Grain & Edge
The Art of Wood

A tree is a pump that stands still

Start with the tree's problem, because wood is the answer to it. A tree has its mouth in the ground and its lungs in the sky, and it cannot walk between them. Water sits in the soil at the bottom. Light falls on the leaves at the top. Something must carry water up — thirty metres up in a big one — every day for a century, with no heart to pump it and no chance of moving somewhere easier.

The answer the tree arrived at is plumbing. It builds itself out of tubes: millions of narrow hollow cells, each one only a few millimetres long, stacked end to end in unbroken files that run from root to crown — every cell a tiny pipe with a stiff fibrous wall around a hollow middle. Water climbs them, cell into cell, the whole height of the tree. Each year a new layer of tubes is added on the outside, and the older tubes deeper in stop carrying water, stiffen, fill with the tree's own gums and resins, and quietly change job from plumbing to structure.

That is all a trunk is. Not a solid rod that happens to have lines on it — a standing bundle of drinking straws, glued along their sides. The tree got its water lifted, and as a by-product got a mast that holds tonnes of branches out into a gale for a hundred winters.

Two words before we go on, both plain. The stiff stuff the tube walls are built from is cellulose — long, strong, thread-like molecules, the same material as cotton and paper. The stuff gluing the tubes to one another and stiffening their walls is lignin — a natural resin, and the reason a tree stands up while a wet paper straw does not. Cellulose is the rope. Lignin is the glue. Rope and glue, in tubes. Hold that and you can derive the rest.

The one fact that explains everything

Here is the sentence the whole course rests on. It is not a poetic comparison. It is a flat description of the thing in your hand:

Pick up any piece of wood and find the tubes. On a sawn face you will see lines running its length: those lines are tubes seen from the side. On the sawn end you will see rings and tiny pores: the same tubes, seen down their throats. The lines and the rings are not a pattern printed on the wood. They are the wood.

The direction the tubes run has a name, and it is the most-used word in the craft: the grain. "With the grain" means along the tubes. "Across the grain" means through them sideways. End grain is the cut face where you are looking straight into the open ends, as though down at a bundle of straws standing in a jar. Long grain is any face that runs along their sides.

Accept the straws and a whole shelf of woodworking lore stops being lore and starts being obvious:

• Why a chair leg carries your weight. You are loading a bundle of tubes end-on, which is the one thing tubes are magnificent at.
• Why firewood splits with a tap but will not be chopped through. Parting straws sideways is nearly free. Severing every straw is expensive.
• Why a board cups, and always in the same direction. Part Three, and it is the most useful idea in this course.
• Why a plane leaves glass one way and torn fur the other. Part Four.
• Why a beautifully made tabletop can tear itself in half one January. That one has cost people months of work, and it was entirely predictable.

Reading the end of a board

The cut end of a log is a diagram of everything the tree ever did. It takes two minutes to learn to read, and it will tell you — before you touch a tool — how that board intends to behave.

In the middle, a small dark dot or soft core: the pith, the original seedling stem that the whole tree was built around. Around it, the rings. Each growth ring is a year's work in any climate with a winter: a band of wide, thin-walled, fast tubes grown in the rush of spring, then a band of narrow, thick-walled, dense tubes grown as the season slows. One pale band plus one dark band is one year. Count from pith to bark and you have the tree's age, and the spacing tells you which years were kind to it.

The outer wood is usually paler: the sapwood, tubes still doing plumbing. The darker core is the heartwood, retired tubes plugged and loaded with the tree's own extracts. Those extracts are why heartwood of oak, sweet chestnut, cedar or yew shrugs off rot and insects while sapwood of the same tree does not. Heartwood is not stronger than sapwood. It is better preserved.

And running the other way — the exception that proves the bundle — are the rays: thin ribbons of cells running like spokes from the pith outward, carrying food sideways across the trunk. In oak they are big enough to see with the naked eye, and they show as bright silvery flecks on the right cut. Rays matter more than they look, because they are a built-in fault line: the direction along which wood splits most willingly of all.

So a piece of wood has three directions, not one, and they behave like three different materials:

• Along the grain — up the tubes, the way the tree grew. Enormously strong. Barely moves with moisture. Splits like nothing at all.
• Radially — from the pith outward, along the rays. The middling direction.
• Tangentially — round the trunk, the way the rings curve, parallel to the bark. The restless direction, and the source of nearly all your trouble.

Enormously strong along, feeble across

Take one drinking straw and stand it on end on the table. Press straight down on it with a finger. It holds a surprising amount before it folds. Now lay the same straw flat and press down on its side. It goes flat instantly, for nothing.

That is the whole of wood's strength in one gesture, and there is no more to it. Along the tubes, a load runs down the walls — rope-like fibres pulling in exactly the direction the load is going. Across the tubes there is nothing but hollow space and side-walls being asked to work sideways, which is not what they were built for.

The two are not close. Pulled along the grain, clear straight-grained wood is one of the great structural materials — weight for weight it holds its own against mild steel, which is why aeroplanes were built of spruce and plywood well into the age of metal, the de Havilland Mosquito of the 1940s among the fastest aircraft of its day. Pulled across the grain, that same wood is perhaps a twentieth to a fortieth as strong. You can tear it with your hands.

Now look at any chair. The legs run with the grain top to bottom, so your weight presses down the tubes. The rails run with the grain end to end, so they are pulled along the tubes. Nobody in the whole history of chairmaking has cut a leg with the grain running across it, and it is not tradition stopping them. A crossways leg would snap under a child.

Short grain — the invisible fault

If wood is strong only along the tubes, then a piece is only as strong as the way the tubes run through it — not the way its outline is shaped. Most broken woodwork comes from here, and it is invisible unless you have been told to look.

Picture sawing a curved chair back out of a wide flat board. The outline is a graceful arc; the tubes in the board still run dead straight, front to back. So at the top of the arc, where the curve turns hardest, the tubes cross the part — entering one edge and leaving the other within a few centimetres. That region is short grain: a place where no fibre runs the length of the part, only stubs of them. It looks exactly like the rest of the wood. It is a crack waiting for an excuse.

Its close relative is run-out: fibres running gradually out through a face instead of along it. A board can look perfectly straight-grained and be full of run-out, because the saw cut a straight line through a tree that grew with a lean or a spiral. Every fibre that leaves through the surface is a fibre that no longer carries load from one end of the piece to the other.

Which is why a tool handle is chosen at its end, not by its price ticket. Look down the end of an ash or hickory axe handle: the growth rings should stand in the plane of the swing, and the lines along the side of the handle should run its length, not wander off the side within a hand's width. A handle with run-out at the neck will break on some swing, at the worst possible moment. Not luck. The fibres either run through or they do not.

So the answer to a curved part is never "saw it from a wide board and hope". It is one of three: rive it — split it, so the fibres are forced to follow their own true line and cannot run out; steam-bend it — bend a straight piece so the fibres follow the curve; or laminate it — glue thin strips around a form, each strip bending with its fibres intact. Three answers, one reason.

Knots — where two bundles met

A knot is not a blemish in the wood. A knot is a branch, cut through, still sitting where it was when the trunk grew around it.

A branch has its own bundle of tubes running out sideways, at right angles to the trunk's. Where the two meet, the trunk's fibres cannot go straight — they part and sweep around the branch like water round a rock, then close up beyond it. A knot is therefore a small region where every rule of direction is locally suspended: fibres pointing three ways at once, end grain surfacing in the middle of a face, and a wake of swirling grain around it.

Everything a knot does to you follows from that:

• It weakens the piece more than its size suggests — the fibres that should have run straight past have been diverted, and diverted fibres carry load badly. A knot at the edge of something that bends matters far more than the same knot in the middle of it.
• It tears out under a plane, because there is no single grain direction to plane with.
• It is harder than the wood around it — branch wood is denser and often resin-soaked — so it blunts edges and stands proud when you sand.
• A dead black knot falls out and leaves a hole, because that branch had already died and was simply buried; the trunk's fibres never joined it. A live knot, pale and continuous with the wood around it, stays.
• It moves differently from its surroundings when moisture changes — which is why knots ring themselves with fine cracks.

None of which makes knots wicked. They are honest, often beautiful, and the right choice for a hundred jobs. But you place them on purpose: away from the edge of anything that bends, away from where a joint must be cut, and out of tool handles altogether.

The axe splits, the saw crosscuts

There are only two jobs in wood: parting the straws sideways, and severing the straws through. Every edge tool ever made is built for one or the other — and once you know which, you can pick up an unfamiliar tool and predict what it wants.

Parting them sideways is cheap. The tubes are only glued to their neighbours along their sides, and that glue line is the weakest thing in the material. A wedge driven in has only to break that side-glue — and once a split starts it runs ahead of the tool on its own, which is why a log bursts with a bang and the crack has reached the far end before the axe is halfway in.

Severing them through is expensive. Now you are cutting the rope itself, millions of walls, one at a time, and nothing happens anywhere except exactly where the steel is.

Look at what those two facts produced in the tool rack:

• A splitting axe or maul is fat — a stubby wedge with steep cheeks. It is not meant to cut deep; it is meant to enter a little and then lever the sides apart. Sharpness barely matters. Shape does everything.
• A felling axe is thin and keen, because felling and limbing cut across the grain, where steel must sever fibres rather than push them apart.
• A froe — a thick blade with the handle at right angles — does not cut at all. You start it in the end grain and lever, steering a split down a log for shingles, laths, basket splints and chair parts. Riven wood can have no run-out, because the split follows the fibres wherever they truly go instead of where a saw wished they went. That is why a riven oak shingle outlives a sawn one, and why traditional chairmakers rive their legs and rails.
• A rip saw's teeth are tiny chisels, filed straight across, paring out the bottom of a cut that runs along the grain.
• A crosscut saw's teeth are tiny knives, filed to points with bevels, each one scoring and severing tube walls before the waste is lifted away. Use a rip saw across the grain and it tears; use a crosscut saw along it and it crawls. Same steel, same arm — only the geometry knows which job it is doing.

The two kinds of water in a piece of wood

The tubes were plumbing. They do not stop being plumbing when the tree is felled — they simply change what they are plumbing for. A piece of wood in your workshop is still taking on and giving up water, every day, in step with the air around it, and it will go on doing that for three hundred years.

But the water sits in two quite different places, and telling them apart is the key to everything in this part of the course.

Free water sits in the hollow middles of the tubes, sloshing about in the pipes. A freshly felled tree is drenched with it — green wood, as it is called, can weigh more in water than it does in wood. When green wood dries, this is what leaves first.

Bound water is held inside the tube walls themselves, clinging to the cellulose fibres, wedged between them and pushing them apart. This is the water that matters.

The moment when the hollows are empty and the walls are still full has a proper name — the fibre saturation point — and in most timbers it falls near thirty parts of water to a hundred parts of dry wood. Above it, wood is heavy and does not move. Below it, every change in the air changes the size of your work. Say it plainly and you never need the phrase again: wood moves only after the walls begin to give up water.

And it never stops. Wood always drifts toward balance with the air around it — damper air, and it takes water back on and swells; drier air, and it gives water up and shrinks. Outdoors under cover in a temperate climate it settles somewhere near a sixth of its dry weight in water. Inside a heated house in midwinter it may fall to half that. Those two numbers are the whole reason furniture built in a damp barn splits in a centrally heated flat.

So there is no such thing as "dry wood" in the sense of finished, settled, safe. There is only wood that has come into balance with a particular place. Move it and it moves.

Movement is across the straws, never along them

Now put the two ideas together — the bundle, and the water in the walls — and the most useful rule in woodwork falls straight out.

When bound water leaves a tube wall, the wall gets thinner. Thinner walls mean each tube's diameter shrinks a little. But the tube does not get shorter, because its length is set by those long cellulose ropes running end to end, and drying does not shorten a rope, it only lets the strands pack closer side by side.

So the whole bundle gets narrower without getting shorter. That is the entire mechanism.

The sizes involved are worth saying out loud, because people underestimate them by a wide margin. Along the grain, from soaking wet to bone dry, a piece changes by roughly a tenth of one per cent — about a millimetre over a metre of length, which is to say about the thickness of a fingernail. Across the grain over the same range it can change by a twentieth of its width or more. In everyday terms: a wide oak tabletop can easily be a centimetre narrower in a heated January than it was in a damp August, while its length does not measurably change at all.

And the two across-directions are not equal either. Movement around the rings — the tangential direction, parallel to the bark — is roughly twice the movement outward along the rays. Nobody is certain of every reason, but the rays are a large part of it: those ribbons of cells run radially, and being oriented across the shrinking direction they act like a great many small straps, restraining the outward movement while doing nothing at all about the movement round the ring.

Two directions of shrinkage, one twice the other, in a material whose rings are curved. Hold that thought for exactly one module, because it is about to explain the single most familiar defect in wood.

Therefore a board cups away from the heart

Here is the derivation, in four steps, and at the end of it you will be able to look at the end of any board and say which way it is going to curl before it does.

One. Wood shrinks about twice as much round the rings as it does outward along them. Two. The rings are curved, because the tree was round — so an ordinary board, a flat-sawn board cut straight through the log like slicing a loaf, is a slice taken across a fan of arcs all centred on where the pith was. Three. The face of that board which stood nearer the bark sits on the longer arc; the face which stood nearer the pith sits on a shorter one. Four. Drying shortens every arc by the same proportion — so the longer, outer arc loses the most actual width. The fan closes a little, and the board's two faces do not shorten by the same amount. (It closes precisely because of step One. If wood shrank equally in both directions the whole fan would simply scale down, every angle unchanged, and nothing would bend at all.)

A board whose top face shortens more than its bottom face has exactly one option. It curls, hollow on the face that shrank the most — the face that stood nearer the bark. (The old shop shorthand for this is that "the rings try to straighten". It is a picture rather than the mechanism, but it points at the same answer, and it is easier to carry in your head at a timber yard.)

Now cut the same log a different way. Stand the board so its width runs outward from the pith, along the rays — a quarter-sawn board. Its rings now cross the end as straight lines running through the thickness. And look what changes:

• It moves half as much in width, because its width is now in the restrained radial direction rather than the restless tangential one.
• It stays flat. Neither face sits on a longer arc than the other, so both shorten by the same amount and there is nothing to cup. It simply becomes slightly thinner.
• It shows the rays. On oak this is the famous silver fleck or figure, which appears on quarter-sawn faces and nowhere else — the ribbons you met in Module 3, cut lengthways at last.
• It costs far more, and now you know exactly why. You cannot slice a round log into radial boards without a great deal of waste and rehandling. You are paying for the geometry.

That is the whole of the flat-sawn versus quarter-sawn argument, and you have just derived it rather than memorised it. Flat-sawn is cheap, wide, boldly figured, and restless. Quarter-sawn is dear, narrow, quietly figured, and steady. Neither is better. They are answers to different questions — and when the answer must not move, you already know which one to buy.

One old rule deserves an honest treatment. When gluing several boards edge to edge into a wide top, many books say alternate them heart-up, heart-down, heart-up, so that the cups oppose each other. It does work, in the sense that you get several small ripples instead of one big dish. But it is a way of managing a problem rather than solving it, and it usually costs you the best face of every second board. The better answers, in order: choose quarter-sawn where flatness truly matters, keep the wood in the room it will live in for a fortnight before you work it, and above all fasten the top so that it is held flat while still being allowed to change width. That last one is Part Five, and it is the whole of joinery.

The four ways a board goes wrong — and how to dry one

Warping is not one thing. It is four, they have separate names, and each has its own cause in the bundle. Once you can name what a board has done, you can usually say why.

• Cup — a curl across the width, the board becoming a shallow gutter. Cause: the tangential and radial directions shrinking by different amounts, as in the last module. Predictable from the end grain.
• Bow — a curve along the length, seen on the face, like a ski. Cause: one face drying faster than the other, or wood of differing character through the thickness.
• Crook (sometimes called spring) — a curve along the length, seen on the edge, so the board is banana-shaped when laid flat. Cause: one edge shrinking more than the other, which happens when the grain runs at a slight angle along the board, or when one edge holds denser wood.
• Twist (also called wind, said as in winding a clock — which is where the cabinetmaker's "winding sticks" get their name) — the four corners no longer lie in one plane, so the board rocks on a flat bench. Cause: nearly always spiral grain, where the tree itself grew with a twist and the tubes wind slowly round the trunk. This is the hardest to fix and the easiest to spot: sight along the board from one end, and if the far end appears tilted relative to the near one, put it back.

And one more, which is not a warp but a failure: checking — splits that open on the surface and at the ends. That one is worth deriving, because it explains how wood must be dried.

Water leaves wood far faster through the open ends of the tubes than through their sides. The difference is not small — something like ten to fifteen times faster along the grain than across it. So the ends of a board dry first and try to shrink first, while the middle is still wet and still full size. The dry ends are gripped by a wet middle that will not let them shrink. Something must give, and since wood pulls apart across the grain at the slightest provocation, the ends split.

Everything in the drying yard follows from that one sentence:

• Seal the ends. Paint, wax, old emulsion, anything that slows the fast exit. You are not keeping water in — you are making the ends dry at the same rate as the middle.
• Stack with stickers — thin strips of dry wood laid across between each layer, in vertical columns so the weight bears through. Air must reach every face equally; a board dried on one side only will bow, for exactly the reason your wetted lolly stick curled.
• Weight the top of the stack and keep it out of sun and wind. Slow is the whole trick. Sun and wind dry the surface long before the core, and a shell of dry wood over a wet core is what tears a board apart from the inside.
• Allow time. The old rule for air drying is roughly one year per inch of thickness — about a year for every twenty-five millimetres — and then a final spell indoors in the room where the piece will live. Kilns exist to buy that time back with careful heat and controlled humidity, not to skip the physics.

With the grain, and against it

Stroke a cat from head to tail and the fur lies down. Stroke it tail to head and the fur stands up, and so does the cat. A plane on a board is doing exactly that, for exactly the same reason.

A plane blade does not slice a shaving off the way a razor slices paper. It wedges under a thin layer and lifts it — and because wood parts along the grain at the slightest excuse, a small split runs ahead of the cutting edge, following the fibres. That split does most of the work. The steel mostly decides where it begins.

Everything then depends on where the fibres are heading:

• If the fibres ahead of the blade rise toward the surface in the direction you are going, the split runs up and out. The shaving breaks free at the surface and the surface left behind is the one you intended. This is with the grain.
• If the fibres ahead of the blade dive down into the wood, the split runs down with them, below the level you meant to cut, and then a lump has to break out to release it. That is tear-out: a pit torn out of your finished surface, leaving you planing below the level you wanted and chasing it. This is against the grain.

So the real skill is not "plane with the grain", which everybody is told and almost nobody is shown how to judge. The skill is reading which way the fibres leave the face you are about to cut. Two ways to do it, and use both:

• Look at the edge of the board. The grain lines there run at a slight angle. Plane in the direction in which those lines are rising toward the face you are cutting — so the shaving climbs out rather than diving under.
• Use your thumbnail. Drag it lightly along the face one way, then the other. One way feels smooth, the other faintly furry, because you are catching fibre ends. Plane the smooth way. This works on painted, dirty and thoroughly confusing boards where the lines tell you nothing.

And then there is wood with no answer at all: figured maple, rippled ash, wood beside a knot, wood from a crotch where two limbs parted. In these the grain reverses every few centimetres, so part of every stroke is against it. That is not a failure of skill, it is a property of the board, and the craft has three honest answers — all of which do the same thing, which is to stop the split ahead of the edge from running:

• Take a finer shaving. A thin shaving cannot lever up a long split. Most tear-out simply goes away when the cut gets lighter.
• Close the mouth of the plane and set the cap iron close behind the edge. The mouth is the slot in the plane's sole that the blade comes up through; the cap iron is a second piece of steel clamped on top of the blade a whisker back from its edge. Narrow the one, bring the other forward, and both do the same thing — hold the wood down immediately in front of the cut and break the shaving over sharply, so the split has nowhere to travel.
• Change the angle of attack. A blade set at a steeper bedding angle — the angle at which the blade lies in the tool — or skewed, or a card scraper (a rectangle of hard steel whose cutting edge is a fine burr turned over along its side) held nearly upright, stops levering and starts scraping. Scrapers and abrasives sever fibres whichever way they point, which is why they can smooth wood no plane will touch, and why they leave a duller surface: fibres severed and slightly torn, rather than sliced.

Sharp severs, dull crushes

Sharpening is treated as a devotional practice, with sects and long arguments. Underneath it is a plain mechanical fact about tubes.

An edge is a wedge. A sharp edge has an apex finer than the wall of a wood cell, so the first thing it meets is a wall, and it parts it. A blunt edge has a rounded apex, thicker than the cell itself. It cannot get between anything — so it presses the tubes flat first, crushing them, and only then drags them apart. What you get is a surface of collapsed, torn, woolly fibres, and the tell-tale of dullness: wood that looks fuzzy rather than burnished, and that goes dark and blotchy when finished, because the crushed fibres drink far more oil than their neighbours.

The straws also tell you which cuts are hard. Cutting along the grain, you are mostly breaking side-glue: easy. Cutting across the grain on a face, you sever tube walls in a line: harder. Cutting end grain — paring or planing the sawn end of a board — is the hardest cut in woodwork, because every tube must be severed square across its open mouth with nothing behind it for support. From which three otherwise arbitrary rules:

• End grain demands your sharpest edge. Anything less crushes the tube mouths, which is why a badly pared shoulder looks grey and compressed.
• Skew the blade. Presenting the edge at an angle slices rather than chops, for the same reason a kitchen knife is drawn through a tomato rather than pressed onto it. Less effort and a better cut, from geometry alone.
• Never run off the far edge. The last few tubes at the far corner have no neighbours holding them, so the blade pushes them off and takes a chunk of your work with them. Cut in from both ends toward the middle, or clamp a scrap block behind the cut so the blow-out happens in the scrap. It is the short-grain problem from Module 5, arriving at a corner.

Glue holds the sides of straws, not their ends

Take two bundles of drinking straws and try to glue them together end to end. The glue runs away down the tubes, there is barely any solid material at the surface for it to grip, and the joint you get comes apart with a twist of the wrist.

Now glue two bundles side by side along their lengths. Every straw lies against a straw over a long area, and the glue has a great deal to hold. That joint is strong.

Wood does precisely this, at a scale you cannot see. A long-grain glue joint — two board edges glued to make a wide panel — is genuinely stronger than the wood itself; break it and it breaks beside the glue line, not on it. An end-grain glue joint is close to worthless, because you are gluing the open mouths of tubes: the adhesive wicks away down the pipes and starves the surface, and what is left is holding almost nothing.

Which explains an entire family of joints at a stroke. Nearly every classic joint is a device for turning an end-grain meeting into a long-grain glue surface, or for adding a mechanical lock that does not depend on glue at all:

• A mortise and tenon — a tongue on one part, a socket in the other — turns the end of a rail into two broad cheeks lying against the long grain of the leg, with shoulders around it — the flat faces the tongue steps down from — to hide small movement and stop the joint racking, which is to say folding out of square like a gate. It is the oldest joint we have: Egyptian furniture in museums is held together with it, and the oak linings of neolithic wells excavated in central Europe, whose timbers are tree-ring dated to more than seven thousand years ago, are jointed the same way.
• A dovetail — wedge-shaped pins and tails — does not trust glue at all in the direction that matters. The wedges physically cannot pull apart one way, which is why drawer fronts, pulled every day of their lives, have been dovetailed for centuries.
• A peg or dowel introduces long-grain surface where there was none, and adds a mechanical lock that survives even if the glue fails. In draw-boring, the hole through the tenon is offset by a shade on purpose, so that driving the peg drags the shoulders tight and keeps them tight. That trick predates reliable glue, and it is why pegged oak frames are still standing.
• A housing (a trench cut across a board), a rebate (a step taken out along an edge) or a groove (a trench running with the grain) each gives a shelf or panel a long-grain cheek to sit against and a shoulder to carry the load.

The one rule of cross-grain construction

Now the two halves of this course meet. Wood is strong along the grain and weak across it. Wood moves across the grain and not along it. Put those together and you have the single rule that separates work which lasts from work which destroys itself.

Picture a tabletop sixty centimetres wide, glued up from boards. Its width will change by several millimetres between a damp summer and a heated winter. Now screw a solid batten across underneath it, tight, in a dozen places. The batten's length lies along its own grain, so it does not change length at all. The top wants to shrink across. The batten will not permit it.

The top is now being stretched across its own grain — the weakest direction the material has. It has three ways out and it will take one: it splits, it pulls the fastenings out, or it crushes itself. Nothing else is on offer. And it happens in the first dry winter, in a heated room, to a piece that was flawless in the workshop.

Worse, the failure can hide. In a damp summer that trapped top tries to expand, cannot, and squashes its own fibres flat — the cells crush and never fully recover. Come winter it shrinks from that new, permanently smaller size, and there is now a gap that never existed before. Old tables show this as open joints which were tight the day they were made. The wood did not shrink away from its maker. He squeezed it first.

So the craft's answer, everywhere, is the same: hold it firmly in one place and let it slide everywhere else.

• Frame and panel. A door or a chest side is built as a frame of narrow parts — uprights and cross-pieces, none of them wide enough to move much — with a wide panel sitting in a groove, not glued, free to shrink and swell inside it. Six centuries of cabinets rest on this idea. Glue the panel in and it splits; let it float and it survives. On old work you can often see the unweathered line at a panel's edge, showing exactly how far it has shrunk since it was made.
• Buttons and slotted fastenings. A tabletop is held to its frame by wooden buttons hooked into a groove, or by screws through slots running across the direction of movement. Tight enough to hold it flat, loose enough to let it breathe.
• Breadboard ends. A strip across the end of a wide panel to hold it flat: glued and pegged at the middle only, with the outer peg holes elongated into slots so the panel can grow outward from the centre in both directions. Done properly, the ends of the strip stand a whisker proud in winter and a whisker shy in summer, and that mismatch is not a defect. It is proof the joint is working.
• Floating drawer bottoms and cabinet backs, sitting in grooves, fastened along one edge at most.
• Expansion gaps at the edges of a floor, hidden beneath the skirting, so that a room full of boards has somewhere to go in August.

The same rule read backwards tells you when you may ignore it. Narrow parts move very little, so a narrow batten glued across a narrow rail is fine. And man-made boards such as plywood are built of thin layers glued with their grain crossing at right angles, so each layer restrains the next and the sheet barely moves in any direction — which is exactly why plywood goes where solid wood would tear itself apart, and why a plywood drawer bottom may safely be glued in all round when a solid one may not.

Building for three hundred years

Take the last two modules seriously and a whole philosophy of building falls out of them, and it is not the modern one. You are not making a rigid object. You are making an arrangement that stays true while every part of it quietly changes size, forever.

That is why traditional timber framing is jointed and pegged rather than glued and bolted: a pegged mortise and tenon can shift a hair with the seasons and come back. It is why a well-fitted drawer is left a touch easy in a humid climate and runs sweetly all year, while one fitted perfectly in February jams solid in August. It is why an old chest with a split top so often turns out to have been improved, at some point, by somebody who screwed it down tight.

The lesson from Hōryū-ji is not that the old ways were magic. It is that people who could not overpower the material learned to arrange it, and arrangement outlasts force. Modern screws and glues are excellent and this course will not tell you to give them up. But they let you overpower a board temporarily — and a board is patient.

Choosing wood — density is mostly air

Balsa and oak are made of very nearly the same substance. That sentence sounds absurd until you remember what wood is.

The material a cell wall is built from — the cellulose and lignin — has almost exactly the same density in every tree on earth. What differs wildly between species is how thick the walls are and how much hollow they enclose. Balsa is a bundle of very wide straws with very thin walls: mostly air. Lignum vitae is a bundle of narrow straws with walls so thick there is hardly any hollow left, and it sinks in water. Same stuff, different plumbing.

From which, without memorising a table: dense wood is stronger, stiffer, harder to dent, harder to cut, blunter on tools, slower to dry, worse at insulating, and — because there is more wall material to hold bound water — it moves more with the seasons, not less. Light wood is the reverse on every count. Neither is superior. A light softwood is the correct choice for a beam that must be lifted by one person, and a dense hardwood is the correct choice for a mallet head.

Which brings us to a word that misleads more beginners than any other. Hardwood and softwood are not measures of hardness. They are a botanical division: softwoods are conifers, the needle-bearers with cones; hardwoods are broadleaved trees with proper flowers. That is the entire distinction. Balsa is a hardwood and you can dent it with a thumbnail. Yew and pitch pine are softwoods and are harder than plenty of hardwoods. If you want hardness, ask about density, or look up a hardness figure — never trust the name.

Two more things worth knowing when you choose a board, both straight from the tubes:

• Some woods have big tubes and some have small ones. Oak, ash and elm lay down a band of large vessels at the start of each spring, so their growth rings begin with a row of visible pores; that is why oak has such bold grain, why it takes stain unevenly, and why the pores must be filled if you want a mirror finish. Maple, cherry, birch and sycamore spread small vessels evenly through the year, so the surface is close and even and takes a fine finish straight from the plane.
• White oak holds water and red oak does not. In white oak the old vessels are plugged from the inside by balloon-like growths called tyloses; in red oak they stay open. This is why barrels, boat planking and outdoor joinery are made from white oak and never red — and you can prove it yourself. Take a short length of red oak, dip one end in soapy water, and blow hard down the other. Bubbles. The straws are open pipes and you have just breathed through a tree. Try it with white oak and nothing happens at all.

Finally, the trick that seems to break every rule in this course: steam bending. If the fibres must run along the curve, and sawing a curve creates short grain, how does a Windsor chair get the bent hoop its back spindles stand in? You wet the wood and heat it — steam at boiling point for roughly an hour per twenty-five millimetres of thickness — and the lignin, the glue between the fibres, softens. The fibres can now slide and shorten against one another on the inside of the bend. What they still cannot do is stretch on the outside: wood has almost no give in tension along the grain, which is why serious bending is done with a steel strap and end stops clamped along the outer face, forcing the whole bend to happen as compression on the inside. Cool it, let it dry in the form, and the lignin sets again in the new shape. Michael Thonet built an industry on this in Vienna from the 1850s, and the bentwood café chair he made by the million is still in production. It is not an exception to the straw model. It is the straw model with its glue briefly warmed.

Therefore, therefore, therefore

If you followed all of that, you were not handed a list of tips. You were handed one fact and made to watch its consequences unfold. Here is the whole chain in one place. Read it as a single sentence with a lot of commas.

Wood is a bundle of straws — long hollow tubes, all running the same way, glued along their sides.

• Therefore it is enormously strong along the tubes and feeble across them — so a chair leg runs one way and never the other, and every load-bearing part is a decision about direction.
• Therefore a part is only as strong as the fibres that run through it, not the outline you sawed — so short grain and run-out are faults you cannot see, and curved parts are riven, bent or laminated rather than cut out.
• Therefore it parts along the tubes for almost nothing and resists being severed across them — so an axe is a fat wedge that levers, a saw is a row of little knives, and a split runs ahead of the tool by itself.
• Therefore water sits in two places — loose in the hollows, and bound inside the walls themselves — so nothing moves until the walls begin to give theirs up, and then everything does.
• Therefore wood swells and shrinks across the grain and barely at all along it — the single cause of cupping, warping, splitting, and every joint that fails in a heated January.
• Therefore, since it moves twice as much round the rings as outward along them, and the rings are curved, a board cups away from the heart — and a quarter-sawn board, whose rings stand through its thickness, moves half as much and stays flat.
• Therefore water escapes fastest out of the open ends of the tubes — so ends check first, ends get sealed, stacks get stickered, and drying is slow on purpose.
• Therefore a blade wedging up a shaving starts a split that follows the fibres — so planing with the grain shears them cleanly and planing against it tears them out, and every adjustment on a plane exists to govern that split.
• Therefore a sharp edge parts cell walls while a dull one crushes them first — so sharpness is not fastidiousness, and end grain, where every tube must be severed at its mouth, demands the keenest edge you own.
• Therefore glue grips the sides of tubes and finds nothing to hold at their open ends — so joinery exists largely to convert end-grain meetings into long-grain surfaces, or to lock them mechanically.
• Therefore, finally, joinery is the craft of arranging straws so that movement cannot destroy the work — hold it in one place, let it slide everywhere else, and it will still be standing in three hundred years.

Now test yourself on something this course never covered. Why does a wooden spoon left standing in the washing-up water end up rough and furry? Why does an old door stick in the damp months and swing free in the dry ones — and why along the closing edge rather than the top? Why is a cricket bat made of willow with its grain running down the blade? Why does a log stored with its bark on rot from the outside, and one stored on the ground rot from beneath? Why do the boards of a boat swell tight when it is put back in the water?

You can answer all of those now, and nobody taught you any of them. That is what an education from first principles buys. Not more answers — better questions, and the ability to derive the answer yourself, standing in front of a board, with nobody around to ask.

The tree spent a century lifting water. The wood has not forgotten how.
Build as though it is still moving, because it is.
— The Guild of the Grain