A Gothic cathedral is sometimes described as a sermon in stone. The description is memorable but misleading, because it implies a single speaker and a clear message. The buildings themselves are more like accumulated conversations, extended over generations, among bishops, master masons, donors and users, each bringing a different vocabulary. Learning to read a cathedral means paying attention not only to what is visible but to the sequence in which it was built, and to the compromises that any long project inevitably imposes.

The cathedral of Chartres, in northern France, is a useful example. Its western facade, completed in the 1150s, is recognisably older in its decoration than the soaring nave behind it, which rose after a fire destroyed most of the building in 1194. The rebuilding was carried out in less than thirty years - remarkable speed for a project of such scale - and the result is an unusually coherent interior. The stained glass, however, was installed over decades, and careful inspection reveals at least three distinct workshops at work, each with a slightly different style of figure drawing and of colour palette. Art historian Dr. Colette Renard has used these stylistic differences to date individual windows with a precision unavailable through documents alone.

Building techniques evolved visibly within a single cathedral. Early thirteenth-century masons at Amiens, for example, used fairly massive piers in the lower levels of the choir. By the time they reached the upper parts of the nave, they had learned to reduce the thickness of the piers without sacrificing stability, probably because new methods of calculating thrust had been developed in the workshops. Amiens therefore preserves, within a single building, the record of an engineering advance. Reading the cathedral from bottom to top is in some sense reading the history of thirteenth-century structural thinking.

Donors left their own traces. The lower panels of many windows show small portraits of the tradespeople - bakers, wheelwrights, dyers - whose guilds had paid for the glass. These panels were not ornaments added at the end; they were negotiated into the design as a condition of the donation. Similarly, the tombs of bishops and nobles, now distributed around the side chapels, were often commissioned decades after the main building work and integrated with more or less success. Some chapels were reshaped to receive a particular tomb; others show the awkward seam where a later monument intrudes on an earlier space.

Theologically, cathedrals were meant to be readable in a particular way. The fa??ades of the great French cathedrals are usually organised as a scheme. At Chartres, the central doorway shows Christ in Majesty, flanked by scenes from the Old Testament on one side and the lives of apostles and saints on the other. This programme was not invented on site; it followed widely circulated theological templates, adjusted to local tradition. Medieval visitors with almost no literacy would have recognised many of the figures from sermons and from illustrated manuscripts. For them, the fa??ade was a summary of Christian history displayed in the form of a permanent image.

Not everything in a cathedral was planned. Fires, storms and wars forced repairs in styles that differed from the surroundings. The rose window on the north transept at Reims, famously damaged during the First World War, was rebuilt using contemporary techniques that make the restored sections visible on close inspection. Liturgical reform also left physical marks. The removal of rood screens - stone or wooden barriers that once separated the choir from the nave - was carried out in many English cathedrals during the seventeenth and eighteenth centuries, opening up views that the original designers had never intended.

Seen in this way, a cathedral is less a finished work than a long, imperfectly coordinated project. It records the ambitions, skills and disputes of many overlapping groups. The great merit of Gothic buildings is that, despite these layers, they usually feel coherent rather than chaotic. Dr. Renard argues that this apparent coherence is the result of a strong shared vocabulary of forms: arches, tracery, buttresses, vaults, which gave successive builders a common language even when they disagreed about the content. Whether or not one accepts her explanation, the experience of standing inside such a building and sensing several centuries at once is difficult to describe and harder still to imitate.