After much mistaken investigation, the leak turned out to be from
the low/reverse servo pivot pin hole. The pin has an o-ring, but
apparently it decided to quit working after 47 years.

Fixing the o-ring meant considerable disassembly: Transfer case,
tail housing and pan. Incidentally, the tail housing is also the
rear trans support, so it's a big job.

After some soul-searching, I decided to try epoxying the hole
in the case where the pin resides. The trans has nearly 200k
miles, before too long it'll have to come out for a major problem.
In the meantime I can look for a good shop to do the job; it's
much bigger than I want to tackle myself.

First problem was seeing the spot. A small mirror on a magnetic
stand worked fairly well, but working, reversed, in a mirror is
quite hard. Second problem was fitting an applicator tool into the
space.

I started with a soft iron wire, shaping the end into a tiny 
spatula and bending the wire until I could touch the point of
application _without_ touching other nearby surfaces. This helped
establish both muscle memory and the size/shape of the applicator.

Then I made a real spatula out of .093" music wire, forging the
blade and bends by using a torch, hammer and anvil.

With a low ATF level, the fluid stayed below the leak if I started
the engine and ran the trans through the gears once a day. That 
meant I could keep the area clean while applying and curing the epoxy.

Cleaning was done with aerosol carb cleaner, using a short length of
very small plastic tube (string trimmer fuel line) as an extension.
This let me scrub the area with Scotchbrite and hose it down, repeatedly.
The retainer tab over the pivot pin obstructs the scrubbing severely,
so I tried to compensate with lots of carb cleaner.

The epoxy was applied to a warm trans case, in hopes it'd flow into
the surface roughness and stick a little better. I just pressed the
epoxy on the end of the pivot pin and pushed it up, under the retainer
tab as far as possible, letting it flow out and down by gravity. 
This was repeated three times over a couple of days. Total thickness
accumulated was not much, perhaps a millimeter.

A test patch on the trans case was epoxied as a reference. It was
more accessible and much easier to clean, but even then the bond was
not very impressive; the epoxy could be peeled with a sharp knife. 
However, it did stick somewhat, and might resist a zero-pressure leak.

There's no doubt the leak will return but it's the best stalling
tactic I could come up with.