Wood moves. Not “might move” or “moves if you get it wrong” — it moves, every year, for as long as the piece exists. Almost every furniture failure we are asked to repair comes down to somebody building as though it did not.
Stickered and racked. A year per 25mm of thickness, and no shortcuts that have ever worked.
It only moves in two directions
Length is effectively fixed. A two-metre board is the same two metres in January and August. Width and thickness are not: they swell as humidity rises and shrink as it falls, and width is where you notice it because there is more of it.
Quartersawn stock moves about half as much across its width as flatsawn, which is why it costs more and why it is worth it on a wide tabletop.
Species
Flatsawn
Quartersawn
Oak
~8mm
~4mm
Walnut
~6mm
~3mm
Beech
~10mm
~5mm
Pine
~6mm
~3mm
Approximate movement across a 600mm width, between a damp summer and a heated winter.
Building for it
Never glue a wide panel into a groove. Let it float.
Fix tabletops with buttons or slotted brackets, never screwed hard through a rail.
Do not cross the grain. A batten screwed across a wide board will split it, given a winter.
Finish both faces equally. A top sealed on one side only will cup towards the unfinished face.
Acclimatise, then build
Bring timber into the room it will live in and leave it a fortnight before cutting joints. A board that arrives at 14% and is built at 14% into a house that sits at 9% will do all of its shrinking after assembly, which is exactly when you do not want it.
You cannot stop it. You can only decide, in advance, where it is going to go.
A toybox in knot-free pine for a customer’s grandchild, built over a wet weekend. Simple work, but a few decisions in it are worth writing down.
Carcass
Through dovetails at the corners, which is more joint than a pine box strictly needs, but it is the kind of thing that gets kept. Bottom in a groove, floating, because even pine this dry will move.
The lid
Soft-close stay, non-negotiable on anything a child opens.
Finger gaps under the lip on both sides, so a lid cannot trap a hand.
Ventilation holes in the back panel, low down where they do not show.
Finish
Hard wax oil, two coats, fully cured for a fortnight before it went anywhere near a child. Pine takes oil unevenly around the knots and there is not much to be done about it except choose boards without many.
The fence that came with the router table was a folded steel channel with two plastic knobs. It flexed visibly under a push stick, which is the one thing a fence must never do.
The replacement is two pieces of 18mm birch ply laminated face to face, with a hardwood lip glued along the bottom edge to keep it flat.
Both faces are split so they can be brought in around the cutter. Sliding faces make a jointing pass possible without shimming anything.
Dust extraction is a 63mm port let into the back of the fence, which pulls perhaps eighty per cent of what the cutter throws.
Total cost was one sheet offcut and an afternoon. It is the single best improvement the table has had.
Most of the questions we get about finishing are really questions about maintenance. What people want to know is not which finish looks best on the day, but which one they can live with in five years when a mug has been left on it.
The same oak offcut, four finishes, one coat each. The differences get larger with the second coat, not smaller.
The two families
Finishes either sit on the wood or soak into it. Film finishes — lacquer, varnish, shellac — build a layer on the surface. They protect well and they fail visibly: a scratch goes through the film and catches the light. Penetrating finishes — oils and hard wax oils — go into the fibres. They protect less and they fail invisibly, which usually means you can fix them with a rag.
Neither is better. The question is who is going to look after the piece, and how much they want to think about it.
Finish
Water resistance
Repairable
Coats
Hard wax oil
Good
Very — spot repair with a rag
2
Danish oil
Moderate
Very
3–4
Shellac
Poor
Yes, dissolves into itself
4–6
Polyurethane
Excellent
No — strip and redo
2–3
What we use, and when
Hard wax oil on nearly everything. It gives a finish close to bare wood, it is repairable by anyone with a cloth, and two coats is genuinely enough. On a dining table that will see daily use, we add a third coat on the top only.
Danish oil on tool handles and small items, where the slightly softer surface is an advantage and the piece is easy to re-oil. Shellac under a wax on anything that will never see a drink — boxes, frames, the inside of cabinets.
Polyurethane almost never, and only on request. It is the toughest thing on the list and the least pleasant to live with: it looks like plastic under raking light, and when it does eventually fail there is nothing to do but strip the whole surface.
Kitchen surfaces
Food contact: anything that will meet food directly — boards, bowls, spoons — gets food-safe mineral oil and nothing else. Not because the others are dangerous once cured, but because mineral oil is the only one a customer can reapply from a supermarket without asking us first.
Test on the offcut
Every board finishes differently, and the sample in the tin lid was made on something that is not your timber. Keep the offcuts from the piece you are building, finish them alongside, and look at them in the light of the room the furniture is going into. It takes an afternoon and it has saved us from several expensive opinions.
If the dovetail is the joint people photograph, the mortise and tenon is the joint that holds the furniture up. Almost every chair, table and door in the workshop depends on one. It is also far more forgiving than it looks, which is why we teach it first.
A haunched tenon for a table apron, dry fitted before the shoulders are trimmed.
Proportions
The old rule is one third: a tenon one third the thickness of the stock, centred. It survives because it works — a third leaves enough cheek on either side of the mortise that the walls do not split, and enough meat in the tenon that it does not shear.
In practice you round to the nearest chisel. A 21mm leg with a 7mm tenon is arithmetically correct and annoying to cut; a 6mm tenon chopped with a 6mm mortise chisel is very slightly weaker and takes half the time. Take the chisel.
Mortise first
Always cut the mortise first and fit the tenon to it. A mortise is a hole with fixed walls and there is very little you can do to adjust it after the fact. A tenon is four surfaces you can pare. Working the other way round means adjusting the thing that cannot be adjusted.
Gauge both walls from the face side, so any error in thickness lands on the back.
Chop from the middle outwards, working back towards each end.
Stop about 2mm short of the ends until the depth is right, then pare to the line.
Clear the chips often. Levering against a packed mortise is how you bruise the top edge.
Cutting the tenon
Shoulders define the look, cheeks define the fit. Cut the shoulders with a knife line and a fine backsaw, and take your time — a shoulder that is not square shows as a gap on the outside of the finished piece, where everyone will see it. The cheeks can be rough; nobody sees them and glue does not mind.
Aim for a fit that goes home with firm hand pressure. If you need a mallet, it is too tight, and driving a tight tenon into a mortise splits the leg — often days later, after the glue has dried and the piece is finished.
Haunches and wedges
A haunch is the small step left at the top of a tenon where it meets the end of a leg. It stops the rail twisting and it fills the groove if the piece is grooved for a panel. It costs one extra saw cut and it is almost always worth it.
Wedges go across the grain of the mortised piece, never with it, and they want kerfs cut before assembly. Two thin wedges beat one thick one. Cut them from the same board as the tenon so they move together and the colour matches.
A joint that needs a mallet to close will need a clamp to stay closed, and a repair to stay together.
Common failures
Nearly every failed mortise and tenon we see in repair work failed for one of three reasons: the joint was too tight and split the leg, the shoulders were not square so only half the joint was ever in contact, or it was glued in end grain and never had a chance. None of those is a skill problem. They are all a patience problem.
