The Craft · How it Works
Hub, spokes, felloes, iron tire. How wood, iron, and fire become a rolling thing that can carry a ton across rough country for a hundred years.
Hero photograph: inkknife_2000 (Flickr). CC BY-SA 2.0
Every wooden wagon wheel starts at the centre. The hub is a barrel of hardwood, turned on a heavy lathe, with a bore down the middle for the axle and a ring of mortises around the outside for the spokes. Wheelwrights almost always turn the hub from elm or hickory. Elm is the traditional choice because its grain is interlocked — it refuses to split when you drive a dozen spokes into its sides. Hickory is the American substitute when good elm is hard to come by, and it's what most Texas shops have used for a hundred years.
The hub has two lipped flanges, one at each end, a little wider than the barrel between them. The flanges keep grease and trail dust out of the bore, and they give the hub shoulders to rest against the axle skein. The bore itself is not a plain hole. It's lined with a cast-iron sleeve called the box, sometimes called the hub liner, which is what actually rides on the iron of the axle. Wood and iron don't wear well against each other; iron and iron, lubricated, will run for decades.
Outside the wheel, at the very end of the axle, sits the linchpin— a tapered iron wedge driven through a hole in the axle. The linchpin is the one piece that keeps the whole wheel from walking off the axle on a rough road. If you ever hear someone call a load-bearing person the “linchpin” of an operation, that's the metaphor. Pull it and the wheel comes off.
A heavy freight wheel takes twelve to fourteen spokes; a lighter buggy wheel can run on eight or ten. The traditional spoke wood is white oak or hickory, always rived— split along the grain rather than sawn across it — so the long fibres run uninterrupted from hub to rim. A spoke that's been crosscut anywhere along its length has a weak point built into it. A rived spoke does not.
Each spoke is shaped with a drawknife and a spoke shave, taking the rived blank from a rough square to a tapered, oval cross-section. The inboard end is cut into a tongue tenon— a flat, rectangular tongue that drops into the matching mortise in the hub. The outboard end is rounded into a smaller round tenon that fits a bored hole on the inside of the felloe. Two different joints on the same stick, and both have to be cut square the first time.
When the spokes are driven into the hub, they are driven hard, with a heavy mallet, until the tongue bottoms out. At this stage the wheel looks like a sunburst — a hub at the centre, a ring of spokes pointing out, and nothing on the rim yet. A good wheelwright will angle the mortises slightly outward so the spokes splay just a few degrees from the perpendicular. That slight splay is the beginning of the dish, which we'll get to in a minute.
The outer wooden ring of a wagon wheel is built from felloes — pronounced fellies, and sometimes spelled that way. A typical heavy wheel uses six or seven felloes, each a shallow curved segment of the circle, joined end-to-end to make the rim. Two spokes land in each felloe, so a twelve-spoke wheel has six felloes, and a fourteen-spoke wheel has seven.
Felloes are cut from white oak — the same wood as the spokes — either bandsawn from a flitch with the grain following the curve, or steam-bent from straight stock for a stronger rim. The straight-cut method is faster and what most American shops used; the steam-bent method is what English and Australian wheelwrights preferred. Either way, the wood grain has to follow the arc, or the felloe will split the first time it takes a hard hit.
Each felloe is joined to the next by a pair of wooden dowels — called dowel pins — that bridge the seam. No glue, no metal. The dowels locate the joint and keep the segments aligned while the iron tire is fitted; once the tire is on, the entire rim is locked under compression. The inner edge of each felloe is bored to receive the round outboard tenons of its two spokes. That bore has to be deep enough to seat the spoke but not so deep that it weakens the rim — a tolerance of a sixteenth of an inch is normal.
The iron tireis the masterstroke. It's a single continuous band of wrought iron — or, in the modern shop, mild steel — that wraps the wooden rim and holds the whole wheel together under compression. There are no bolts. The iron is forged into a closed ring that is deliberately a touch smaller than the wooden rim it will cover. Roughly an eighth to a quarter of an inch smaller in circumference, depending on the wheel.
To measure how long the iron needs to be, a wheelwright uses a small rolling tool called a traveler— sometimes a travelling wheel. It's a hand-held disc with a counter and a marker on the rim. You roll the traveler once around the outside of the wooden rim, count the revolutions, and that's the length of iron you cut. Then you roll the same traveler around the inside of the forged tire to confirm that the iron is shorter by the right amount. It is the only tool that can measure a curve directly, and every shop has one.
The tire is then heated — either in a long forge fire or laid flat in a ring of brushwoodbuilt around it on the shop floor — until it glows a dull cherry red. At that temperature the iron has expanded enough to drop down over the wooden rim. The wheel is set flat on a tire platform, the glowing ring is dropped on with tongs, and the moment the iron seats, the whole assembly is doused with water — lifted into the slack tub, or quenched by buckets poured around the rim. The steam goes up. The iron contracts. Every felloe seam is pulled tight against every other, every spoke tenon is locked into its mortise, and the wheel that was loose ten seconds ago is now a single structural object.
If the wheelwright cut the iron right, the wheel rings when you tap it. That is the ringing test: a true wheel, struck on the rim with a small hammer, sounds like a bell. A dead thud means a loose joint somewhere — usually a spoke that didn't seat — and the wheel goes back on the bench. The test is more than two centuries old. Read more in the Wikipedia entry on wheelwrights.
Look at a wagon wheel head-on, from the outside of the wagon, and you'll notice it isn't flat. The spokes splay outward; the rim is set further from the hub than the inboard ends of the spokes. The wheel is, in other words, slightly cone-shaped. That intentional cone is called the dish, and it is one of the smartest pieces of geometry in the whole trade.
There are three reasons for it. First, country roads have always been crowned— higher in the middle than at the shoulders — so water sheds off. A wagon rolling down a crowned road is always tilted slightly outward, and a dished wheel sits perpendicular to the ground at exactly that tilt. Second, every wagon swings side-to-side as it rolls; the dish lets each spoke take that side-load in pure compression instead of bending. A straight spoke under a side blow snaps; a dished spoke just pushes harder into the hub. Third, the dish gives wheels enough clearance to mount onto an axle that flares outward at the ends, which most working axles do.
The dish is set in three places at once: in the slight outward angle of the spoke mortises in the hub, in the round tenons at the outboard ends of the spokes, and in the diameter of the iron tire when it's shrunk on. A wheelwright who gets one of those three out of register ends up with a wheel that wobbles or, worse, dishes in the wrong direction — concave inward, which is called negative dish and will fail under load. The Wikipedia article on wheel dishing walks through the physics if you want the long version.
Every wheelwright works from the same handful of steps. Tools and wood vary by shop and by century; the sequence does not.
A seasoned billet of elm or hickory is mounted on a heavy wood-turning lathe and shaped into the familiar barrel with lipped flanges at each end. The axle bore is cut on a true centreline. Get the centreline wrong here and nothing downstream is fixable.
Twelve to fourteen mortises are laid out around the hub's barrel, evenly spaced, and cut by hand with a heavy mortise chisel. Each mortise is angled outward by a few degrees — that's where the dish is born.
White oak or hickory is split along the grain into spoke blanks. Each blank is worked down with a drawknife and spoke shave; the inboard end is cut into a tongue tenon, the outboard end into a round tenon with a hollow auger.
Each spoke is driven into its hub mortise with a heavy mallet until the tongue bottoms out. Then the felloes are sawn or steam-bent to the correct radius, bored to receive the spoke tenons, and drilled at each end for the dowel pins that bridge their seams.
The traveling wheel is rolled once around the wooden rim. That length, minus a fraction, is the iron the smith cuts. The strap is bent into a ring, the ends are forge-welded into a closed hoop, and the tire is checked once more on the traveler.
The tire is brought to a dull cherry red in a ring of fire, lifted with tongs onto the wooden wheel, and quenched in the slack tub. The wheel is then sleeved with its cast-iron box, ring-tested with a hammer, and matched against the wheel built next to it on the bench.
The wagon wheel as we've been describing it — hub, spokes, felloes, iron tire, dish — is the end of a very long line of refinements. The oldest wheels in the archaeological record are nothing like it. They're slabs. The Ljubljana Marshes wheel, dug out of a Slovenian bog and dated to roughly 3130 BC, is a solid oak disc about 28 inches across, joined from two halves with ash dowels, with a square hole through the middle for an axle. That same form — the solid-disc wheel — turns up on Sumerian war carts in the Royal Standard of Ur from around 2500 BC, and on terracotta models from the Indus Valley from around the same period. Bronze Age Crete and the early Cycladic islands used the same disc on ox-carts well into the 2nd millennium BC. The wheel was already old when writing was new.
The spoked wheel arrives later and changes everything. The earliest confirmed spoked wheels come off Andronovo-culture chariots near the southern Urals and date to roughly 2000 BC. Within four centuries the spoked wheel had spread to Egypt, Anatolia, and the Mycenaean world; Tutankhamun's tomb (c. 1323 BC) was buried with six chariots whose wheels have four or six bent-wood spokes and rawhide-bound felloes. The reason the spoke caught on is weight. A solid oak disc heavy enough to take a war chariot weighs forty pounds and resists every turn. A spoked wheel of the same diameter weighs eight or ten and accelerates like a different machine.
Dishing— the deliberate cone built into the wheel — is much later, and as far as the surviving evidence shows, it's a European medieval innovation. Roman cart wheels were generally flat. The dished wheel appears in English and northern-European wainwright work from roughly the 13th and 14th centuries, exactly when crowned roads and heavier farm carts came into common use. The two arrive together because they need each other: a crowned road wants a dished wheel, and a dished wheel needs a crowned road to sit level on.
The iron tireis older than the dish but younger than the spoke. Celtic wheelwrights in what's now Austria and southern Germany were shrinking forged iron hoops onto wooden wheels by the 1st century BC— examples have been pulled from La Tène-era graves at Hochdorf and elsewhere — and the Romans adopted the technique on their heavier baggage wagons. Then, oddly, the method nearly vanishes for a thousand years. Most medieval European wheels used strakes — short curved iron plates nailed to the rim, six or eight per wheel — rather than a single continuous hoop. The single shrunk-on tire comes back into common shop practice in England in the late 17th centuryand spreads through the colonies in the 18th. By the time George Sturt was writing about his grandfather's shop in Farnham, Surrey, in the early 1900s, the technique was so settled it read as ancient. It is, and it isn't.
The shrink-fit isn't a trick. It's thermodynamics applied honestly. Wrought iron has a linear thermal expansion coefficient of roughly 12 × 10-6 per degree Celsius— mild steel is essentially the same, about 11 to 13 × 10-6. Multiply by a hundred degrees and you get a length change of around 0.12%. Multiply by the working temperature of a cherry-red tire — roughly 700 to 800°C above shop ambient — and the iron is about 0.9 to 1.0%longer than it was on the bench. On a 48-inch (152.4 cm circumference, 121.9 cm diameter) wheel, that's a little over a centimetre of growth, or close to half an inch. That extra half inch is what lets the glowing tire drop down over a wooden rim it was forged to be too small for.
Working backwards from the wheel, the wheelwright cuts the iron strap so that, after forge-welding into a closed hoop, its cold inside circumference is between an eighth and a quarter of an inch shorterthan the wooden rim it has to grip. On a 48-inch wheel that's an undersize of about 0.2 to 0.5% — well inside the expansion you get from a single forge-fire heat. The wheelwright doesn't run the calculation in degrees and coefficients on the shop floor. The number is in the traveler-count: rolls of the traveler around the wood, minus a known fraction of a roll, depending on the diameter and the wood's seasoning. The math is the same; the wrappers are different.
Cooling is its own piece of physics. Once the glowing hoop is seated, the wheel goes into the slack tub — the wide, shallow water bath that sits next to every forge — or the rim is doused with buckets and wet sacking as the shop hands run a circle around it. The point isn't just to cool the iron. The point is to cool it fast and evenly. A tire that cools slowly hardens unevenly across its width and can warp; a tire that cools on one side before the other can pull the wheel into a slight oval. The slack tub gives a near-uniform quench in seconds. The steam coming off the rim isn't showmanship. It's the shop's thermometer.
Miss the calculation by even a sixteenth of an inch and the wheel will tell on you. Cut the iron a sixteenth too long and the tire will seat without pulling the joints up — the wheel sits together but isn't under compression. Strike it with the hammer and instead of ringing it gives a dead, short note. The trade calls that a wheel that rings false. The fix is to cut the tire off, shorten it, weld it again, and redo the heat. Cut the iron a sixteenth too short and you get the opposite problem: the tire won't seat all the way down even at full heat, or it seats and then over-tightens as it cools and splits a felloe along the grain. A felloe that cracks under shrink-fit is a felloe you replace before anything else moves — you cannot pull a cracked rim segment back into compression by sanding it.
The first thing that wears on a working wagon wheel is the iron tire. A freight wagon hauling six thousand pounds across rocky Texas country grinds about a sixteenth of an inch off its outer rim every two or three thousand miles, depending on the road. When the iron is down to half its original thickness, or when it starts to develop hairline cracks at the forge-weld, the wheel goes back to a shop to be re-tired. Re-tiring is the same heat-shrink operation as the original fitting, only the wooden wheel is already broken in. The old tire is cut off with a chisel, a new strap is forged, traveler-measured, welded into a hoop, heated, and dropped on. A well-built wooden wheel will outlast three or four iron tires.
The other thing the trade watches for is the loose tire, and this is where climate matters more than most modern people realise. White oak shrinks across its grain by about 4 to 8% as it dries from green to bone-dry, and even a fully seasoned wheel keeps moving a fraction of a percent between humid summer and dry winter. In a dry Texas July, wooden felloes can lose enough diameter that a tire which was tight in May rattles by August. The fix is old, and you can still find it in barn yards: park the wagon in a shallow pond or creek for a night or two, let the wood swell back to its fitted size, pull it out, and roll on. If the tire never tightens up — if the wheel has dried past the point of useful swell — the only honest repair is to pull the tire, cut a quarter-inch out of it at the weld, reforge, and shrink-fit again to the new, smaller rim.
Hub grease is the second leg of maintenance. Historically the lubricant was a paste of linseed oil and graphite— the graphite was sold in the 19th century as black lead — sometimes thickened with tallow or beeswax for hot weather. The grease went between the cast-iron axle box and the iron axle skein, packed in fresh every few hundred miles. A wagon driver who could hear his own wheels was a driver who hadn't greased recently. By the late 1800s, petroleum axle grease in tin pails — the famous Mica Axle Grease made by Standard Oil, with flakes of muscovite mica suspended in the petroleum — had replaced linseed and graphite on most working farms. The mica did the same job the graphite had: a soft, plate-like solid lubricant that smeared into the bearing surface and stayed there.
Felloes are replaced one at a time. If a single rim segment cracks — a hard rock strike, a frost split, a hidden grain defect that finally let go — the wheelwright pulls the tire, knocks the cracked felloe loose at its dowel joints, shapes a new oak segment to exact match, dowels it in, and shrinks the tire back on. Two adjacent felloes is borderline. Three and the whole rim is suspect; at that point the trade wisdom is to rebuild the wheel from the hub out. Spokes are replaced even less often, because a spoke failure usually means the hub has gone too. An old wheelwright's saying: a felloe is a repair, a spoke is a rebuild, a hub is a new wheel.
The economics are why the craft survived as long as it did. A pair of working wheels in 1880 cost a farmer roughly two weeks' wages. A re-tire was an afternoon's work and a few dollars. A new felloe was less. As long as wheels could be maintained for a fraction of replacement cost, the wagon stayed on the road; the moment maintenance stopped being cheaper than buying a Ford, the wagon went into the barn. A wheel that gets re-tired every five thousand miles, greased every few hundred, and ponded once a summer will outlast its builder. There are working examples in living-history museums today that are more than 120 years old and still roll true.
A wagon wheel was almost never sold by itself. They came in pairs — built on the same bench, on the same day, from the same flitch of oak, by the same hand. The reason is wear. Two wheels on the same axle with even slightly different diameters will track the wagon to one side and load the axle unevenly. A pair built together, dished together, and ringed together stays true together. Old livery stables would mark each pair with the same stamp; you can still see those stamps in museum wheels.
The traveleris worth a paragraph of its own. It's an iron disc, four to eight inches across, with a counter and a starting mark, mounted on a wooden handle. The wheelwright rolls it around any curve and counts revolutions to measure the length of that curve. No tape measure can do that without slipping. You can find the Library of Congress's collection of wheelwright photography for examples of the tool in use — the LOC wheelwright archive has dozens.
And the linchpin, again, because it's the single most important small piece of iron on a wagon: a tapered wedge dropped through the end of the axle, just outboard of the hub, that physically prevents the wheel from walking off. On a heavily loaded freight wagon, the linchpin is checked at every stop. Lose it on a downhill run and the wheel goes one way and the wagon goes the other. The phrase is more accurate than most people who use it ever realise.
You can buy a wagon wheel today — bandsawn, glued, bolted, factory-finished, hanging on the wall of a steakhouse. It will not roll a thousand miles. The wheel described above will. The difference between the two is not the materials. It's the method.
Every step in the sequence above does a job that no other step can do. The dish handles side-load on a crowned road. The heat-shrink locks every joint in compression so nothing vibrates loose. The traveler measures the curve no tape can measure. The ring test tells you whether the wheel is true before you ever load it. The linchpin, small as it is, keeps the wheel on the axle through ten thousand miles of bad country. Skip any of these and you've built a wall ornament.
It's also the reason a real wheel is slow to build. A pair of working wheels — from raw oak to ringed and paired — takes about a month of dedicated bench time, spread across the seasoning, the turning, the mortising, the forge work, and the fitting. That's a long time. It is also the price of a wheel that lasts a lifetime. Read more about the wider trade in The Lost Art of Wagon Making and the family of wagons it produced in our history of the covered wagon.
At Healin' Wheels, the workshop in Midway, Texas is set up to build wheels exactly this way — on a hub lathe, around a forge, with a traveler hanging from the wall. The first wagons are not built yet. The bench is, the tools are, and the method is the one above. When we adapt these wagons for wheelchair-accessible riders, the wheels do not change. The bed does. The wheels are the same wheels they have always been.
The literature on wheelwrighting is small, mostly old, and mostly still in print or scanned. These four are the ones we point people to when they want primary material instead of summaries.
Talk to the Shop
Kelly takes a small number of calls a week from folks who want to talk wheels — restorers, history teachers, ranchers, and the occasional curious neighbour. The shop is in Midway. The phone is below.
