Vernacular house form

Why do houses built before air conditioning suit their climates better than newer ones do?

Travel across a hot dry country and the old houses look uncannily alike: thick earth or stone walls, small deep-set windows, pale surfaces, rooms turned inward around a shaded court. Travel across a hot wet one and the old houses are the opposite in almost every respect — light timber, lifted off the ground, wide overhanging eaves, tall rooms, walls that leak air on purpose. Then look at what has been built in both places since about 1960 and the two increasingly resemble each other, and neither their climate.

The flattering explanation is that older builders understood their climate deeply and we have forgotten. Before accepting that, ask a harder question: where would that understanding have come from? Very few of those builders had measured anything. So what process produced a fit that looks designed?

Start with the two extremes and check that the forms really are doing something physical, not merely looking traditional.

In a hot dry place the daily temperature swing is large — scorching afternoons, cool nights. A wall of thick earth or stone does not so much block heat as delay it: the outside heat takes many hours to work through, arriving inside long after the sun has gone, by which time the night air is carrying it away again. So the mass is not insulation, it is a time-shift, and it only pays where the nights actually cool. Small windows keep radiant gain out; the courtyard shades itself for most of the day and dumps heat to a clear night sky.

In a hot wet place none of that works. The nights stay warm, so a delayed arrival of heat just means heat at midnight, and thermal mass becomes a liability. What the body needs there is air movement and dryness, so the house becomes light, lifted clear of damp ground, tall enough for hot air to rise away, and deliberately permeable — shaded openings on both sides so a breeze can pass through. Deep eaves keep rain off those open walls. Every choice is the reverse of the desert house, for a reason.

Good. So the forms are genuinely climate-responsive. Now, the mechanism. Did anyone work this out?

Ask what happens to a builder who gets it wrong. The house is uncomfortable in the worst month, the roof fails in the rains, the earth wall slumps — visible to the whole settlement, and expensive. Get it right and the house stands, is bearable, and becomes the model the next builder copies, because copying what is standing is what you do with no other information. Do you see that no understanding is required anywhere in this loop? Variation comes from ordinary imprecision and local materials, selection from climate and discomfort, transmission from imitation. That is an adaptive process, and its product looks designed for the same reason a beak does.

Which tells us the conditions it needs: many repetitions, so the same people building in the same place over generations; a climate that holds roughly steady; and a penalty for bad form that lands on someone who can respond.

Now ask what air conditioning did to that loop. It did not make anyone stupider. It made the climate stop selecting. A poorly shaded, thin-walled, wrongly oriented house is now perfectly comfortable; you simply pay for it on the electricity meter. The feedback that used to arrive as unbearable afternoons now arrives as a bill, which is easy to absorb, arrives months later, and is not visibly attributable to the shape of the building. So form was released to follow other pressures — land cost, road frontage, catalogue plans, standardised materials, fashion, and building codes written nationally rather than regionally. Add speculative construction, where the person who builds the house never lives in it, and the loop is cut at a second point: the signal, even if it did arrive, would not reach the builder.

Now the honesty this argument needs. We are comparing surviving old houses with all new ones, and only the good old ones survived — straightforward survivorship bias, which inflates how wise the past looks. Vernacular houses also came bundled with behaviour: shutters closed at midday, seasonal migration between rooms, and a tolerance for a far wider range of indoor temperature than we now accept. Measured against modern expectations, with modern appliances heating them from inside, they often underperform their reputation. The claim that holds up is narrower and still interesting: the old forms were shaped by their climate, and the new ones largely were not.

The correction worth making is to the word "knew." It is tempting to say traditional builders knew about thermal lag and stack ventilation. Almost none did, in the sense of holding a theory. What existed was a body of practice that encoded those effects without naming them, and the difference matters, because knowledge transfers by explanation while practice transfers only by continuity. Once the chain of copying broke — through migration, industrial materials, or a single generation building differently — the encoded solution was simply gone, with no text to recover it from.

It also follows that reviving vernacular form by imitation misses the point. Copying a courtyard into a climate that never had one reproduces the appearance of adaptation without its cause. The transferable thing is not the form but the questions it answered: how large is the daily swing, does the night cool, is the air dry enough for evaporation to help, where does the rain come from. Contemporary climate-responsive design asks those directly — faster than centuries of selection, provided the answers are recalculated for a climate that is itself moving.

• What happens to the loop under a shifting climate, where the forms selected by the last two centuries may no longer be the right answer.
• Why regional building codes and cheap glazing spread standardised forms so quickly across very different climates.
• Whether measured performance of traditional houses matches their reputation once modern comfort expectations and internal heat loads are applied.

Every term the collection defines is gathered in the glossary.