The Hardwood That Cupped the First Summer
Why this matters
Cupping is the hardwood callback that tempts a crew into the worst possible repair: sanding the floor flat while it is still cupped. The board edges get cut down, the wood dries out over the winter, and the floor comes back crowned, with the centers now high and the finish worn through on every board. This case follows a whole-floor cup to its source and shows why the fix started under the house and the sander came last, if at all.
The complaint
In late July a homeowner called about solid red oak, 3/4 inch thick and 4 inches wide, nailed down over a 3/4 inch plywood subfloor in March across about 900 square feet of living room, dining room and hall. Every board had "gone wavy." Under a low light across the floor, each board showed raised edges and a dished center, and the seams had tightened so there were no visible gaps anywhere.
Cupping is the board's edges standing higher than its center. It happens when the bottom of the board is wetter than the top: the wetter side swells, and a board pinned along its edges by its neighbors bows. The cupped shape is a direction arrow, and it points down.
The house was a ranch on a vented crawlspace. The job file showed the flooring delivered and installed the same day, with no moisture readings recorded for the wood or the subfloor.
The section that tells the story
SECTION ACROSS TWO BOARDS, NOT TO SCALE
edge high edge high edge high edge high
|\ /| |\ /|
| \_____________/ | | \_____________/ |
| | | |
|_________________| |_________________|
face 9 pct (shallow pins) back 13 pct (deep)
=================================================
3/4 in plywood subfloor, 16 pct from below
----------+-----------------------+--------------
| joist | joist
| |
crawlspace air 85 pct RH, 78 F, vents open
^ ^ ^ ^
..... bare soil, no ground cover, damp .........
The readings were taken with a pin meter set to red oak, shallow pins for the face, insulated pins driven to near the back of the board, and the subfloor read from the crawlspace side. (The meter's species setting matters: red oak read on a generic or wrong-species setting can be off by a point or more, enough to blur a gradient this size.) The interior was 74 F and 55 percent RH with the air conditioning running.
Working through the candidates
The tech started with what could produce a whole-floor cup, then struck causes as the evidence ruled them out.
A water event from above. A dishwasher leak, a pet bowl, a mop left wet. These cup a floor locally, and the worst damage sits near the source with stains or dark seams. Here the cup was uniform from the front door to the back hall, with no stain and no wet spot. And water from above wets the face first, which would read wetter on top. The face read 9 percent and the back 13. Struck.
A plumbing leak below. Possible, because it wets from below. The tech crawled the space with a light: supply and drain lines dry, no stained joists or puddles under fixtures, and a leak would leave a wet zone with a center, not a floor wet evenly from end to end. Struck.
Interior humidity alone. At 55 percent RH inside, oak settles at roughly 10 percent moisture content. That is a humid summer house, not an extreme one, and humidity from the room acts on the face. A floor responding to the room would be swelling from the top, and the face here was the driest layer measured. Not the driver.
The wood installed too dry for the site. The install record had no readings, so this could not be checked against the day of install, and it is a real contributor on same-day delivery. A floor laid well under its in-service moisture content takes on moisture after install and swells as a whole, which tightens seams and can buckle a floor off the subfloor. That explains the closed seams. On its own it does not explain a gradient top to bottom. Kept as a contributor, not the cause.
Moisture coming up from the crawlspace. The crawlspace air read 85 percent RH over bare soil that was visibly damp, with the foundation vents open to humid summer air. Plywood exposed to that air from below for months reached 16 percent. The board backs sat on that plywood at 13 percent while the faces, conditioned by the house, held 9. A 4-point gradient through a 3/4 inch board is what produces a uniform cup over every joist bay. This one fit every reading and the season. Confirmed.
The summer timing was part of the evidence. Installed in March, the floor saw cool, drier crawlspace air for its first months. The cup arrived when warm, humid outdoor air flowed through the vents into a space whose soil was already wet, and summer air entering a cooler crawlspace rises toward saturation.
What the record would have shown
Two readings on install day would have caught this. The hardwood makers and the National Wood Flooring Association (NWFA) installation guidelines call for the flooring and subfloor to be within a set moisture difference before nailing, commonly 4 points for strip under 3 inches wide and 2 points for boards 3 inches and wider (see related: Wood Moisture Content and the Match to the Subfloor, which owns the rule). This was a 4-inch board, so the limit was 2 points. Those same guidelines, in substance, expect the crawlspace earth covered with a vapor retarder before wood flooring goes in above it; check the current edition and the flooring maker's instructions. A tech crawling the space at the estimate would have seen bare damp soil and written it up as a condition to correct before install.
Same-day delivery also meant nobody checked whether the wood had reached in-service conditions. Acclimation is the wood reaching the moisture content it will live at with the HVAC running, checked with a meter, with a day count only where the maker states one.
The fix, in the order it has to happen
1. Correct the crawlspace first. Until the source stops, nothing done to the floor holds. The shop wrote the condition up for the homeowner and referred the crawlspace work to a crawlspace or waterproofing contractor, since it is not flooring work. The options were a ground cover vapor retarder over the soil with the vents left open, or a closed and conditioned crawlspace. The International Residential Code covers both in section R408 as adopted by your jurisdiction: ventilation requirements for a vented space, with reduced vent area permitted where the soil carries a Class I vapor retarder, and the requirements for a closed, conditioned one. Which suits this house, its drainage and its climate was for that contractor and the homeowner to decide. The homeowner chose to close and condition the crawlspace. Working in a crawlspace is a confined, dirty space: the crew member entering it carries a light and has a partner topside, watches for live wiring and animals, and wears a respirator if the space has mold growth or disturbed insulation.
2. Let the floor come back on its own. With the crawlspace dry, the subfloor gives up its moisture, the board backs follow, and much of the cup relaxes as the gradient closes. The shop set monthly readings at the same three points: face, back, subfloor.
3. Evaluate only after the readings settle. The shop's gate was the face and back within 1 point of each other and the subfloor within the 2-point match of the flooring, held across two monthly readings, with the house conditioned normally. Here that arrived in November: face 8, back 9, subfloor 10 percent.
4. Judge what remains. The tech looked at the floor standing, in normal light, which is how finished work is judged, not on hands and knees with a flashlight raking across it. About two-thirds of the floor had relaxed flat to the eye. The dining room still showed a slight cup on boards near an exterior wall. The options were to leave it, or to sand flat and refinish now that the moisture was stable. The homeowner chose to sand the dining room and screen and recoat the rest for an even sheen.
Where this step forks. Two findings at evaluation would have changed the plan. Boards that stay cupped after the gradient has closed have taken compression set: their edges were crushed against their neighbors while swollen, and the fibers do not spring back. Mild set sands out; severe set, or set with cracked tongues, means replacing those boards (see related: How to Replace Damaged Boards in a Nail-Down Hardwood Floor). And boards lifted off the subfloor with nails pulled through, rather than cupped in place, mean the floor buckled from overall expansion. That points back at the wood having gone down too dry, and it calls for re-securing or relaying a section, not sanding. Neither showed up here, but both are why the evaluation is a walk with a meter and not a decision made from the office.
Why the order is not negotiable. Sanding in July would have taken the raised edges off boards that were still wet underneath. When the backs dried in winter, the board centers would have stood proud of the edges, a crowned floor with the finish already thinner at the centers. Sanding is the last step, after the moisture holds still, never the first.
How it was confirmed
The following July, the first humid season after the crawlspace was closed, the tech took the same three readings: face 9, back 10, subfloor 11 percent, with the crawlspace at 52 percent RH. No new cup. The seams showed slight seasonal gaps the following winter, which is normal for solid strip and was explained to the homeowner at the walkthrough.
References
- National Wood Flooring Association installation guidelines, current edition, on moisture difference and crawlspace preparation.
- International Residential Code section R408, Under-Floor Space, as adopted by your jurisdiction.
- Hardwood flooring manufacturer's installation instructions and the moisture meter maker's species corrections.
- See related: Wood Moisture Content and the Match to the Subfloor; How to Take and Log Wood Moisture Readings Before Hardwood; How to Acclimate Flooring and Confirm It Is Ready.
- See related: How to Sand and Refinish a Hardwood Floor.