What is this board going to do?

The premise

A tabletop that fits perfectly in a workshop in August can split its own frame in a heated dining room in January. That is not bad workmanship. It is a 900 mm oak panel losing six millimetres of width because the air around it changed, and being fastened by someone who did not leave it anywhere to go.

Almost every wood movement calculator online applies one shrinkage figure to every species and treats grain direction as a footnote. Both choices are badly wrong. Tangential shrinkage across the woods on this site ranges from 3.8% to 12.7% — a factor of more than three — and for any given board, whether it was flatsawn or quartersawn changes the answer by up to a half.

Provenance

The USDA Forest Products Laboratory has been measuring this since the 1930s and publishes the results in the Wood Handbook, a federal document with no copyright attached. Everything here traces back to it, or to Tropical Timbers of the World for the imported species it does not cover. Anything drawn from that second source is labelled as such: the two studies used different methods, and blending them without saying so would be dishonest.

The moisture equation is checkable, which is the part that matters. The Handbook states that its published moisture table was calculated from a specific equation, so that equation can be implemented and the results compared against the printed table. This one reproduces all twelve published reference values to four decimal places. That comparison is printed under sources.

Three places this disagrees with the Handbook, on purpose

The Wood Handbook publishes ready-made movement coefficients in one table and raw shrinkage data in another, and for a handful of tropical species the two do not agree. Its pre-computed figure for teak is 32% below what its own shrinkage measurements give. The convenient table carries figures from much older editions; the raw data was refreshed and the derived table was not. This site derives coefficients from the raw data using the Handbook's own stated method — which reproduces its published table exactly for 30 of the 32 species checked, and disagrees only where the published table is stale.

Second: shrinkage does not begin at the same moisture content for every wood. The Handbook assigns 22% rather than the usual 30% to eleven species, teak and western redcedar among them. That is a 36% difference in the divisor, and a calculator that hardcodes 30% is wrong by more than a third on those woods.

Third: there is no published coefficient for riftsawn timber anywhere in the Handbook. Several calculators offer a riftsawn option regardless. This one offers it too, and says on the results that it is an interpolation rather than a measurement.

What it will not do

It will not predict what an individual board does. The Handbook is blunt on this and the sentence is worth repeating: accurately predicting the shrinkage of an individual piece of wood is impossible, though the average of a quantity of pieces can be predicted well. Shrinkage varies by roughly 15% between boards of the same species, and rather more in commercial lumber than in the clear samples the figures were measured on. Treat every result here as the centre of a band.

It also will not let an absurd input through quietly. A ten-point moisture swing entered into most calculators returns an arithmetically perfect answer that would have you leaving 38 mm of clearance on a dining table. The arithmetic is not the problem. Nothing on the first page of results flags it, so these tools do.