Well I took the junction box cover off tonight and it was filled with water again. I made a better suction and ended up shop vac'ing about 2 gallons of water out of the entire length of PVC. I would have never thought that water would ever get into the PVC and I have no idea how it is getting through.
Water will get into any buried conduit, especially if the ends surface in different areas (indoors/outdoors, 2 different buildings, 2 different floors of the same building, etc.)
Air flow through the conduit will carry atmospheric moisture into it. When the temperature underground is less than above, that can cause the moisture to condense. The same thing can happen with buried cable that has voids inside the jacket. This is why direct-bury CAT5 and phone cable is flooded with a gel, and why DB romex has a solid jacket, while regular romex has air space in it.
Sealing the conduit at each end will minimize water accumulation, the tighter the better.
I fought water accumulation in conduits and cables for a long time years ago before I understood the principles involved. I first learned of this with submersible pressure transmitters in water tanks. The rise and fall of the water and resulting pressure on the cable jacket would pump air in and out of the exposed cable end. Over a period of months or a couple of years, the jacket would get enough water condensed in it to wick down to the electronics in the transmitter and destroy it.
Manufacturers now vent the transmitter cable jacket through a capillary tube connected to a desiccant tube or bellows to keep water vapor out.
Even for non-buried cables, junction boxes and hollow fiberglass antennas, diurnal pumping (the slight change in barometric pressure over 24 hours) plus weather related changes in BP, can pump humid air in and then a drop in temperature allow it to condense. Again, it can take a long time, but I've had sealed J boxes get wet inside. Now I drill a small vent hole to keep that from happening, and most antennas come from the mfr vented.
