What makes wood move, and in which direction
Wood cells hold water in two places: free water in the cell cavities, and bound water in the cell walls themselves. Free water leaves first and changes nothing dimensionally. It is only when the cell walls start giving up bound water — below the fibre saturation point, around 28% moisture content for most species — that the wood begins to shrink. Above that point it does not move at all; below it, movement is close to proportional to moisture content.
The movement is wildly unequal in the three directions. Along the grain it is negligible, on the order of 0.1% across the entire range from green to oven dry. Across the grain it is large, and it is larger tangentially — around the growth rings — than radially, typically by a factor of about two. Red oak shrinks 0.00369 of its width per point of moisture content tangentially and 0.00158 radially, a ratio of 2.3.
That ratio is why grain orientation is a design decision and not a matter of taste. A flatsawn board, with its growth rings running roughly parallel to the face, presents its tangential direction across the width and moves the full amount. A quartersawn board, with the rings running through the thickness, presents its radial direction and moves less than half as much. It is also why flatsawn boards cup and quartersawn ones stay flat: on a flatsawn board the face nearer the bark is more tangential than the face nearer the pith, so the two faces shrink by different amounts.
Moisture content in service is set by the surrounding air, not by how well the wood was dried. Kiln-dried lumber brought into a heated house in January will keep drying until it reaches equilibrium with that air, and it will take moisture back up in July. Drying is the starting point; the environment decides the rest.
The dimensional change coefficient
The USDA Forest Products Laboratory publishes, in the Wood Handbook, a dimensional change coefficient for each species in each direction. It is the fraction of the dimension that changes for each one-point change in moisture content, and it turns the whole question into one multiplication: ΔW = W · C · (M₂ − M₁).
Three things about that expression are worth being precise about. First, C is a fraction, not a percentage — red oak's tangential value of 0.00369 means 0.369% of the width per point. Second, the reference width W is the width at the starting moisture content; the error from using the finished width instead is well under a percent of the answer and does not matter at shop tolerances. Third, the moisture difference is in points of moisture content, not percent change in moisture content: going from 6% to 12% is six points, not a doubling.
Which coefficient applies depends on how the board was sawn. Flatsawn stock uses the tangential value; quartersawn and riftsawn use the radial. A glued-up top from random boards contains some of each, so the mean of the two is a defensible working figure — but if you want a safe number for joinery clearance, use the tangential value and accept that the panel may move slightly less than predicted.
The linear fit is calibrated across roughly 6% to 14% moisture content, which happens to cover almost everything a piece of interior furniture experiences. Outside that band the relationship curves, and the coefficient overstates movement at the very dry end. The calculator warns you when your range strays outside it.
To turn a predicted movement into joinery, you need one more decision: where the panel is anchored. A tabletop screwed solidly at its centre and slotted everywhere else splits its movement between the two halves, so the outermost fastener sees half the total. Anchor the panel at one edge instead — often the front edge, so the overhang stays constant — and the far edge sees all of it.
Worked example: a 36 in red oak tabletop
You are building a flatsawn red oak dining table, 36 in across the grain. Your shop reads 8% moisture content in the wood on glue-up day. The house it is going to runs about 6% in winter and about 12% in an August with the windows open.
- Coefficient. Flatsawn, so use the tangential value for red oak: C = 0.00369 per point.
- Moisture swing. 12 − 6 = 6 points.
- Total seasonal movement. 36 × 0.00369 × 6 = 0.797 in, or 20.2 mm. Call it 13/16 in.
- Percent change. 0.00369 × 6 × 100 = 2.214% of the width.
- Swell from build to damp. 36 × 0.00369 × (12 − 8) = 0.531 in wider than the day you glued it.
- Shrink from build to dry. 36 × 0.00369 × (6 − 8) = −0.266 in, so a quarter inch narrower.
- Fastener travel, centre-anchored. 0.797 ÷ 2 = 0.399 in at each outer fastener. Slot for a #8 screw with a 0.164 in shank: rout the slot 0.399 + 0.164 = 0.563 in long, so a 9/16 in slot, and centre it on the screw's position at your build moisture content.
Compare the same top quartersawn: 36 × 0.00158 × 6 = 0.341 in, which is 43% of the flatsawn movement. That single choice cuts the movement by 0.456 in, which is the difference between an obvious seasonal gap at a breadboard end and one you have to look for.
Seasonal movement of a 36 in panel, 6% to 12% MC
| Species | Tangential C | Flatsawn (in) | Radial C | Quartersawn (in) |
|---|---|---|---|---|
| American beech | 0.00431 | 0.931 | 0.00190 | 0.410 |
| Shagbark hickory | 0.00411 | 0.888 | 0.00259 | 0.559 |
| Red oak, northern | 0.00369 | 0.797 | 0.00158 | 0.341 |
| White oak | 0.00365 | 0.788 | 0.00180 | 0.389 |
| Hard maple | 0.00353 | 0.762 | 0.00165 | 0.356 |
| Yellow birch | 0.00338 | 0.730 | 0.00256 | 0.553 |
| Yellow-poplar | 0.00289 | 0.624 | 0.00158 | 0.341 |
| Black walnut | 0.00274 | 0.592 | 0.00190 | 0.410 |
| White ash | 0.00274 | 0.592 | 0.00169 | 0.365 |
| Douglas-fir, coast | 0.00267 | 0.577 | 0.00165 | 0.356 |
| Southern pine | 0.00259 | 0.559 | 0.00165 | 0.356 |
| Black cherry | 0.00248 | 0.536 | 0.00126 | 0.272 |
| Honduras mahogany | 0.00238 | 0.514 | 0.00172 | 0.372 |
| Ponderosa pine | 0.00216 | 0.467 | 0.00133 | 0.287 |
| Eastern white pine | 0.00212 | 0.458 | 0.00071 | 0.153 |
| Teak | 0.00186 | 0.402 | 0.00101 | 0.218 |
Beech flatsawn moves 0.931 in where eastern white pine quartersawn moves 0.153 in — a factor of 6.1 across the same panel width and the same six points of moisture. Species and sawing choice together matter more than anything else you can decide.
Turning the number into joinery
The figure to design around is the total seasonal range, not the shrink or the swell alone, because the joint has to survive both ends of the year. What changes with your build moisture content is where in that range you start.
Build at the middle of the range and the panel moves half the total in each direction, which is the easiest case: a fastener slot centred on its screw has half the travel available on each side. Build at the dry end and the panel only ever swells, so all the clearance has to sit on one side of the screw. This is the practical reason to bring lumber into the shop and let it sit until it stops changing — you want to know which case you are in.
Rules of thumb for the common joints follow directly. A tabletop fastened to an apron needs slots or figure-eight fasteners giving the travel the calculator reports, plus the screw shank diameter. A breadboard end is glued only at its centre few inches, with the outer tenons pinned through elongated holes sized to the travel from the centre to that pin. A panel in a frame-and-panel door must float in its groove with clearance equal to the swell from build MC to the damp end, and the groove must be deep enough that the shrink from build MC to the dry end does not pull the panel out of it. Drawer bottoms and case backs follow the same logic.
Two things reduce the number rather than accommodate it: quartersawn stock, which roughly halves it, and a stable substrate. Plywood and MDF move about a twentieth as much as solid wood across their faces, which is why veneered panels are the standard answer for a wide, fixed surface. A veneered top on a plywood core can be screwed down solid.
Finish does not stop movement. A film finish slows the exchange of moisture — a well-finished piece responds to a change in humidity over weeks rather than days, and both faces should be finished so they respond at the same rate — but the equilibrium the wood eventually reaches is set by the air, not by the finish. Confirm where you are starting with the wood moisture content calculator.
Mistakes that split a top
- Screwing a solid top down through fixed holes. The commonest failure in amateur furniture. The top cannot get narrower, so it splits instead — usually along the glue line nearest the middle.
- Gluing the whole length of a breadboard end. The breadboard runs long grain across the top's cross grain. Glue it all the way and one of the two has to give.
- Gluing a floating panel into its frame. A panel glued at the edges is a fixed panel, and it will crack or force the frame joints apart.
- Treating a fitted drawer as dimensionally stable. A drawer fitted with a shaving of clearance in February will bind in July. Fit for the season you are in and allow for the other.
- Building at whatever moisture content the lumber arrives at. Boards straight off a truck can be at 12% or more. Sticker them in the shop until a meter shows two consecutive stable readings.
- Using the tangential coefficient for a quartersawn top, or the radial one for flatsawn. The two differ by roughly a factor of two, so this is not a small error in either direction.
Where the coefficients come from
The tangential and radial dimensional change coefficients used here are the values published by the USDA Forest Service Forest Products Laboratory in the Wood Handbook: Wood as an Engineering Material, the reference every wood technology text builds on. They are species averages measured on small clear specimens, so an individual board — especially one with irregular grain, tension wood or a wide density range between earlywood and latewood — can differ noticeably. Use them for design clearance, not for a precision fit, and give yourself margin.
Related calculations and the limits of this one
This page answers one question: how far a given width of a given species moves between two moisture contents. It does not tell you what those moisture contents will be. That depends on the equilibrium moisture content of the air in the room, which is a function of relative humidity and temperature, and a well-heated house in a cold climate can drop below 5% while a coastal summer pushes past 13%. If you do not know your own numbers, put a hygrometer in the room for a year before you commit to a design that depends on them.
Nor does it model cup, bow or twist. Those come from differential movement — one face drying faster than the other, or the tangential direction moving more than the radial across a single flatsawn board — and they are the reason a wide flatsawn board cups away from the bark side as it dries. Alternating the growth ring direction in a glue-up spreads that cupping into a series of small waves rather than one large curve, which is a cosmetic improvement rather than a cure.
Two neighbouring calculations come up in the same projects. If you are working out how much stock to buy for a top, size the boards with the cut list calculator and remember to leave width for the movement plus a final trim. If the panel is a shelf rather than a top, its stiffness is usually the binding constraint rather than its movement — check it with the shelf sag calculator. The same shrinkage arithmetic, with different coefficients, governs clay shrinkage and fabric shrinkage, both of which ask you to build oversize by a known percentage for the same underlying reason.