On Wed, 02 Dec 2015 00:13:36 -0700, Don Y
wrote:

On 12/1/2015 8:00 PM, wrote:
On Tue, 01 Dec 2015 21:46:03 -0500, wrote:

On Tue, 01 Dec 2015 19:26:18 -0700, Don Y
wrote:

On 12/1/2015 7:19 PM, Don Y wrote:
The outlets in the back of the house are on a 20A GFCI.

For those who didn't connect the dots, this is a GFCI *breaker*,
not a GFCI *outlet* daisy-chained to feed the others.
Does it trip with nothing connected? If not, it's not a dauber or
anything on the main circuit. It has to be on the "load". Outdoor
lights on a GFCI can be a problem - lots of possibilities for
"leakage"

If there is absolutely nothing connected, a neutral fault will not
trip the GFCI. The slightest load will trip it tho. (it splits the
neutral current)
The other thing to remember is the faults are additive. If you are
leaking a couple ma in a couple places, it will work until the total
gets up around 5ma or so.

You're ignoring the fact that the breaker latches and HOLDS on the
"second attempt". This suggests the problem is related to the
change in load characteristics from the previous "brief lighting
event" (fraction of a second when power surged into the lamps
just before the breaker trips on the first attempt).

I'm at a loss to explain how anything else could appear and
then disappear so readily!

If it is a clump of moist material, you may be cooking the moisture
out a little with the first attempt. I am not a fan of the "inrush"
theory of GFCI tripping. (or most other myths about what trips them)
This is a simple comparison of the current between the hot and the
neutral through a buck wound transformer. If you really think it is an
overcurrent situation, plug it into a non GFCI circuit to test it