Shrinkage and movement by species
Radial and tangential shrinkage for 106 woods, the ratio between them, and the coefficients derived from them. Sortable — click any heading.
| Species | Radial % | Tang. % | T/R | C₃ | Cₜ | FSP | SG | Janka |
|---|
The column nobody else publishes
T/R is tangential shrinkage divided by radial shrinkage, and it is the number that predicts cupping. Movement across the width reports how far a board changes size; the ratio reports how hard it fights itself. A wood that shrinks a lot but shrinks evenly stays flat. A wood that shrinks modestly but very unevenly will curl.
Sort by it and the first entry is a surprise. Eastern white pine has the worst T/R ratio in the entire dataset at 2.90 — worse than any oak, worse than beech. Nobody thinks of white pine as unstable, because its absolute movement is small and so the width change is unremarkable. But a wide flatsawn white pine board cups disproportionately hard, and that is invisible in every shrinkage table published online because they only list the two raw figures and never divide them.
The other direction is just as useful. Black walnut sits at 1.42 and northern red oak at 2.15 with broadly comparable tangential shrinkage. Walnut's advantage over oak is not that it moves less — it barely does — but that it moves evenly, so it stays flat. That is most of the reason walnut has the reputation it has, and no table on the web shows it.
Teak's reputation is about the wrong number
Teak is famous for stability, and it earns that: 5.8% tangential shrinkage is genuinely low. But its T/R ratio is 2.32, which puts it in the worst dozen woods here. It is stable in the sense that it does not move much, not in the sense that it resists cupping. Flatsawn teak in a wide board will still curl.
Fibre saturation point is not 30% for everything
Shrinkage begins when the cell walls start giving up water, and the moisture content at which that happens is conventionally taken as 30%. The Handbook itself assigns 22% to eleven of the woods listed here, in a footnote to its own coefficient table: western redcedar, northern white-cedar, both redwoods, teak, genuine mahogany, iroko, greenheart, kokrodua and the keruings.
That is not a rounding difference. It is a 36% change in the divisor, and it flows straight into every coefficient derived from it. A published calculator stating flatly that fibre saturation is assumed to be 30% for all species is wrong by more than a third on teak and western redcedar — by the Handbook's own footnote.
Two specific gravities, ten per cent apart
Specific gravity always counts the wood ovendry, but the volume it is measured against varies, and the Handbook uses both. Domestic species are given on volume at 12% moisture content; imported species are given on green volume. Across the common woods those two bases differ by about 10% on average, and by 15% for basswood and beech. They are quoted interchangeably all over the web under one heading.
Every figure here says which basis it is on. Where only the green-volume figure exists, the weight calculator declines to convert rather than guessing.
Asked at the bench
Why are some rows marked AH-607?
Those species are not in the Wood Handbook at all, and the figures come from Tropical Timbers of the World — also USDA, also public domain, but a different study with different methods. They are labelled because mixing two datasets silently would make them look more comparable than they are.
Where is osage orange?
Not here, because no USDA source publishes shrinkage, specific gravity or hardness for it that I could find. Figures for it circulate online with no traceable origin. An empty row is more honest than a borrowed one.
Why is there no riftsawn column?
Because no riftsawn coefficient exists in any primary source. The Handbook classifies lumber as plainsawn below 45 degrees of ring angle and quartersawn above it, and uses the word "rift" only as an oak flooring grading term. Several calculators offer a rift figure without mentioning that they invented it. The tools here will give you one and tell you it is an interpolation.
Are these figures for green to ovendry? I never dry wood that far.
They are, and you never will. Green-to-ovendry is simply the reference range the measurements were made over. The coefficients in the two derived columns are what you actually use — they express change per single percentage point of moisture content over the normal in-service range, which is where all real work happens.