Guédelon looks like a medieval castle that has somehow survived unfinished for eight centuries.
It has round stone towers, arrow loops, curtain walls, vaulted rooms, timber roofs, a fortified gateway and wooden scaffolding wrapped around sections that are still rising. Men and women quarry stone, shape timber, forge iron and mix mortar around it.
But there is no ancient ruin underneath any of it.
In 1997, there was no castle here at all.
Guédelon is being built from scratch in the twenty-first century as if its builders were working in the thirteenth. The aim is not to create a film set or a modern imitation of a medieval fortress. It is an enormous experiment: if we use the materials, tools, geometry and working methods available to medieval builders, what actually happens when we try to build a real castle?
That question has kept the construction site going for almost thirty years.
And its story begins only about ten minutes away, at the Château de Saint-Fargeau.
In 1994, Michel Guyot, one of the owners responsible for restoring Saint-Fargeau, was studying the enormous château with archaeologists. Investigations beneath and within the later building revealed traces of a much older medieval castle buried inside the history of the present structure.
Saint-Fargeau had been rebuilt so many times that understanding exactly how its earlier phases had been constructed was difficult.
From that problem came an extraordinary idea.
Instead of examining another ruined castle and trying to work backwards, why not reverse the process?
Why not build one?
If archaeologists could watch an entire medieval-style construction project from the first foundation trench to the final roof timber, they could test explanations that were almost impossible to prove from ruins alone.
Michel Guyot developed the idea with Maryline Martin, who became Guédelon’s co-founder and later its president. A scientific committee, architects, historians, archaeologists and craftspeople became involved.
Then they needed somewhere to build it.
The site they eventually chose in the forest near Treigny was almost ideal.
It contained an old quarry of iron-rich sandstone. There was extensive woodland for structural timber and firewood. The ground contained clay. Water was available. Sand, earth and natural pigments could be found in the surrounding landscape.
This is fundamental to Guédelon.
A medieval castle was not simply a building placed on a site after thousands of tonnes of material had been delivered by lorry from industrial suppliers.
Transport was difficult and expensive.
Whenever possible, builders used what was available close to the construction site.
Stone had to come from somewhere. Trees had to be felled. Lime had to be produced. Iron needed to be worked. Clay became tiles. Sand and lime became mortar. Rope, baskets, carts and scaffolding all required their own materials and labour.
At Guédelon, the landscape therefore became part of the experiment.
The first stone was laid on 20 June 1997.
The construction site opened to visitors the following year.
But simply deciding to build “a medieval castle” would not have been precise enough. A castle built in 1050 would look and function differently from one built in 1250 or 1450. Architecture changed, weapons changed and the political organisation of France changed.
Guédelon therefore needed a date.
The builders effectively placed themselves in 1229.
They also invented a lord.
His name is Guilbert Courtenay.
He is not a real historical person. He is a carefully constructed fictional landowner whose social position allows historians to decide what sort of castle he could realistically have afforded.
According to the scenario, Guilbert was born in 1199 and is a relatively modest lord of Puisaye, subordinate to the more powerful Jean de Toucy.
This fictional biography is much more important than it may sound.
Without it, nothing would prevent the designers from selecting the most impressive medieval feature they could find from a dozen different castles and combining them into an unrealistic fantasy fortress.
Guilbert does not have unlimited money.
He cannot build a royal Louvre.
He requires a defensible residence appropriate to a minor lord of the first half of the thirteenth century.
The result is therefore a château-résidence: a fortified lordly home rather than an immense royal fortress.
Its architecture belongs broadly to the system developed during the reign of Philip II Augustus, King of France from 1180 to 1223.
Philip Augustus transformed castle building across the expanding Capetian kingdom. Fortifications became increasingly regular and standardised.
A typical “Philippian” castle used high masonry curtain walls arranged around a polygonal enclosure. Round towers projected from the corners so defenders could fire along the faces of the walls. Arrow loops were arranged at different levels. One tower was generally larger and stronger than the others, while the principal entrance could be defended by two round towers.
Guédelon follows those principles.
That also explains why its towers are round.
Earlier castles frequently used square towers. The corners of square towers created weak points and blind areas. A cylindrical tower distributed impacts more effectively and gave defenders a wider field of fire.
The castle therefore represents a very specific period in the development of European military architecture.
But Guédelon is not an exact copy of any surviving castle.
There is nothing to copy.
Every architectural decision has to be justified by studying real thirteenth-century buildings, archaeology, written evidence, iconography and the results of previous experiments on the site.
Sometimes the experiment proves an assumption wrong.
Then they change it.
That is perhaps the biggest difference between Guédelon and a normal construction project.
Normally, rebuilding something because research has changed would be considered a failure.
Here it can be the result.
The purpose is not simply to finish the castle as quickly as possible. The purpose is to understand how it could have been built.
The quarry is one of the clearest examples.
Large blocks of sandstone are extracted from the ground close to the castle. Quarrymen examine the natural fractures in the rock and use wedges and hand tools to detach useful pieces.
The stone then passes to the masons and stonemasons.
Not every stone requires the same amount of work.
Large quantities of irregular stone can be used within the thickness of walls, while carefully cut blocks are required for corners, doorways, windows, vault ribs, arrow loops and other structural details.
A stonemason therefore does not simply smash rock until it becomes rectangular.
He has to understand where the finished stone will sit in the building.
Templates and geometric drawings determine its shape. Mallets and chisels remove material progressively until the stone fits the position intended for it.
Measurements are established using simple geometric methods rather than digital surveying equipment.
Ropes, compasses, straightedges, squares and proportional geometry can produce remarkably sophisticated results.
One of the symbols frequently associated with medieval geometry is the knotted measuring rope. Whatever the exact historical use of particular knot arrangements, ropes and basic geometric constructions allow builders to produce right angles, arcs, circles and proportional divisions without modern instruments.
The mortar holding much of the castle together is also fundamentally different from modern Portland cement.
Traditional masonry uses lime mortar.
Limestone is heated to produce quicklime. Water is then added to slake it, and the resulting lime can be combined with sand to make mortar.
Producing the lime is an experiment in itself.
Guédelon operates a lime kiln developed in collaboration with archaeologists. A firing can continue for approximately 72 hours without interruption.
That means somebody has to maintain the fire.
The process cannot simply be programmed into a computer and left overnight.
The walls themselves reveal another misconception about medieval castles.
They are not usually made from perfectly shaped blocks stacked through their entire thickness.
The visible faces can contain carefully selected or dressed masonry, while the enormous core between them can be packed with rubble and mortar.
A thick wall therefore consumes astonishing quantities of relatively ordinary stone.
Then the builders have to lift it.
There is no tower crane.
Instead, Guédelon uses lifting devices based on medieval evidence, including the spectacular treadwheel crane, sometimes called a squirrel cage.
A person walks inside a huge wooden wheel.
The rotating wheel winds a rope around an axle, producing enough mechanical advantage to raise heavy loads.
The principle is simple, but seeing it work changes the way a medieval cathedral or castle looks.
Every stone high in a tower had to be moved against gravity somehow.
Mechanical engineering did not begin with steam engines.
Medieval builders were extremely capable of using levers, pulleys, winches, treadwheels and carefully designed scaffolding to multiply human strength.
The scaffolding itself is timber.
Carpenters work with oak and other local wood, often beginning with entire tree trunks rather than arriving at the site with perfectly standardised industrial beams.
Axes and adzes are used to square the timber.
This leaves tool marks that archaeologists can compare with marks surviving on medieval roof structures.
That is one of Guédelon’s most useful characteristics.
The experiment creates archaeological evidence.
A researcher can observe exactly what mark a particular tool produces, how timber behaves after being worked in a particular direction or how much labour is required to make a certain joint.
Carpenters then assemble roofs, floors, doors, bridges, hoardings and the temporary wooden structures required by the masons.
Many joints rely on carefully cut connections and wooden pegs rather than modern metal brackets.
Where iron is necessary, it has to be made by the blacksmiths.
The forge produces nails, hinges, straps, tools and fittings.
Even a door can become a major project.
The two huge leaves of the fortified gateway require timber shaped by carpenters and ironwork produced individually by the forge.
In 2026 another seemingly minor part of the castle illustrates the amount of labour hidden inside medieval architecture.
The team is installing wooden panelling beneath the roof structure of the lord’s chamber using thirteenth-century methods.
The work requires several different crafts, including carpenters, joiners and workers splitting timber.
The blacksmiths alone are expected to produce around 2,000 nails for this panelling.
Two thousand individually forged nails for part of one room.
It is an excellent measure of how misleading the finished castle can be.
When we see an old building today, the objects inside it appear static.
At Guédelon, every component exposes the chain of labour required to make it.
A nail begins as iron.
A roof begins as trees.
A tile begins as clay.
A wall begins in a quarry.
A painted chapel begins with minerals and plants that have to become pigments.
The site therefore includes far more professions than masons.
There are quarrymen, stonemasons, carpenters, blacksmiths, woodcutters, tile makers, carters and rope makers, but also basket makers, painters, dyers, gardeners and specialists working with plants.
Large woven baskets can be used for transporting material.
Wool is dyed using plants.
Natural pigments are prepared for wall paintings.
Clay is shaped and fired into tiles.
The castle therefore gradually generates an entire small economy around itself.
That is probably closer to the reality of a medieval building site than imagining a group of masons working alone.
The workers need tools.
The tools require blacksmiths.
The blacksmith needs charcoal.
Charcoal requires wood.
Stone needs transport.
Transport requires carts, ropes and baskets.
Workers need food.
Animals and people require access routes, water and organised storage.
A major building site affects everything around it.
Guédelon also extends beyond the castle.
A few hundred metres through the forest stands a working medieval watermill.
This was created in partnership with Inrap, the French National Institute for Preventive Archaeological Research.
Archaeologists had excavated the remains of a twelfth-century watermill at Thervay in the Jura.
But archaeological remains can tell only part of the story.
They reveal surviving pieces, dimensions and traces of mechanisms. They do not automatically explain exactly how efficiently the complete machine worked.
So Guédelon built one.
The reconstructed mill actually grinds grain and produces flour when there is sufficient water.
It is experimental archaeology in its purest form: archaeology supplied the evidence, craftspeople produced a working hypothesis, and operating the machine generates new information.
The castle has worked in exactly the same way.
In 2002, Guédelon completed its first ribbed vault.
In 2003, another type of vault followed in the basement of the great tower.
The lord’s residential range began rising in 2004.
By 2008 its first major roof structures were being raised.
Fireplaces became operational in 2009.
The residential roof structure was completed in 2010.
The largest vault was brought under load in 2011.
Wall painting began in 2012.
Construction of the chapel became a major focus from 2014, with its vault completed the following year.
A second wall walk was nearing completion in 2016.
One of the towers received its roof structure in 2018, and the Chapel Tower roof followed in 2019.
From 2020, the monumental entrance between two round towers became one of the principal areas of work.
In 2022 another tower received its roof structure.
In 2023 one tower was covered with wooden shingles.
The building therefore changes visibly between visits separated by several years.
This is not a reconstruction in which scaffolding has been left in place to create atmosphere.
The scaffolding moves because the castle really is moving upwards.
By 2026, masons are still raising the fortified gateway between its two towers. The room intended to contain the portcullis mechanism has risen above the entrance, and when conditions allow, visitors can enter this new area.
Work also continues on the eastern curtain wall facing the gardens, including its parapet and nine battlements.
At the same time, the panelling project inside the lord’s chamber is testing another largely forgotten aspect of medieval carpentry.
So after almost thirty years, Guédelon remains unfinished.
That was originally not expected to take quite so long.
But speed gradually became less important than research.
New archaeological studies produce new information. Experiments reveal better interpretations. Sometimes something that appeared convincing on paper proves difficult or implausible once craftspeople actually attempt to construct it.
That is precisely why the project exists.
A drawing can show that a medieval crane should work.
A full-size crane tells you how many people it needs, how quickly it lifts a stone, where the rope wears, how the timber flexes and how workers communicate while using it.
A historian can calculate how much stone a wall contains.
A working quarry reveals how long extracting that stone actually takes.
A surviving medieval roof can be measured.
Felling, squaring and assembling comparable timbers reveals what its construction demanded from real people.
Guédelon therefore sits somewhere between a castle, laboratory, construction site, open-air museum and school.
Calling it simply a “replica castle” misses almost everything interesting about it.
There is no lost Château de Guédelon being recreated.
The fictional lord Guilbert Courtenay never lived here.
Instead, the fictional history gives the experiment rules.
If it is 1229, a building method documented only in the fifteenth century cannot simply be added because it is convenient.
If Guilbert is a modest lord, he cannot suddenly afford the architectural programme of a king.
If a particular material would have required implausible transport, the builders need to question whether another solution would have been more realistic.
Those restrictions are what turn the project into historical research rather than medieval-themed entertainment.
The connection with Saint-Fargeau is therefore much deeper than simple geography.
At Saint-Fargeau we see the end result of nearly a thousand years of alteration.
A medieval fortress became a later château. Towers were reused. Interiors were rebuilt. New architectural fashions covered older structures. Archaeologists then had to peel those layers apart to understand how the building had evolved.
Guédelon reverses the process.
Here we can watch those layers being created.
A quarry becomes a block.
The block becomes a wall.
A tree becomes a beam.
A beam becomes a roof.
Iron becomes a hinge.
Clay becomes a tile.
And slowly, over decades, an empty patch of forest becomes a castle.
In 1997 Guédelon consisted largely of quarry, trees, soil and an idea.
By 2026 there are vaulted rooms, painted walls, working fireplaces, roofed towers, a lord’s residence, defensive curtain walls, a chapel, a fortified gateway and an entire network of workshops surrounding them.
Yet perhaps the most important result is not the building.
It is everything the builders have learnt while making it.
That was the idea born from Saint-Fargeau in the 1990s: if ruins cannot answer every question about medieval construction, build a new castle and ask the questions while the stones are still being laid.
Nearly thirty years later, they are still doing exactly that.