Wood doesn't just move seasonally. Under steam, it bends permanently. No kerfs, no fake geometry — the part that comes out of the form is actually curved. That's what happens when you heat lignin past its glass transition point and let it reharden in a new shape.
Lignin is the binder that holds the wood's cellulose fibers in position. Dry, it's rigid. Steam it to 100°C and it goes plastic. The fiber bundles start moving with you rather than against you. The window once the piece leaves the box is 30 to 90 seconds. That's it. The lignin starts stiffening almost immediately as it cools. If you don't have the form set up, the clamps open, and the bending strap fitted and ready before the piece comes out, you've already lost the bend.
The bending strap goes on the tension face. The outside of the curve. The convex side. This trips people up every time because it seems backward. The strap isn't there to prevent failure. It's there to relocate it. By limiting elongation on the outside, you push the failure mode to the compression side of the bend, where wood tolerates considerably more deformation. Without a strap, anything steeper than a very gentle radius will split on the outside face. The strap goes on the outside. Remember that.
The Species Question
Straight, clear, consistent grain is the first requirement. Non-negotiable. Interlocked grain creates unpredictable tension failures across the grain direction. Figured stock fails catastrophically. Not gently. Not in a way you can recover. This isn't a refinement point, it's a wall. If the grain in the bending section isn't straight and clear, the piece is scrap before you start.
Ring-porous species (white oak, ash) have large vessel clusters in the earlywood that create structural discontinuities under bending stress. They bend, but they want conservative radii and solid strap support. Diffuse-porous species like cherry and hard maple bend more cleanly.
For the BC Interior palette:
Ash — the best bending species if you can source it. High plasticity, consistent grain, forgiving on radius. Not common in Interior yards. Worth calling around specifically for it when you have a bending project coming.
Cherry — excellent. Straight-grained clear cherry bends cleanly, holds the form well, and finishes beautifully. One of the better picks available locally.
Hard maple — good for moderate radii with proper strap support. Select your grain carefully. Don't push the radius.
White oak — acceptable for gentle to moderate bends, quartersawn only. Flat-sawn white oak's ray structure is a liability under bending stress. Treat flat-sawn white oak the same as a species that won't bend.
Bigleaf maple — don't. The interlocked grain that makes it spectacular on a panel makes it unpredictable under steam. Put it down.
Western red cedar and yellow cedar — no. The compression-facing brittleness that makes them good under edge tools is exactly why they fail in the form. Avoid.
In the Steam Box
One hour per inch of thickness. That's heat penetration time, not surface temperature. A 1" part needs a full hour. A 1½" part needs 90 minutes. Undersizing steaming time is the most common reason first bends fail, and the failure doesn't always look like what you'd expect.
Wet steam only. The generator needs water in the chamber throughout the session. Dry heat desiccates the wood instead of hydrating it. Lignin only softens in the presence of moisture. Check your water level before the piece goes in.
Box design basics: tight interior, condensate drain at the low end, vent at the high end, dowels or rods to keep the part off the floor so steam contacts all faces. None of this is complicated. The box just needs to stay at temperature and stay wet.
Rehearse the bend dry before the piece touches the box. Not once — several times. You need to know exactly which face goes where, where each clamp seats, and how the strap positions. Fumbling with a clamp for 30 seconds is the difference between a good bend and scrap.
When Bends Fail
There are four failure modes.
Tension failure — the outside face splits. The strap wasn't used, wasn't tight enough, or the radius was too steep for the species. Not recoverable. The piece is scrap. More strap support and a more generous radius are the corrections for the next attempt.
Compression failure — the inside face buckles and corrugates. Radius too tight for the section thickness or the species. Move to lamination bending at that radius, or open the radius until the species tolerates it. Compression failures look bad, but they're telling you something useful: the radius-to-thickness math doesn't work for what you're trying to do.
Springback — the part moves back toward its original shape after you release it from the form. This is normal. All steam-bent parts do it. You calibrate by overbending on a test piece, measuring the return, and adjusting the form accordingly. The amount varies by species and thickness. You need test blanks before you commit a finished part.
End checking — splits at the end grain. The end grain dried faster than the interior during steaming. Seal end grain with melted wax or commercial end-grain sealer before the piece goes in the box. Preventable.
When Not to Steam Bend
Lamination bending is the other path, and in some situations it's the cleaner choice.
Use it when: the species isn't a bending candidate, the radius is too tight for the section thickness, you need compound curves (steam bending is single-plane), you don't have steam infrastructure, or you're making multiple identical parts that need to match.
Strip thickness math: inside radius divided by 200, in the same units. A 10" inside radius uses 0.05" (1.27mm) strips. A 20" inside radius uses 0.10" (2.54mm) strips. Slow-set PVA handles most furniture work. Epoxy for structural parts. Springback applies to lamination bending too, though it's less than steam. Still needs calibration on a test piece before you commit.
The first time a bent back rail holds its curve out of the form, something clicks. The material didn't just submit to a plan. It participated. Lignin reshaped by its own chemistry, now permanently curved. That's not metaphor. It's what thermal plasticization actually does.
Not a kerf-bent fake. The real thing.