I finish all of it. The underside of a shelf, the back of a case, the interior of a drawer, the bottoms of feet. Not just the faces that will be seen.
It began as discipline and became structural logic. Wood is hygroscopic — it exchanges moisture with the air through any unfinished pore, and a finish does not stop that exchange. It only slows it. The Forest Products Laboratory's guidance is that finishes protect wood from moisture only when applied evenly to all surfaces, and that unequal coatings can cause unequal shrinkage and lead to warp. Not everyone weighs that the same way, and the dissent is worth stating properly rather than in passing. Bob Flexner has argued for years that this is a myth — that a finish makes no difference to whether a top warps, because it only slows vapor exchange and cannot stop it. He does not stop at the mechanism. He points out that pre-1920s furniture was rarely finished underneath, and that when those tops cupped they cupped on the top side, which is not what an unbalanced finish predicts. He puts the real blame on repeated wetting of the top face, on wood that was never dried properly, and on pieces moved between climates. He is not wrong about any of that. He also grants that finishing both sides for looks and feel is perfectly legitimate — which covers two of the three reasons I am about to give.
Where I differ is narrower than it sounds. An unbalanced finish does not hold one face dry. It makes one face slower. When humidity climbs, a raw underside takes up moisture ahead of a sealed top, swells, and the top goes hollow. When the air dries out again the underside gives it up first and the cup runs the other way. All else equal, the unfinished face leads — and all else is not equal when the water lands on the top, which is Flexner's case and a fair one. Finishing both faces alike does not stop the wood moving. Nothing does. It keeps the two faces moving on the same schedule.
I should say plainly that an oil creates far less of that imbalance than a film finish would. The Laboratory's warning is aimed at coatings that exclude moisture well, and tung oil is not one of those. Balancing the faces here is a floor, not a defense against the failure they have in mind.
The second reason is longevity, and here the ground is firmer. Pieces live in the real world. Water finds the underside. A plant overwatered on a shelf, a spill that runs over an edge and clings underneath, a mop that hits a base. Raw end grain is a wick. Moisture moves along the grain far faster than across it — the drying trades seal the ends of green stock for exactly that reason — and the feet of a piece standing on a wet floor are the worst place to leave bare wood.
Tung oil penetrates and polymerizes inside the fiber, and that network slows water at first contact. It is worth being exact about what that buys. It is a barrier for the length of a spill, not the length of a season — leave the piece in humid air for two weeks and the wood reaches equilibrium regardless. But most water damage is an incident, not a season, and an incident is what the oil is good against.
A closed, unventilated interior is also where mildew is most likely, which is reason enough to care what state it is in. I would not claim the oil prevents it. What a sealed interior gives you is a surface that wipes clean instead of drinking whatever reaches it. It does not take much, either — one thinned coat on a hidden face is usually enough. Not to seal it, which one coat will not do, but so the surface is saturated and wipeable rather than raw.
The third reason is integrity. A piece of furniture is not a facade. It is an object that will be moved, turned over, repaired, inherited. Someone will see the bottom. A future restorer will judge the care by what was done where no client was looking. Finishing the unseen is a way of saying the object is whole, not staged.
A working detail follows from it. Finish the hidden faces before assembly and squeeze-out sits on the surface instead of soaking into bare pores, where it blocks the finish and leaves a pale ghost around the joint. The rule that comes with it is strict: mask anything that still has to take glue. Oil in the fiber is the enemy of a glue joint, and there is no film to scrape off — you have to cut the surface away, and in end grain and open pores even that is unreliable.
There is a practical cost — more oil, more time, more cure days for surfaces no one will photograph. And one that is easy to learn the hard way: an oil finish goes on off-gassing for weeks after it is dry to the touch, and in a closed space with no air exchange it has nowhere to go. Close a drawer on a fresh interior and it will hold that smell for months. The standard advice is not to put an oil finish inside a drawer at all, and the reason is sound. I do it anyway, and the discipline is the price: interiors get finished early and left open, and the piece is not shut up until the oil has stopped breathing.
In Sacred Noir work, where pieces are meant to be held and kept close, I treat the hidden surfaces with the same saturation as the visible ones. The piece should be finished like a body, not a mask — sealed everywhere it breathes, so it can move as one thing.