The Hundred-Year Chair · a chapter

The Keel

On August 6 I sat in 22F, a window seat on the flight from Brisbane to Queenstown, and spent about an hour drawing a hinge. The sketch shows one curved leaf wrapped around a pin, with the wood grain and cut surfaces carefully hatched. Under the ink are traces of earlier attempts in pencil. It should not have taken that long.

I had been reading Walter Isaacson's biography of Leonardo da Vinci. I loved the notebook drawings, especially the fine lines that could make a little mechanism seem worth studying for a long time. I had bought a roughly A5-sized Midori notebook and some 0.2-millimeter pens. Truth be told, I was practicing my Leonardo.

Ink section through a hinge pin and curved leaf, with hatched cut surfaces and wood grain.
The hinge study, August 6. A pin would let the two pieces meet at different angles. The hatching took longer than the idea.

There was a chair problem behind the drawing. I had settled on two triangles, but their corners still had to be made. I wanted someone with a chop saw to be able to cut the wooden members and put them together without reconstructing my side-view drawing on the shop floor. A hinge seemed promising because it could serve at any of the corners. Put a hole in each member and run a pin through; the same connector could accommodate different angles.

The hinge did not settle how to attach the hardware to the wood, locate it accurately, or keep the assembly from wobbling. I moved on to brackets. Soon I was pursuing one bracket that could fit every corner, preferably with one part number. A triangular connector with the chair's three angles looked like a good answer. Turn it one way for one corner, another way for the next.

Over the following week I drew more than forty variants and built at least six prototypes. I glued wooden blocks into corners. I 3D printed brackets in PETG that snapped when I tightened the bolts, then triangles in ABS that did not. One two-piece printed frame was elaborate enough to be an assembly jig, except that I was planning to leave it in the chair.

Three triangular plywood test frames standing on tile, with black, white, and metal-colored connectors at their corners.
Three ways of joining the same triangular frame. The nearest has black printed gussets; the others test different connectors.

My notebook filled with versions: bolted blocks, bronze castings, wedges, thin webs, one triangle nested inside another. On August 15 I starred a rounded universal bracket that could serve all three corner angles. I was getting better at solving the problem I had set myself.

Nine notebook sheets showing alternative corner connectors, rotated so the handwritten annotations read horizontally.
Nine sheets of corner studies. The starred universal bracket is dated August 15. Each sketch tries another way to connect the wooden members at a corner.

For a week I had effectively forgotten something Peter Grove's fifth graders could have told me. Fix the lengths of a triangle's three sides and the angles are fixed too. Join three sticks with pins through their ends: the pins can turn while you assemble them, but once the last pin goes in, the triangle cannot change shape without a stick changing length or a connection giving way. I did not need each bracket to insist on an angle. I needed it to locate a corner.

I knew this. The hinge sketch was an attempt to exploit it. But as I worked on the hardware, I had gone back to treating every corner as a separate angle to be held. The useful dimension was the distance between connections, not necessarily the overall length of a piece of wood. A hole in the right place could locate a pin. The end of a stick could locate against a shoulder cut in another. I had been putting a great deal of effort into the shape of the bracket when I should have been looking at where the parts met.

On the morning of August 18 I was looking again at the chair's long diagonal, with the little corner blocks attached to it. The blocks and the diagonal could be one part.

"uh oh. Another idea," I typed to Claude. "That diagonal not only has all the information, it has all the hard joints."

Reduce the side of the chair to lines and mark the six joints. Four lie on the long diagonal: where it meets the seat, the front leg, the back, and the tie. I had been treating those as four separate problems. They were four places on one board.

Give that line fifty millimeters of width and it becomes a piece of wood. Cut three notches into its edges and angle its upper end to meet the seat. Now the four joints have places to land. Instead of positioning a collection of connectors during assembly, I could cut their positions into the board before assembling anything. It looked like a little rafter with the notches carpenters call birdsmouths.

Three stages at the same scale: a two-triangle chair with six joint dots; a fifty-millimeter gray band along the diagonal with four dots; and the isolated board with three notches and an angled seat bearing.
Four of the six joints lie on one line. A single board can locate them: three at notches and one at its upper end. A rafter with birdsmouths, or, with a change of metaphor, a keel with fish mouths.
Notebook page headed THE LEDGER BOARD, showing the diagonal with corner blocks, a pin-jointed chair diagram, and possible names in the margins.
August 18: the diagonal, the chair reduced to lines and pins, and the proposed combined part. At the bottom I wrote, "THIS ENCODES ENTIRE CHAIR (almost)."

The "almost" was doing useful work. The board would locate the joints in the side frame. I still needed the lengths of the other members, the width of the chair, and a way to fasten everything together. But much of what made the chair difficult to lay out could be concentrated in one part, made twice.

I called it a datum strut, then hunted for a better name. My notebook says LEDGER BOARD, SPINE, AXIS, DATUM. I asked Claude for the name of the marked stick carpenters use to transfer positions without measuring each one again. A story pole. That was close: a board carrying the information needed to place other parts. Except that mine would stay in the chair after the layout was done.

Claude favored story pole. I preferred another word in our exchange: keel. It suggested the long structural member around which I could build the rest. "So, let's call it the keel," I wrote. "And while we're at it, let's standardize names: back, seat, keel, front, tie. I won't forget what these mean."

With the new name, the birdsmouths became fish mouths. It was a keel, after all. In the earlier versions I had added a little wedge of wood behind an acute corner. Now I could cut the corresponding shape out of the keel. The end of the adjoining member would bear against the wood that remained.

Three notches on each keel replaced six separate corner blocks. The back and seat still needed a connection to each other, as did the front and tie. I eventually designed a small aluminum bracket for those joints. I had not abolished brackets, but I no longer needed to solve every junction with one.

Close view of a pale plywood chair with a dark diagonal keel, showing notches where the back, front leg, and tie meet it.
The notches give the adjoining parts a place to sit and a shoulder to bear against. Fasteners hold the assembly together.

A notch could both position a part and carry force through contact with the timber. It still needed a fastener: a contact that carries a push may open under a pull, and the chair would be picked up and dragged as well as sat on. But I could now separate the jobs. Cut the shoulder where I wanted the part to sit, then provide a connection that kept it there.

Each joint position along the keel became a station, measured from the same end of the board. If I misplaced one station and measured the next from it, I would carry the mistake forward. Measuring every station from one starting point kept those errors from accumulating. And because the two keels had the same profile, I could stack two boards and cut them together. When I separated them for the two sides of the chair, the notches would match.

That changed what I was asking of a future builder. The complicated layout could be done once, on the keels. The other parts could be cut and brought to the places waiting for them. A carefully made pair of keels could also be supplied to someone who wanted to do the rest of the woodworking. I had started by trying to make every joint clever. Now I had one wooden part worth making accurately.

Side elevation of a Keel chair against a plain wall, with two triangles sharing the long diagonal.
The long diagonal runs from the rear foot to beneath the seat. The other members meet it at the positions established by its notches.

When I saw the first keel chair on the bench, I thought, "nailed it." I had felt much the same about earlier versions. This one did not stop the revisions either. I changed materials, panel thicknesses, and brackets. The larger test was whether the parts would fit without my intervening to make them fit.

On September 6 I rebuilt the maple chair. Its keels were stained black; the pale plywood made their shape easy to see. I replaced the seat and reduced the thickness of the tie panel. The chair was beginning to look like something I could make again.

Maple Ebony chair from above and in front, with pale plywood panels, black-stained maple keels, and visible furniture-bolt heads.
Maple Ebony: pale plywood panels on maple keels stained black. This is the maple chair, not the subsequent cherry build.

That evening I cut a pair of keels in cherry and made a new chair. Every part was cut to the drawing and then assembled. No hole was reamed to meet another hole. No shoulder was pared back because the drawing and the wood disagreed. Assembly took about five minutes.

Until then I had cut close, offered the part up, found the half millimeter, and taken it out with a block plane. The chair came right because I was standing there to make it come right. Someone else would have the drawings but not my experience of where they needed correcting. With the cherry chair, I finally got through an assembly without reaching for the plane.

I was still the person cutting and assembling the parts. The independent test would have to wait for another builder. But the drawing and the wood had agreed for once, and I had begun to trust the drawings coming back from Claude. Sometimes enough to stop checking them.

Notes

A chapter of The Hundred-Year Chair, drafted and refined in public. The same text lives in the book; the interactive tools and the fully-sourced technical notes live here. home · the book · the technical notes · hundredyearchair.com