to me is that on the plug I pulled, there are coke spots
around the insulator but no typical oil "slick" from bad rings on the
plug electrode itself.  The plug consequently never got fouled so the
motor kept running smoothly until its sudden death.

But I know it was burning oil because there was blue smoke under load
at high rpm (say an uphill downshift on the highway), and there was
never a drop on the ground.

I'm assuming the valve stem seal was done correctly unless there's a
reason that assumption should be checked.  In any case there was no
blue smoke at startup anymore as it was prior to the valve job.

So here's where I'm stuck.

Did I screw up by assuming a clean #1 + good compression in #2-4 =
good rings in #1?  Is there anything else I could have looked at to
ascertain the condition of the dead cylinder?

If the bottom end was too wore out for a valve job, why did only #1
start burning a lot of oil and the other plugs remained dry?  (And it
is a *lot* of oil for one cylinder to be drinking a quart in 300 miles
all by itself.)

My only prior experience with a valve job on another 200k 4 banger had
been that it went perfectly and the motor used even less oil
afterwards.  Is the Accord more in line with others' observations of
what typically happens after a valve job?

I just want to be able to reconcile my "observations" when I first
tore down the motor with the "experiment" and the "results"... then
I'll be able to sleep :-)