Knowing Through Making

How do you know that something is wrong, before you can explain why?

A mason can look at a proportion and know that it is off. A carpenter may wait before beginning the next task because the material does not yet feel ready. Someone who has lived around a courtyard for years often knows which verandah will be comfortable at four in the afternoon and which one will not. None of them necessarily needs a formula in order to act with precision.

This is a slightly uncomfortable proposition for architecture, because we are trained to trust things that can be drawn, written, dimensioned and specified. We like the measurable. A drawing can be checked, a specification can be quoted, a calculation can be repeated. Knowledge, preferably, should leave paperwork behind.

But does it always?

Following Rules Without Spelling Them Out

A builder may judge a proportion, sequence a task, or spot an inappropriate relation without ever stating the abstract rule behind it. This does not mean that there is no rule. It only means that knowing how to work with a rule and knowing how to describe that rule are not the same thing. A theoretical framework might name the rule after the fact; the practitioner does not need to.

This becomes particularly interesting in traditional building practice. Classical Indian architectural texts specify proportional relationships between the principal elements of a temple—the sanctum, mandapa, tower—using modular systems such as tāla. But the historical relationship between those written instructions and what builders actually did is considerably harder to establish.

The documented practice of V. Ganapati Sthapati gives us a useful glimpse.¹ Here, proportional competence survives through family and workshop training, but it is not frozen. It adapts. The interesting question is not whether the proportions existed, but whether the craftsman had to consciously calculate the relationship every single time. The available evidence does not tell us that he did. It is entirely possible for repeated calculation to become judgement.

We do this all the time. You do not calculate the trajectory of your foot every time you climb a staircase; you simply know when the next step feels wrong.

The Building Knows More Than the Drawing

Consider a courtyard house. Its inhabitants gradually learn where to sit, sleep, work, talk, dry clothes, move furniture, open doors, close shutters, or avoid the afternoon sun. They follow temperature and shade bodily rather than numerically. The verandah that was comfortable at nine in the morning may become intolerable by three, while another edge of the courtyard becomes preferable. The house appears static; occupation is not.

One could describe this afterwards using dimensions, orientation, thermal mass, air movement, shade and radiation. The inhabitant need not. The competence is practical before it becomes analytical.

This is why I increasingly suspect that a building can carry knowledge which a drawing merely represents. A drawing tells us where things are; a building allows us to discover what those relationships do.

Architecture Measured Against the Body

This is not uniquely Indian. Roxana Waterson’s studies of South-East Asian houses describe building traditions in which spatial hierarchies—front and back, ancestral and everyday space—operate systematically without requiring their inhabitants to formulate a theory of those relationships.² In several Indonesian and Malay traditions, dimensions may even be derived from the body of the builder or householder rather than from an abstract standard.

There is something wonderfully direct about that. Not a universal metre, but this person, this house, this relationship.

Australian Aboriginal songlines make the proposition even clearer. Territorial and navigational knowledge can be transmitted orally, through story, song and performance, rather than by map.³ The landscape has not become less precise because the knowledge is not drawn. The precision simply resides elsewhere.

Perhaps our problem is not that unwritten knowledge is vague. Perhaps we have trained ourselves to recognise precision only in the forms in which we usually record it.

Duration Matters Twice

Old traditions give us one advantage: time. Not because old knowledge is automatically better—ancientness is not intelligence—but because a practice repeated through different generations, materials, sites and circumstances leaves behind more evidence.

One example can be an accident. Repeated relationships are harder to dismiss.

Duration matters twice. For the practitioner, repetition can turn an action into habit, judgement, procedure, eventually even institution. For us, looking from outside, repetition gives us something to compare: what changes, what stubbornly remains, where the rule bends, and where it breaks.

This is where variation becomes useful. If everything remained identical, we would learn very little about the underlying rule. Difference is evidence.

A Rule Is Not a Recipe

This may be the simplest distinction in the whole argument. A recipe gives you an answer; a rule helps you produce another answer when the circumstances change.

Ganapati Sthapati could retain inherited proportional and spiritual commitments while adapting engineering decisions to new circumstances.¹ That is not copying. Copying produces the same answer regardless of the question. A rule survives because the answer can change.

Julia Hegewald’s study of Jain temples along the Karnataka coast makes this especially visible.⁴ Along the wet coast, timber and mud construction worked beneath wide, projecting roofs suited to heavy monsoon rain. On the drier Deccan plateau, builders worked more readily in stone. Over time, available materials and technologies changed; stone became more common along the coast, and reinforced concrete entered the available vocabulary too. Each material behaved differently and demanded a different structural response.

Yet the roof still had to throw rain away.

The specification changed; the rule connecting form to function did not.

This, to me, is far more interesting than stylistic continuity. A copied form tells us that somebody remembered an appearance. An adapted rule suggests that somebody understood a relationship.

Buildings Can Be Teachers

Where, then, does such knowledge live? Sometimes in people, sometimes in institutions, and sometimes, perhaps, in the thing being made.

Edward Simpson describes shipbuilding workshops in Kachchh where vessels are assembled from thousands of pieces of timber without plans, technical drawings or sophisticated machinery.⁵ The apprentice learns beside the ship. At first he is told exactly what to do. Later, less. Eventually, he is expected to know.

This is a very different model of teaching from the one we have become accustomed to in architectural school. There is no moment at which the entire system is handed over as information. Competence arrives piecemeal, and responsibility follows it. First: do this. Later: you know what needs doing. Eventually: decide.

The teacher withdraws as judgement appears.

That is pedagogy too.

Some Things Have to Be Done

The hereditary bronze-casting workshops of Swamimalai demonstrate another part of the same problem.⁶ Clay and wax behave differently according to heat, moisture, timing and handling. One may explain this and one may write it down, but at some stage someone has to touch the material.

Too wet. Too dry. Too hot. Too early. Not yet.

There is no insult to theory in saying this. Some knowledge becomes available through doing in ways description alone cannot reproduce. The hand is learning, but so is judgement.

This is also why I am suspicious of calling such knowledge “intuition.” Intuition sounds accidental, mysterious, almost magical. This is not magic. It is repeated, corrected, tested and disciplined.

Can an Institution Remember?

Knowledge does not necessarily have to belong to an individual either. At the Chidambaram temple, custodial knowledge within the Dikshitar community is distributed across a collective institutional structure rather than concentrated in one person.

That matters. If all knowledge lives in a single expert, the death of the expert can be catastrophic. If responsibility, procedure and memory are spread across many people, the system has redundancy—much like a structure. Remove one member and it may continue to stand; remove enough of the system and something else happens.

When the Chain Breaks

This is where the romantic story of traditional knowledge needs to end.

Embodied knowledge does not survive simply because it is embodied. A body dies. A workshop closes. A patron disappears. A trade becomes unviable. A son chooses another profession. A teacher has nobody left to teach.

The knowledge survives only while the conditions that allow one person to teach another remain in place.

The decline of older karkhana systems is important for precisely this reason. When workshops and apprenticeship chains weaken, the architecture may remain while the ability to reproduce its intelligence begins to disappear.

This produces a peculiar kind of survival. The building survives. The photograph survives. The measured drawing survives. Perhaps even the dimensions survive. And yet something has been lost.

We know what it looked like. We may know exactly how large it was.

But we no longer know what to do next.

A building can survive while the capacity to reproduce its intelligence does not.

Perhaps that is why recovery matters—not to reconstruct some perfect, uninterrupted past, and certainly not to copy old buildings, but to ask a more difficult question:

What did people know how to do here—and how did they learn to do it?

I am beginning to think that the answer may tell us more about architecture than the building alone ever could.