Years later, Jane noticed that the refrigerator was running more than normal, and we both noted that the frost level on the holding plate now never reached more than about 3" high.
I confess that despite my Chemical Engineering background (or perhaps because of it...), holding plates seemed a little mysterious to me. Ours is a stainless steel tank about 12" x 20" x 4" in which the evaporator coils are submerged in a dilute glycol solution (propylene, not ethylene since this is an application where leaks could conceivably get on food. Ethylene glycol is a deadly poison, while propylene glycol is a food additive). The idea is that when the compressor runs, ice forms in the tank (in the form of slush, due to the glycol), thus storing the 'cold' for later release. It is kind of a formalized and self-contained "Blue Ice" system. This is a design which works well on a boat where the engine is run maybe once a day and it is desired to not use power when it is off. Anyway, I was reluctant to get involved with the holding plate.
Now I really had no choice. Presuming that the lowered frost level indicated the level of fluid in the tank, it was obviously leaking. Where? An examination of the tank showed that the little fitting on the bottom was once again loose - probably bumped again somehow. So I broke it off completely - way too easily. Nothing came out of the resulting 1/4" hole.
OK, next step - find the leak - all I'd have to do was put more fluid in the tank and see where it comes out. So how to do that? Taking a deep breath, I drilled a 3/8" hole in the far upper sidewall to gain entry for a hose for filling. Voila! As soon as the drill penetrated, the water (no longer any glycol - just water) gushed out the hole in the bottom! The hole really did communicate with the tank interior, but the water did not drain earlier because there was no way for air to enter the tank to take its place. Mystery solved.
Now... how to seal a 1/4" hole in an almost inaccessible location, and a new 3/8" hole that was easily accessible... against the possibility of some internal pressure (should it freeze solid), and proof against future bumping?
The one in the 1/4" hole on the bottom is not visible, but the 3/8" one in the upper sidewall can be seen in the picture. The tank is now filled with a solution of 10% propylene glycol (aka "RV antifreeze") and the holding plate works beautifully.