The Riven
A froe is an L-shaped blade used to cleave wood along its grain. The maker positions the blade and strikes once; the wood does the rest. The split follows the fiber structure — weaving around knots, curving where the grain curves, producing a piece whose surface is entirely intact fibers. No fiber is severed. The result is stronger, more water-resistant, and more dimensionally stable than anything a saw could produce from the same billet. But each piece is unique. The grain chose the split line, not the maker. You cannot blueprint a riven piece because the information that determines its shape is not available until the wood reveals it.
Sawing solved this problem by ignoring it. A saw cuts across fibers indiscriminately, producing pieces of predictable dimension regardless of internal structure. Every board is the same width. Every board has severed fibers on both faces, wicking moisture into the wood's interior. The industrial revolution chose sawing because it scales: one specification produces a thousand identical pieces. Riving does not scale because it cannot generalize. Each piece requires reading the specific billet — its growth history, its knots, its internal tensions. Peter Follansbee, reviving seventeenth-century green woodworking, describes the riving technique as a conversation with each individual piece of wood. The conversation cannot be had in advance.
This is not a parable about craft nostalgia. The distinction between riving and sawing is the distinction between following structure and imposing structure, and it appears everywhere that materials have internal organization.
Sashimono — Japanese cabinet-making — uses interlocking joints without nails, screws, or adhesive. The joints are not arbitrary geometry. They are designed around the fact that wood is hygroscopic: it expands and contracts with humidity changes, and it does so asymmetrically, moving more across the grain than along it. A sashimono mortise-and-tenon joint is cut to permit movement in the direction the wood wants to move. Constrain it in the wrong axis and the joint will destroy itself over a single season. The joint succeeds by accommodating the material's tendency, not by overriding it.
The word kigumi — wood assembly — treats each piece as a diagnostic problem. The joiner reads the specific wood: where the grain runs, how it was dried, what movement it still carries. Two pieces from the same tree may require different joints. The knowledge is in reading, not in a catalog of joint types. Cataloging joints is like cataloging diagnoses — useful, but the skill is in seeing which one the specific case requires.
Dry-stone walling operates on the same principle at a different scale. A dry-stone wall uses no mortar. Each stone is selected and placed according to its individual shape, weight distribution, and center of gravity. The Inca walls at Sacsayhuamán fit so precisely that paper cannot be inserted between the stones — not because the stones were cut to specification, but because each stone was read and placed where its specific geometry made it load-bearing. Galloway dykes in Scotland stand after two centuries of Atlantic weather. Aran Islands walls on Ireland's west coast let gale-force winds pass through their gaps rather than resisting them — reducing wind load on the structure, turning what looks like incomplete construction into pressure relief.
A dry-stone waller's knowledge is embodied. It cannot be fully written down. You learn it by handling stone — feeling weight, reading fracture planes, sensing which orientation lets this particular stone lock against its neighbors. No two walls are alike because no two collections of stones are alike. The wall is a record of every diagnostic decision the builder made, stone by stone, from base to cap.
Cob building is earth construction: subsoil, water, fibrous material, sometimes lime. It is built in lifts — each course applied and left to dry before the next. The material determines the construction schedule. Its shrinkage rate sets the pace; its compression strength sets the wall thickness; its thermal mass determines where in the wall section to place openings. Cob walls at Hayes Barton in Devon are more than five hundred years old. They survived not because they were built to resist the environment but because they were built to behave the way their material behaves. Earth wants to settle, crack along certain lines, absorb and release moisture on a diurnal cycle. The cob wall lets it.
Concrete is the material that most completely overrides internal structure. Liquid concrete fills any formwork, hardens into whatever geometry was specified, and requires no reading of the material at all. There is no grain to follow, no shape to diagnose, no internal organization to accommodate.
But modern reinforced concrete has a design life of fifty to one hundred years. Carbonation advances inward from the surface; when it reaches the steel reinforcement, corrosion begins, expanding the rebar and cracking the concrete from inside. The material's ongoing chemistry — the one thing the engineer did not specify — is what destroys it. The imposition holds until the material's own process reasserts itself.
Roman marine concrete tells the other half of the story. Made with volcanic ash, lime, and seawater, it was placed and then left to its own chemistry. Marie Jackson and colleagues showed in 2017 that Roman concrete in marine environments strengthens over centuries. Seawater infiltrating the matrix produces Al-tobermorite and phillipsite crystals that grow in microcracks, progressively reinforcing the structure. The material was never finished. It continued its own chemical process after placement, and that process — the one the Romans did not design but did not prevent — is what makes two-thousand-year-old harbor breakwaters more durable than modern equivalents built to specification.
Modern Portland cement is designed to be chemically inert after curing. The engineer's intent is that the material should stop reacting — should be done, fixed, imposed. Roman concrete was never done. It followed its own chemistry for millennia. The material that was allowed to keep working outlasted the material that was required to stop.
Following structure requires more knowledge than imposing structure. The knowledge is diagnostic — specific to the instance, learned through contact with the material, not fully transferable through specification. A saw does not need to know the grain. A nail does not need to know the wood's movement. Mortar does not need to know the stone's shape. Formwork does not need to know the concrete's chemistry. Imposition scales precisely because it treats the material as interchangeable.
But the structures that follow tend to outlast the structures that impose. The reason is not mystical. It is mechanical. A structure that follows the material's internal organization does not need to resist the material's tendency to return to that organization. It is already there. A structure that imposes external geometry must continuously resist the material's preference — wood wants to move along its grain, stone wants to settle into its natural angle of repose, concrete wants to continue reacting. Every imposition is a bet that the override will outlast the tendency. Over long enough timescales, the tendency wins.
The cost of following is legibility. You cannot write a specification for a dry-stone wall. You cannot mass-produce riven shingles. You cannot blueprint sashimono joints without seeing the specific wood. The knowledge required to follow structure is exactly the knowledge that resists being written down — which means it resists being scaled, standardized, taught in classrooms, or encoded in building codes. The industrial revolution chose imposition not because it was better but because it was writable.
This is the trade: legibility for longevity. Every system that scales does so by ignoring the specific case. Every system that lasts does so by attending to it.