Seasonal movement in timber floors: gaps, cupping and plank width

At a glance

Change in width of one board free to move, for a 3 point change in moisture content. Flat-sawn and quarter-sawn use the Wood Handbook (2021) Table 13-5 coefficients for white oak (0.00376 and 0.00194 per point). A laid floor moves less than a free board. Estimates, not predictions.

The estimator: width, coefficient, swing

A timber board takes up water from damp air and gives it up to dry air, and it changes size as it does. Nearly all of the change is across the width and the thickness. Along the length it is small: the Wood Handbook gives 0.1 to 0.2 % from green to oven-dry for most species, against 5.6 and 10.5 % across the width of white oak (WH4). So the question for a floor is how wide the board is, how it was cut, and how far the humidity of the room swings.

The Wood Handbook gives a way to put a number on it for the small changes a floor sees in a building, between about 6 and 14 % moisture content (WH13, equation 13-3). It is three steps, and the rest of this sheet takes them in turn.

1. Choose the swing. The change in moisture content, in percentage points, between the driest and dampest conditions the room will hold (section 2).
2. Choose the coefficient C. The change in width, as a fraction of the width, for one point of moisture content, for the species and the cut. Use the tangential value for flat-sawn boards and the radial value for quarter-sawn (section 3).
3. Multiply. Movement = width × C × swing. For a 140 mm flat-sawn white oak plank and a 3 point swing: 140 × 0.00376 × 3 = 1.6 mm.

The handbook says the equation underestimates the true change by about 5 %, and that because commercial boards are seldom perfectly flat-sawn or quarter-sawn it will tend to overestimate the change in width (WH13). Treat the answer as an estimate to the nearest half millimetre, not a prediction. The figure is for one board left free to move; what a laid floor does is in section 4.

Step 1: from room humidity to a moisture swing

Wood does not respond to the humidity of the air directly. It settles at a moisture content that depends on the humidity and, much less, the temperature of the air around it. The Forest Products Laboratory table (WH4, Table 4-2) gives the figures below for air at 21.1 °C. Across 15.6 to 26.7 °C the same humidity moves the answer by a few tenths of a point, so the humidity is what matters.

Take the swing from the two ends of the range you expect the room to hold. The sources we read name different ranges, so the swing depends on whose range you adopt:

Moisture content from Table 4-2 at 21.1 °C; the swing is the difference. The 5.8 point example is ours. All the ranges are advice from the named source, not a standard. A UK maker's own worked example runs from 15 °C and 60 % RH to 21 °C and 35 % RH, which it gives as 11.14 % to 6.95 %, a loss of over 4 points, and about 1 % contraction at 0.22 % per point (MAKER); our table gives the same two ends as 11.1 and 6.9.

No source we read gives the real indoor humidity swing of a building. The honest way to find it is a logger in the room, as the sources ask for (DIST, MAKER). Until you have one, pick the range the heating and ventilation design is meant to hold, and say which you chose.

Step 2: the coefficient for the species and the cut

The coefficient is the one number that changes with the timber. Table 13-5 of the Wood Handbook (WH13) lists a radial and a tangential value for each species, per 1 % change in moisture content, valid between 6 and 14 %. The handbook says to use the tangential value for the width of flat-sawn boards, the radial value for quarter-sawn, and tangential for mixed or unknown ring orientation (WH4). The table shows some of the species laid as floors.

Wood Handbook (2021), Table 13-5; the mm columns are the coefficient times 100 mm. The handbook's species are North American. European oak, the usual timber for UK floors, is not in the table.

• Quarter-sawn is about half for oak, less so for walnut. For the two oaks listed the radial value is 45 and 52 % of the tangential; for black walnut it is 69 %. The handbook says quarter-sawn boards shrink and swell less in width and cup less (WH3, Table 3-1). The US trade body says the same of rift and quarter-sawn flooring and of narrow boards over underfloor heating (NIG, appendix H).
• Rift-sawn is not in the table. The handbook has no rift category. It calls rings at 30 to 60° to the face bastard sawn (WH3). Between the two columns is a fair guess, and a guess is all it is: ask the partner for the figure for the cut it will supply. See timber cuts explained for what each cut means.
• One maker publishes a single figure. A UK maker's guide gives 0.22 % dimensional change per 1 % change in moisture content (MAKER). It does not say for which product or which cut. It sits between the radial values of the oaks and walnut, which is plausible for a board with a stable construction, but we cannot say that is the case.

From one board to a floor: why the gap rules are smaller

It is tempting to add up the board movements across a room and size the perimeter gap to match. The table shows why that does not work. Free movement across a 4 m room of flat-sawn oak is about 45 mm, and the perimeter gaps in the guidance we read are 10 to 19 mm.

The US trade body gives the same arithmetic, a board 2¼ in wide moving 1/16 in, which over 16 boards is an inch and over a 10 ft floor more than three inches, and then says: “in actual practice, however, change would be diminished in a complete floor, as the boards' proximity to each other tends to restrain movement” (NIG, Coefficients of Change). A floor that is bonded, nailed or held at its edges is restrained, and the movement shows as small gaps at the joints, as stress, or as cupping, not as the full free figure. None of the sources we read says how much of the free movement is taken up, so we do not give a restraint factor.

• Ask for the gap and the joints as a set. The perimeter gap and any internal movement joint depend on the fixing method (bonded, nailed or floating), the width of the room and the humidity range. The workshop states them for the floor it will make, and they go on the layout drawing. See floor transitions and movement joints.
• Wide rooms. The US trade body says spacing between rows may be needed to minimise expansion on floors wider than 20 ft (about 6.1 m), with a washer or removable spacer every few rows or starting from the centre (NIG, chapter 9). Its magazine calls such a built-in gap an internal expansion joint, or washer row, and says it should not be filled (NMAG).
• Width. Movement is in proportion to width. The same magazine puts it as an 8 in plainsawn white oak plank shrinking twice as much as a 4 in plank of the same cut and species (NMAG).

Cupping, gaps that stay open, and what to ask

A board cups when its two faces are at different moisture contents: the drier face shrinks more, and the board bends towards it. The drawing shows the idea with the distortion exaggerated. A quarter-sawn board cups less than a flat-sawn one (WH3, Table 3-1), and an engineered board is more stable than a solid one but, the US trade body says, shows low humidity in a different way, by dry cupping (NMAG).

“Crowning” for a board that rises in the middle is trade usage; we did not find it defined in a source we could read.

What to tell the client at handover

Seasonal gaps are the symptom clients ring about, and they are easiest to deal with if the client was told at handover. The US trade body says the only way to remove seasonal change completely is to control the room's conditions, and that humidifying in winter also helps the rest of the building (NMAG). A UK maker asks for a logger that alerts when humidity falls below 35 %, so that a plant or another moisture source can be added (MAKER).

The humidity range to quote is a choice, and the sources differ (30 to 50 %, 40 to 65 %, 45 to 65 %). Quote the range the partner gives for its floor and the one the building can hold, and put both in the note below. The specification wording gives a note to adapt.

• What to hand over: the range, the logger and where it is, the expected gap size from the estimator, the instruction not to fill gaps, and who to call.
• What not to promise: that there will be no gaps. A solid floor in a heated building will show some.

The sources behind the keys

The keys in brackets above point to this table. The US body's figures are in US units, converted. Suppliers are named by type; names and links are in our notes for review.

The floor is designed for room conditions of [45 to 65] % relative humidity and [15 to 25] degrees C, held for the life of the floor, as stated by [workshop] for [product]. At the moisture content swing of [3.5] points between the driest and dampest design conditions, movement across the width of one [140] mm [flat-sawn / quarter-sawn] [species] board is calculated at [1.6] mm (width x coefficient [0.00376] x swing). The workshop is to state the coefficient used and where it comes from. Seasonal gaps at the joints up to [ ] mm in the dry season are expected movement and are not a defect [contract adviser to confirm this clause].

Perimeter gap [ ] mm, and internal movement joints at [ ] m centres where the floor is wider than [ ] m, as stated by [workshop] for the fixing method [fully bonded / nailed / floating], shown on layout drawing [ref] and covered by [skirting / threshold profile]. Gaps are not to be filled. A temperature and humidity logger is to be placed at floor level in [each room] from [7] days before delivery until handover; the record is to be issued to [designer] at handover.

Your floor is solid timber. It takes up moisture in damp weather and gives it up in dry weather, so in the heating season expect fine gaps between some boards, up to about [ ] mm. They close when the air is damper. This is movement, not a fault: please do not fill them. Keep the room humidity between [ ] % and [ ] %; a logger is fitted in [location], and if it stays below [ ] % add humidification. Contact [name] with questions.

Standards and sources

• USDA Forest Products Laboratory, Wood Handbook (FPL-GTR-282, 2021): chapter 13 (equation 13-3, Table 13-5), chapter 4 (Table 4-2) and chapter 3 (Table 3-1), read. US species.
• NWFA Installation Guidelines (revised 02.2016), Wood Floor Facts: Relative Humidity (2018) and Hardwood Floors magazine, 'Tis the Season (2017, updated 2026): read. US trade body; US units converted.
• A UK distributor's installation guidelines (Jan 2024) and a UK maker's wide plank guide (2023): read; suppliers, not independent bodies.
• BS 8201:2011: BSI lists it as Current (31 October 2011). Not read: confirm its advice on gaps.

Questions we are asked

How much does a wood floor move with the seasons?

It depends on the width of the board, the species and cut, and how far the room humidity swings. Estimate it as width times the dimensional change coefficient times the change in moisture content. For a 140 mm flat-sawn white oak plank and a 3 point swing that is about 1.6 mm; quarter-sawn it is about 0.8 mm. A laid floor moves less than a free board because the boards restrain each other.

How big a gap in winter is normal?

We found no source that sets an acceptable gap width. The sources say small seasonal gaps are natural movement, not a fault. The estimator gives a ceiling for each joint, and the partner workshop should state the gap it expects for the floor and fixing method it supplies, so that the figure can be agreed before the floor is laid.

Does quarter-sawn flooring move less?

Yes. The Wood Handbook says quarter-sawn boards shrink and swell less in width and cup less. On its coefficients quarter-sawn moves about half as much as flat-sawn for the oaks, and about two thirds as much for black walnut.

Does an engineered floor move less than solid?

The US trade body says engineered flooring is more dimensionally stable but shows low humidity differently: it can dry-cup when humidity stays below the maker's range. We found no published coefficient for engineered boards, so ask the partner for its movement figure and humidity range.

Should seasonal gaps be filled?

No, say the US trade body. Filler, some of it as hard as wood, can crush the board edges when the wood swells in damp weather and leave worse gaps. Gaps built in as washer rows should not be filled either.

What humidity range should I specify?

The ranges differ: 30 to 50 % from a US trade body, 40 to 65 % from a UK maker and 45 to 65 % from a UK distributor. Specify the range the building will actually hold and the one the partner gives for its floor, and ask for a logger to show it.

Is a wider plank always a bigger risk?

Movement across a board is in proportion to its width, so a plank twice as wide moves twice as much, other things equal. Cut matters as much: a quarter-sawn plank of the same width moves roughly half as much. The partner confirms the widths it recommends for each construction.