A castle under construction can be more revealing than a finished castle. Before plaster hides a wall or a roof covers a joint, choices remain visible. At Guédelon in Burgundy, France, present-day craftspeople use a building project to investigate medieval construction. The castle is a result, but the questions encountered while making it are also part of the work.
A modern project with a historical frame
Guédelon began building in 1997. Its official account describes an experimental archaeology project concerned with the methods and organization of a construction site in the first third of the thirteenth century. This is a contemporary undertaking informed by historical evidence. The people at work are not medieval builders accidentally preserved in a French forest.
The project draws on surviving castles, building accounts and visual sources. Inspecting a wall can reveal dimensions and traces of workmanship, while a manuscript image may suggest a tool or operation. Neither necessarily gives a complete set of instructions. Guédelon creates an opportunity to test what a proposed technique demands when someone actually attempts it.
That difference is the heart of experimental archaeology. An idea that looks persuasive on a drawing may require an awkward sequence, an unavailable tool or far more material than expected. A successful experiment can show that something is possible under particular conditions. It cannot, by success alone, prove that every medieval builder used the same method.
Look for the dependency between trades
A stone wall encourages attention to the mason. But shaping stone depends on tools, moving it depends on transport, and placing it at height depends on lifting arrangements and temporary works. A building site makes these dependencies harder to overlook. One craft’s finished object becomes another craft’s starting condition.
The official construction chronology gives a concrete example in its account of the great doors completed during the 2025 season. Carpenters worked on the timber, stonemasons made pivot stones, and blacksmiths supplied metal fittings and nails. An impressive door was therefore also a negotiated meeting between materials. Its weight and movement had to be accommodated by people whose work would finally appear as one object.
For a visitor, an ordinary question such as “Where did that nail come from?” can be more productive than asking only how tall the wall has become. Follow the answer backwards. What tool made it? Who maintained that tool? What fuel and material were needed? A castle is an accumulation of these smaller chains, many of which disappear behind its final appearance.
An excavated mill becomes a testable proposal
The French archaeological institute Inrap describes its collaboration on Guédelon’s watermill in a 2014 press release. The reconstruction drew on remains of two watermills excavated at Thervay in the Jura in 2008, one Carolingian and another dating between the mid-eleventh and late-twelfth centuries. Archaeological remains supplied the starting evidence; a working reconstruction allowed further questions to be explored.
A watermill is especially useful for this kind of investigation because its parts must cooperate. A plausible-looking arrangement is not enough. Water must drive the mechanism, movement must reach the stones, and the grain must be processed. Reconstruction can expose gaps between an interpretation of remains and the requirements of operation. It can then suggest new questions to bring back to the excavation record.
The mill also prevents the castle from becoming an isolated object. Buildings require food, materials, energy and surrounding work. Considering a mill alongside walls expands the picture from monumental architecture to a functioning environment. It makes the less conspicuous infrastructures of daily life part of the inquiry.
What is being asked now?
As checked in October 2026, Guédelon’s own chronology describes the work as continuing. The 2026 season includes experiments with timber panelling for a vaulted ceiling in the aula, or great hall. Its questions concern the wood, the dimensions and splitting of boards, their attachment and their possible finish. These are current research questions, not a claim that all answers have already been recovered.
No promised completion date is needed to understand the project. A target date can become stale, and a finished exterior does not mean that every historical question has been settled. The chronology is more informative because it records which tasks and uncertainties occupy a particular season.
The experiment needs a record
The PROMESS research project at CNRS/LIRIS used Guédelon as a setting for work on multisensory recording and presentation. Its stated interests include both the physical environment and craft gestures. This acknowledges a difficult problem: a completed timber joint does not preserve every movement, sound or judgment involved in making it.
Recording cannot capture everything either. A camera angle can hide a tool edge, and a successful result can omit the failed attempt that clarified the problem. The useful record connects evidence, decisions, conditions and outcomes closely enough that another person can understand what was tested. It makes the experiment available for criticism as well as admiration.
A useful way to browse the site from home
Open the official chronology and compare two neighbouring years. Pick one object that required more than one trade. Then identify a question that was still open. If a film or interactive feature fails, the written entries provide a clear account of the work without requiring a player.
Guédelon’s strongest invitation is to see historical knowledge as something that can be questioned through making. The stones matter, but so do the decisions between them: a sequence tried, a difficulty encountered and a method revised in response to evidence.
