Wood movement will destroy your build — here's the arithmetic

A tabletop glued up in a humid August shop and delivered to a heated house in January is running an experiment whether the builder intended it or not — the wood is going to lose moisture, and losing moisture means losing width. Most of the "the wood just cracked, no idea why" stories on any woodworking forum are that experiment running to its logical conclusion. The arithmetic behind it is simple enough to do on the back of a cut list, and simple enough that there's no excuse for skipping it on anything wider than a pencil.

What's actually moving, and why it stops

Wood shrinks and swells across its width and thickness as its moisture content changes — length barely moves at all, which is why this is a width problem, not a length problem. Below roughly 30% moisture content (the fiber saturation point), the water still in the wood is bound directly into the cell walls, and as that bound water leaves, the walls themselves get thinner and the board gets narrower. Above that point, any additional water is just sitting as free liquid in the hollow centers of the cells, and pulling it out doesn't change the wood's dimensions at all — which is the whole reason kiln drying can knock a green board from 60% MC down to 30% without the board changing size, and then every remaining point of moisture loss below 30% costs real, measurable width.

The relationship is linear once you're below that threshold: a species' total shrinkage percentage (published as its green-to-oven-dry shrinkage, separately for the radial and tangential directions) divides by the fiber saturation point to give a movement coefficient, and that coefficient times the starting width times the change in moisture content gives the dimensional change in inches. No calculus, no lookup table beyond one number per species — just a species constant and two moisture readings.

A worked case: the top that shrank in January

Take a flatsawn black walnut tabletop, glued up 42 inches wide, finished and delivered at the end of a humid summer when the shop (and the wood) sat around 12% moisture content. By the time the first heating season is over, the house's indoor air has pulled that board down to something closer to 6% MC — a six-point swing that's entirely ordinary for a heated home in a four-season climate. Black walnut's tangential shrinkage figure is 7.8%, so the movement coefficient is 7.8 ÷ 30 ÷ 100 = 0.0026 per inch per point of moisture change. Multiply that out: 42 × (6 − 12) × 0.0026 = −0.655 inches. Two-thirds of an inch of width, gone, on a board that looked completely finished the day it left the shop.

Cut the same top from quartersawn stock instead and the number changes because the movement axis changes: walnut's radial shrinkage is 5.5%, giving a coefficient of 0.00183 and a movement of 42 × (6 − 12) × 0.00183 = −0.462 inches for the identical humidity swing — noticeably less, but still not nothing, and still enough to open a real gap in any construction that assumed the board was done moving once it was dry to the touch.

Now compare that to a narrow part on the same project — say a 3-inch-wide interior rail, going the other direction, gaining moisture from 6% back up to 12% as summer humidity returns. Same species, same coefficient, opposite sign: 3 × (12 − 6) × 0.0026 = +0.047 inches. Less than a sixteenth of an inch. This is the part of the arithmetic that trips people up before they've run it themselves: it isn't that wood "moves a lot" or "moves a little" as a species trait alone — width is doing as much work in the equation as the species coefficient is, which is exactly why the same lumber behaves completely differently in a drawer rail than it does in a glued-up top.

What actually breaks

A rigidly restrained wide panel is where this arithmetic turns into a repair bill. A tabletop screwed straight through an apron with no slotting, a breadboard end glued along its entire length instead of just at the center, or a wide panel glued into a dado on all four sides — every one of these designs assumes the wood is a fixed dimension, and every one of them is wrong by exactly the number this equation produces. When the movement has nowhere to go, something gives: the screw hole elongates and cracks the apron, the breadboard end splits the panel it was supposed to be capping, or the panel itself splits along a growth ring because that's genuinely the weakest path available to relieve the stress. None of this is a material defect or "bad wood" — it's a design that didn't leave room for a number that was always going to be nonzero.

Building the allowance in, not discovering it later

The fix isn't avoiding movement-prone species or over-drying everything to some theoretical zero-humidity target — it's sizing the joint to the number instead of guessing at it. A floating panel in a frame gets a groove deep enough to swallow the panel's own worst-case shrinkage and swelling, not just today's fit. A breadboard end gets elongated screw holes sized to the same number, fastened only at the center so the ends are free to slide. And a top attached to an apron uses figure-eight fasteners or slotted clips instead of screws driven straight through, specifically so the top can walk in and out of its rigid understructure across the seasons without pulling anything apart.

Timing the glue-up matters almost as much as the joinery

Building a wide panel at whichever moisture extreme the shop happens to be sitting at when the humidity is furthest from the piece's eventual destination is asking the joinery to absorb the full swing in one direction only. Gluing up a tabletop in a damp August shop for delivery into a dry, centrally-heated winter home means the panel only ever shrinks from the day it's finished — every bit of the movement this equation predicts shows up as the wood pulling away from whatever restrains it, never as it pushing outward. Building the same top in a dry winter shop for the same destination splits the difference: the panel arrives close to its final moisture content already, and whatever swing is left runs in both directions across the year instead of only one. None of this makes the arithmetic optional — it just changes which side of zero the number the calculator returns is going to be, and a joint built to expect motion in only one direction is a joint that's already halfway to the same failure this post opened with.

Running the actual numbers for a specific board — its species, its width, and the real humidity swing it's going to see — takes less time than sanding the piece you're about to glue up, and it's the difference between an expansion gap you planned for and a crack you didn't. Try the wood movement calculator with your own project's width and species, check the destination's likely humidity range on moisture content targets by region, and see breadboard end for how the slotted-screw fix is actually built. If you're trying to work out exactly how deep a groove or how wide a gap a specific panel needs, how much gap should I leave for a breadboard end gives a fast single-number answer; for the meter-reading side of this problem — knowing your actual starting moisture content in the first place — see reading moisture content.