The Saxonalist · a data project

Tile, Tin, Grass

The villages time has saved — and the ones it is erasing

Roof by roof, we judged 27 Saxon villages in Transylvania, on satellite imagery and two decades of aerial archives. Three materials, three fates: what endured, what changed in order to live, and what left nothing behind but grass.

Bogdan Pencea · July 2026

scroll ↓

Homesteads in Dacia, 2005

The same place, today

Dacia (Stein) · 2005 orthophoto

Map of the Saxon villages

Southern Transylvania · 27 villages in the study

The roofs of Viscri

Viscri (Deutsch-Weißkirch) · 448 buildings

Prejmer

Prejmer (Tartlau) · 1,842 buildings

Dacia

Dacia (Stein) · what remains

In 2005, on the edge of the village of Dacia, stood a cluster of Saxon homesteads: houses fronting the street, barns behind, roof tile baked by the sun.

Today, in the same spot, there is a meadow.

No one demolished anything. The houses were simply left alone — and in twenty years, the land took them back.

It happened in silence, across an arc of villages stretching two hundred kilometres between Brașov and Sebeș.

Eight hundred years ago, German settlers built hundreds of settlements here. The community left in the 1990s. Now the villages themselves are beginning to leave.

To measure what remains, we descended onto every single roof.

Esri and ANCPI imagery · realigned building footprints · human verdicts on random samples

This is Viscri — the village adopted by a king. 448 buildings, judged one by one.

97% of its roofs are still ceramic tile — many so old that, from above, the moss makes them look like grass.

Each brick-red patch = one tile-roofed building, human-confirmed

Fifty kilometres east, near Brașov, Prejmer tells a different story. The village is alive and prosperous, deep in commuter country.

But prosperity has a colour of its own: the blue and grey of tin, replacing tile homestead by homestead.

Here only 86% of the roofs are still tile. Abandonment, on the other hand, is close to zero.

And in Dacia, the village where this story began, the threat is not tin.

It is disappearance. Of the buildings it had in 2005, roughly 1 in 4 no longer exists.

Each red ring = one building confirmed gone, on the 2005–2016 aerial archive

Two erosions, two fates. The Saxon villages do not die the same way. Some are eaten by prosperity: they stay inhabited but lose their substance — tile gives way to tin, house by house, precisely in the villages with a healthy economy. Others keep their substance almost intact and lose everything instead: in Richiș and in Dacia, old tile dominates whatever remains — it is just that less and less remains.

Between these two erosions stand the exceptions that prove the rule: Viscri, saved by the fame of a king, and Mălâncrav, adopted by a foundation. The data points to an uncomfortable conclusion: villages do not survive because they are beautiful. They survive because they are known.

We put this thesis to the test, with numbers: each village’s notoriety — twelve months of views of its Wikipedia articles — set against the two axes. On material, the verdict is blunt: notoriety does not matter at all. On abandonment, the sign is the expected one — the better-known half of the list averages 1.8% of buildings confirmed gone, the obscure half 2.6%, and all four villages with a ruination profile sit in the obscure half. But across 27 villages the difference stays below the threshold of statistical significance. The thesis remains what it was: a hypothesis — suggested by the data, not yet proven by it.

With all 27 villages measured and human-verified, the first erosion acquires a geography: five of the six villages with the least tile — Feldioara (68%), Hărman (72%), Sânpetru (73%), Cristian (76%) and Hălchiu (78%) — sit in the commuter belt of Brașov. The most economically alive villages are precisely the ones losing their substance fastest.

Dacia, 2005 → 2026

drag the handle · the same patch of village, twenty years apart

⟷

2005

today

ANCPI 2005 orthophoto (left of the handle) and current satellite imagery (right), aligned on the same grid. Sources: ANCPI, Esri.

Down in the grass

Dacia, 12 July 2026 · photographs: Bogdan Pencea · the locations match ruins confirmed on the aerial archive

A mound of vegetation that used to be a homestead — the spot from the satellite image that opens the article — This is the place from the opening of this article — the cluster of homesteads in the 2005 satellite image, seen from the ground, today. The roof sheets have melted into the grass; the land has taken back the rest.

Ornamented Saxon gable next to a collapsing house, in Dacia — The two erosions in a single frame: the Saxon gable with its ornaments still legible, the house next door already going, the fence under moss.

Wall with decorative frames and no roof, in Dacia — The decorative frames are intact; the roof is gone entirely. On the automated maps the house still exists — on the ground, it is a footprint without a roof.

Goats sheltering in a collapsed barn, in Dacia — The ruin has become a shelter: the village goats, under the fallen beams of a barn.

Two ways to lose a village

each dot = one village · horizontal: 95% confidence interval for tile · vertical: from confirmed ruins up to the ceiling with uncertain cases

How we measured

For every village we extracted the footprints of all buildings (OpenStreetMap or Microsoft Building Footprints), realigned them on the satellite image, and human-judged a random sample of ~90 buildings: tile, another material, or no roof. The percentages carry 95% confidence intervals.

“Gone” is not a guess: every building reported as a ruin was checked against the ANCPI orthophotos of 2005, 2012 and 2016 — it exists in the archive, it no longer exists today. Phantom outlines from the automated datasets were removed from the calculation.

One acknowledged limit: in the peri-urban communes of the Brașov commuter belt, part of the tile we count is modern tile on new houses — there, the percentage measures the material, not the heritage.

Data and methodology: The Saxonalist project · July 2026. The tile percentages are human-validated for all 27 villages in the study, and the abandonment figures are arbitrated, village by village, against the 2005–2016 aerial archive; where the archive does not allow a verdict, the uncertainty is carried explicitly as an interval.

Next: all ~430 Saxon settlements

The study is expanding to the whole of Saxon Transylvania — every settlement, both erosions. Leave your e-mail and we will write to you once, when it is published.

Research begun and written at Viscri 9 — a guesthouse in the village with 97% tile.