How we know what it looked like

When you see a careful reconstruction of a place that no longer exists, a Roman street, a medieval hall, a vanished city skyline, a reasonable question is: how could anyone possibly know? The building is gone. No one alive saw it. Is this just an educated guess with good production values? Sometimes, honestly, parts of it are. But the gap between guesswork and reconstruction is real, and it is filled by a surprising number of independent kinds of evidence, each blind to the others, that have to agree before anyone draws a confident line. Knowing what those evidence streams are is the difference between trusting a picture of the past and being fooled by one.

The ground keeps the floor plan

The first witness is the site itself. Even when everything above waist height is gone, foundations, post-holes, drains, and floor surfaces usually survive, and they fix the one thing that is hardest to fake: the plan. Where the walls ran, how wide the doors were, where the hearth sat, which way the street went. Archaeology recovers the geometry of a place with a precision that no later description can match, because the earth does not exaggerate.

A few sites are almost cruelly generous. At Pompeii, buried fast and deep, the streets, shops, counters, graffiti, and even the shapes of the dead survive, the last of these as plaster casts poured into the voids their bodies left in the ash. There you are not inferring the town. You are nearly walking it. Most sites give far less, but even a stripped foundation answers the first and most important question, which is not what colour the wall was but where the wall was. Reconstruction starts from the plan, because the plan is the part the ground refuses to lose.

Paper remembers the contents

Foundations give you the empty shell. To furnish it you turn to writing, and the most valuable writing is rarely the grand chronicle. It is the boring paperwork. Probate inventories, drawn up to value a dead person's goods room by room, walk you through a house naming what stood in each space, down to the pots and the bedding. Account books record what a building cost, in what materials, paid to which trades, which tells you what it was made of. Tax rolls, leases, and court records pin down who lived where and how the space was divided.

This is unglamorous evidence and it is gold, because it was written without any thought of impressing posterity. A merchant listing his late father's kitchen had no reason to lie about whether there was a second table. Historians reconstructing the texture of past domestic life lean heavily on exactly this kind of record, precisely because it is so artless. The most trustworthy witness to ordinary life is the document that never expected to be read by us.

Pictures help, and pictures lie

Paintings, prints, and drawings are the obvious source, and they are genuinely useful, but they are the evidence that needs the most discipline. An image was made by someone, for someone, with a purpose, and that purpose bent it. A view commissioned to flatter a duke makes his palace bigger and his town tidier. A devotional scene puts a contemporary city behind a biblical event because the painter knew no other city to draw. Perspective itself is a recent and uneven convention; much older imagery encodes rank or meaning in size rather than distance.

So a single picture is a clue, not a verdict. Read against the foundations and the documents, it can confirm a roofline, a material, the cut of a window, the way a crowd dressed. Read alone and trusted, it will hand you the painter's flattery as if it were the building. The honest method treats images as testimony from an interested party: worth hearing, never taken at its word.

The land, the timber, and the bones

Some of the most powerful evidence is the kind that cannot lie because it was never trying to communicate at all. Environmental archaeology reads pollen, seeds, and insects out of old soil layers to tell you what actually grew around a place and what was farmed, so the fields and gardens in a reconstruction need not be invented. Dendrochronology, the matching of tree-ring patterns, can date a surviving timber to the year it was felled, and sometimes to the forest it came from. Isotope analysis of bones and teeth reveals what people ate and where they grew up, which is how we know that a given graveyard held locals or incomers. Traces of pigment clinging to stone, recovered under ultraviolet light, return the colour that weather stripped away.

None of these sources knows what the others say. The pollen does not consult the account book; the tree rings have never seen the painting. That is exactly why agreement between them is convincing. When the soil, the timber, the documents, and the foundations independently point at the same picture, the picture has earned its confidence.

The honest part is the gaps

A trustworthy reconstruction is as clear about what it does not know as about what it does. We can often be precise about the plan, the materials, and the broad palette, and quite confident about dress and daily objects. We are usually guessing at the exact paint colour of an ordinary house, the precise faces in a crowd, the particular junk in a particular corner. Sound and smell, as a rule, we have lost entirely.

The discipline, then, is not to reconstruct everything with equal confidence. It is to build firmly where the evidence is firm, to infer carefully where several weak signals line up, and to mark, rather than hide, the places where the honest answer is that we do not know. A reconstruction that pretends to certainty it has not earned is a kind of lie, however beautiful. One that shows its confidence and its limits is doing the real work. Knowing what the past looked like is not a single act of knowing. It is many separate witnesses, none of them complete, made to stand in the same room until they agree.